From a7245c0f624b90a50109ed4775af39b6492d500d Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Thu, 6 Aug 2026 11:34:48 +0800 Subject: [PATCH 01/14] feat(deploy): add pause/resume commands; JSON-only output for dataset/finetune/deploy - Add `bl deploy pause` and `bl deploy resume` (console domain, first console-auth commands in deploy group) via modelInstance start/stop APIs - Add core deploy/lifecycle.ts with input-wrapped console gateway calls - Switch all dataset/finetune/deploy commands to JSON-only output, removing text formatting logic - Expose usage/charge_type in finetune get, model_name/expire_time in finetune checkpoints with near-expiry warning - Update deploy delete hint to suggest `bl deploy pause` --- packages/cli/src/commands.ts | 4 + .../commands/src/commands/dataset/delete.ts | 14 +- packages/commands/src/commands/dataset/get.ts | 24 +-- .../commands/src/commands/dataset/list.ts | 26 +-- .../commands/src/commands/dataset/upload.ts | 11 +- .../commands/src/commands/dataset/validate.ts | 42 +---- .../commands/src/commands/deploy/create.ts | 17 +- .../commands/src/commands/deploy/delete.ts | 14 +- packages/commands/src/commands/deploy/get.ts | 23 +-- packages/commands/src/commands/deploy/list.ts | 35 +--- .../commands/src/commands/deploy/models.ts | 151 ++++++------------ .../commands/src/commands/deploy/pause.ts | 85 ++++++++++ .../commands/src/commands/deploy/resume.ts | 84 ++++++++++ .../commands/src/commands/deploy/scale.ts | 19 +-- .../commands/src/commands/deploy/update.ts | 22 +-- .../commands/src/commands/finetune/cancel.ts | 14 +- .../src/commands/finetune/capability.ts | 68 +++----- .../src/commands/finetune/checkpoints.ts | 48 +++--- .../commands/src/commands/finetune/create.ts | 17 +- .../commands/src/commands/finetune/delete.ts | 12 +- .../commands/src/commands/finetune/export.ts | 22 +-- .../commands/src/commands/finetune/get.ts | 50 ++---- .../commands/src/commands/finetune/list.ts | 56 ++----- .../commands/src/commands/finetune/logs.ts | 59 ++----- .../commands/src/commands/finetune/watch.ts | 28 ++-- packages/commands/src/index.ts | 2 + packages/core/src/deploy/index.ts | 1 + packages/core/src/deploy/lifecycle.ts | 99 ++++++++++++ skills/bailian-finetune/reference/deploy.md | 100 ++++++++++-- skills/bailian-finetune/reference/finetune.md | 4 +- skills/bailian-finetune/reference/index.md | 4 +- 31 files changed, 584 insertions(+), 571 deletions(-) create mode 100644 packages/commands/src/commands/deploy/pause.ts create mode 100644 packages/commands/src/commands/deploy/resume.ts create mode 100644 packages/core/src/deploy/lifecycle.ts diff --git a/packages/cli/src/commands.ts b/packages/cli/src/commands.ts index 2cbda30d..bd18f4d5 100644 --- a/packages/cli/src/commands.ts +++ b/packages/cli/src/commands.ts @@ -80,6 +80,8 @@ import { deployScale, deployUpdate, deployDelete, + deployPause, + deployResume, tokenPlanListSeats, tokenPlanCreateKey, tokenPlanAssignSeats, @@ -198,6 +200,8 @@ export const commands: Record = { "deploy scale": deployScale, "deploy update": deployUpdate, "deploy delete": deployDelete, + "deploy pause": deployPause, + "deploy resume": deployResume, "token-plan list-seats": tokenPlanListSeats, "token-plan create-key": tokenPlanCreateKey, "token-plan assign-seats": tokenPlanAssignSeats, diff --git a/packages/commands/src/commands/dataset/delete.ts b/packages/commands/src/commands/dataset/delete.ts index 5e18766a..4c1e888b 100644 --- a/packages/commands/src/commands/dataset/delete.ts +++ b/packages/commands/src/commands/dataset/delete.ts @@ -1,5 +1,5 @@ -import { defineCommand, detectOutputFormat, deleteDataset, type FlagsDef } from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { defineCommand, deleteDataset, type FlagsDef } from "bailian-cli-core"; +import { emitResult, emitBare } from "bailian-cli-runtime"; const DELETE_FLAGS = { fileId: { @@ -19,20 +19,18 @@ export default defineCommand({ async run(ctx) { const { settings, flags } = ctx; const fileId = flags.fileId; - const format = detectOutputFormat(settings.output); if (settings.dryRun) { - emitResult({ action: "dataset.delete", file_id: fileId }, format); + emitResult({ action: "dataset.delete", file_id: fileId }, "json"); return; } const response = await deleteDataset(ctx.client, fileId); - if (settings.quiet || format === "text") { - emitBare(`Deleted ${fileId}.`); - emitRequestId(response.request_id, settings.quiet); + if (settings.quiet) { + emitBare(fileId); } else { - emitResult(response, format); + emitResult(response, "json"); } }, }); diff --git a/packages/commands/src/commands/dataset/get.ts b/packages/commands/src/commands/dataset/get.ts index 2c1cf714..f9a41451 100644 --- a/packages/commands/src/commands/dataset/get.ts +++ b/packages/commands/src/commands/dataset/get.ts @@ -1,5 +1,5 @@ -import { defineCommand, detectOutputFormat, getDataset, type FlagsDef } from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { defineCommand, getDataset, type FlagsDef } from "bailian-cli-core"; +import { emitResult, emitBare } from "bailian-cli-runtime"; const GET_FLAGS = { fileId: { @@ -19,10 +19,9 @@ export default defineCommand({ async run(ctx) { const { settings, flags } = ctx; const fileId = flags.fileId; - const format = detectOutputFormat(settings.output); if (settings.dryRun) { - emitResult({ action: "dataset.get", file_id: fileId }, format); + emitResult({ action: "dataset.get", file_id: fileId }, "json"); return; } @@ -45,19 +44,10 @@ export default defineCommand({ description: file.description ?? "", }; - if (format === "json") { - emitResult({ ...item, request_id: response.request_id }, format); - return; + if (settings.quiet) { + emitBare(item.file_id); + } else { + emitResult({ ...item, request_id: response.request_id }, "json"); } - - // text / quiet - emitBare(`file_id: ${item.file_id}`); - emitBare(`name: ${item.name}`); - emitBare(`size: ${item.size}`); - if (item.md5) emitBare(`md5: ${item.md5}`); - if (item.purpose) emitBare(`purpose: ${item.purpose}`); - if (item.created_at) emitBare(`created_at: ${item.created_at}`); - if (item.description) emitBare(`description: ${item.description}`); - emitRequestId(response.request_id, settings.quiet); }, }); diff --git a/packages/commands/src/commands/dataset/list.ts b/packages/commands/src/commands/dataset/list.ts index 7bc858b4..63bbfeb2 100644 --- a/packages/commands/src/commands/dataset/list.ts +++ b/packages/commands/src/commands/dataset/list.ts @@ -1,5 +1,5 @@ -import { defineCommand, detectOutputFormat, listDatasets, type FlagsDef } from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId, formatTable } from "bailian-cli-runtime"; +import { defineCommand, listDatasets, type FlagsDef } from "bailian-cli-core"; +import { emitResult, emitBare } from "bailian-cli-runtime"; const LIST_FLAGS = { page: { type: "number", valueHint: "", description: "Page number (default: 1)" }, @@ -23,7 +23,6 @@ export default defineCommand({ exampleArgs: ["", "--purpose fine-tune", "--purpose evaluation --page-size 20", "--output json"], async run(ctx) { const { settings, flags } = ctx; - const format = detectOutputFormat(settings.output); if (settings.dryRun) { emitResult( @@ -33,7 +32,7 @@ export default defineCommand({ page_size: flags.pageSize, purpose: flags.purpose, }, - format, + "json", ); return; } @@ -46,7 +45,6 @@ export default defineCommand({ const files = response.data?.files ?? []; const total = response.data?.total; - // Normalize to consistent structure for both text/json output. const items = files.map((item) => ({ file_id: item.file_id ?? "", name: item.name ?? "", @@ -54,20 +52,10 @@ export default defineCommand({ purpose: item.purpose ?? "", })); - if (format === "json") { - emitResult({ items, total, request_id: response.request_id }, format); - return; - } - - // text / quiet - if (items.length === 0) { - emitBare("No dataset files found."); - return; + if (settings.quiet) { + for (const item of items) emitBare(item.file_id); + } else { + emitResult({ items, total, request_id: response.request_id }, "json"); } - const headers = ["FILE_ID", "NAME", "SIZE", "PURPOSE"]; - const rows = items.map((i) => [i.file_id, i.name, i.size, i.purpose]); - for (const line of formatTable(headers, rows)) emitBare(line); - if (total !== undefined) emitBare(`\nTotal: ${total}`); - emitRequestId(response.request_id, settings.quiet); }, }); diff --git a/packages/commands/src/commands/dataset/upload.ts b/packages/commands/src/commands/dataset/upload.ts index a3c61045..13635095 100644 --- a/packages/commands/src/commands/dataset/upload.ts +++ b/packages/commands/src/commands/dataset/upload.ts @@ -1,6 +1,5 @@ import { defineCommand, - detectOutputFormat, uploadDataset, validateDataset, parseDatasetSchemaFlag, @@ -11,7 +10,7 @@ import { ExitCode, type FlagsDef, } from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { emitResult, emitBare } from "bailian-cli-runtime"; const UPLOAD_FLAGS = { file: { @@ -80,7 +79,6 @@ export default defineCommand({ `Supported schemas: chatml, dpo, cpt, tts, image.`, ); } - const format = detectOutputFormat(settings.output); // Image schema allows larger ZIPs (1 GB vs 300 MB for text). const isMediaSchema = schema === "image"; @@ -129,7 +127,7 @@ export default defineCommand({ validate: !flags.noValidate, schema: schema ?? "auto", }, - format, + "json", ); return; } @@ -142,11 +140,8 @@ export default defineCommand({ if (settings.quiet) { emitBare(file.file_id); - } else if (format === "text") { - emitBare(`Uploaded ${file.name} → file_id=${file.file_id}`); - emitRequestId(request_id, settings.quiet); } else { - emitResult({ ...file, request_id }, format); + emitResult({ ...file, request_id }, "json"); } }, }); diff --git a/packages/commands/src/commands/dataset/validate.ts b/packages/commands/src/commands/dataset/validate.ts index 8bf0d292..142fbdf5 100644 --- a/packages/commands/src/commands/dataset/validate.ts +++ b/packages/commands/src/commands/dataset/validate.ts @@ -1,26 +1,13 @@ import { defineCommand, - detectOutputFormat, validateDataset, parseDatasetSchemaFlag, - formatIssue, BailianError, ExitCode, - type ValidationResult, type FlagsDef, } from "bailian-cli-core"; import { emitResult, emitBare } from "bailian-cli-runtime"; -function formatStats(result: ValidationResult): string[] { - const out: string[] = []; - if (result.stats.totalRecords !== undefined) out.push(`records: ${result.stats.totalRecords}`); - if (result.stats.sampledRecords !== undefined) - out.push(`sampled: ${result.stats.sampledRecords}`); - if (result.stats.bytes !== undefined) out.push(`bytes: ${result.stats.bytes}`); - if (result.stats.durationMs !== undefined) out.push(`took: ${result.stats.durationMs}ms`); - return out; -} - const VALIDATE_FLAGS = { file: { type: "string", @@ -79,8 +66,6 @@ export default defineCommand({ `Supported schemas: chatml, dpo, cpt, tts, image.`, ); } - const format = detectOutputFormat(settings.output); - if (settings.dryRun) { emitResult( { @@ -89,38 +74,17 @@ export default defineCommand({ full: flags.fullValidate, schema: schema ?? "auto", }, - format, + "json", ); return; } const result = await validateDataset(filePath, { fullValidate: flags.fullValidate, schema }); - if (format === "json") { - // For json output we always emit the structured result, exit code conveys validity. - emitResult(result, format); - } else if (settings.quiet) { + if (settings.quiet) { emitBare(result.valid ? "ok" : "fail"); } else { - const status = result.valid ? "PASSED" : "FAILED"; - emitBare(`Dataset validation ${status} for ${result.filePath}`); - const stats = formatStats(result); - if (stats.length) emitBare(` ${stats.join(" · ")}`); - - if (result.errors.length) { - emitBare(`Errors (${result.errors.length}):`); - for (const error of result.errors.slice(0, 20)) emitBare(formatIssue(error)); - if (result.errors.length > 20) { - emitBare(` … and ${result.errors.length - 20} more.`); - } - } - if (result.warnings.length) { - emitBare(`Warnings (${result.warnings.length}):`); - for (const warning of result.warnings.slice(0, 10)) emitBare(formatIssue(warning)); - if (result.warnings.length > 10) { - emitBare(` … and ${result.warnings.length - 10} more.`); - } - } + emitResult(result, "json"); } if (!result.valid) { diff --git a/packages/commands/src/commands/deploy/create.ts b/packages/commands/src/commands/deploy/create.ts index 96c75d92..15f6b2a0 100644 --- a/packages/commands/src/commands/deploy/create.ts +++ b/packages/commands/src/commands/deploy/create.ts @@ -1,6 +1,5 @@ import { defineCommand, - detectOutputFormat, createDeployment, pickPlanStrategy, STRATEGIES, @@ -11,7 +10,7 @@ import { type CommandContext, type FlagsDef, } from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { emitResult, emitBare } from "bailian-cli-runtime"; const CREATE_FLAGS = { model: { @@ -122,7 +121,6 @@ async function runCreate( const model = flags.model as string; const name = flags.name as string; const plan = (flags.plan as string | undefined) || defaultDeployPlan(modality); - const format = detectOutputFormat(settings.output); // Plan-specific behaviour is owned by core `plans.ts`. The strategy resolves // the plan-specific body fragment (mu may auto-pick a template from the @@ -146,7 +144,7 @@ async function runCreate( }; if (settings.dryRun) { - emitResult({ action: "deploy.create", body }, format); + emitResult({ action: "deploy.create", body }, "json"); return; } @@ -155,17 +153,8 @@ async function runCreate( if (settings.quiet) { emitBare(deployment?.deployed_model ?? ""); - } else if (format === "text") { - emitBare(`Created deployment.`); - if (deployment?.deployed_model) emitBare(` deployed_model: ${deployment.deployed_model}`); - if (deployment?.status) emitBare(` status: ${deployment.status}`); - if (deployment?.plan) emitBare(` plan: ${deployment.plan}`); - emitBare( - `\nNext: track readiness with: ${identity.binName} deploy get --deployed-model ${deployment?.deployed_model ?? ""}`, - ); - emitRequestId(response.request_id, settings.quiet); } else { - emitResult(response, format); + emitResult(response, "json"); } } diff --git a/packages/commands/src/commands/deploy/delete.ts b/packages/commands/src/commands/deploy/delete.ts index bd6dd85e..d3f386d8 100644 --- a/packages/commands/src/commands/deploy/delete.ts +++ b/packages/commands/src/commands/deploy/delete.ts @@ -1,13 +1,12 @@ import { defineCommand, - detectOutputFormat, deleteDeployment, getDeployment, BailianError, ExitCode, type FlagsDef, } from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { emitResult, emitBare } from "bailian-cli-runtime"; const DELETE_FLAGS = { deployedModel: { @@ -38,10 +37,9 @@ export default defineCommand({ async run(ctx) { const { settings, flags } = ctx; const deployedModel = flags.deployedModel; - const format = detectOutputFormat(settings.output); if (settings.dryRun) { - emitResult({ action: "deploy.delete", deployed_model: deployedModel }, format); + emitResult({ action: "deploy.delete", deployed_model: deployedModel }, "json"); return; } @@ -55,7 +53,8 @@ export default defineCommand({ if (status && status !== "STOPPED" && status !== "FAILED") { throw new BailianError( `Deployment ${deployedModel} is ${status}. Only STOPPED / FAILED deployments can be deleted. ` + - `Stop it first via the platform console, or pass --skip-precheck to attempt deletion anyway.`, + `Run \`bl deploy pause --deployed-model ${deployedModel}\` to pause it first, ` + + `or pass --skip-precheck to attempt deletion anyway.`, ExitCode.USAGE, ); } @@ -69,11 +68,8 @@ export default defineCommand({ if (settings.quiet) { emitBare(deployedModel); - } else if (format === "text") { - emitBare(`Deleted ${deployedModel}.`); - emitRequestId(response.request_id, settings.quiet); } else { - emitResult(response, format); + emitResult(response, "json"); } }, }); diff --git a/packages/commands/src/commands/deploy/get.ts b/packages/commands/src/commands/deploy/get.ts index 9c53e96f..09bff852 100644 --- a/packages/commands/src/commands/deploy/get.ts +++ b/packages/commands/src/commands/deploy/get.ts @@ -1,5 +1,5 @@ -import { defineCommand, detectOutputFormat, getDeployment, type FlagsDef } from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { defineCommand, getDeployment, type FlagsDef } from "bailian-cli-core"; +import { emitResult } from "bailian-cli-runtime"; const GET_FLAGS = { deployedModel: { @@ -22,10 +22,9 @@ export default defineCommand({ async run(ctx) { const { settings, flags } = ctx; const deployedModel = flags.deployedModel; - const format = detectOutputFormat(settings.output); if (settings.dryRun) { - emitResult({ action: "deploy.get", deployed_model: deployedModel }, format); + emitResult({ action: "deploy.get", deployed_model: deployedModel }, "json"); return; } @@ -33,7 +32,7 @@ export default defineCommand({ const deployment = response.output ?? response.data; if (!deployment) { - emitBare(`No data returned for ${deployedModel}`); + emitResult({ deployed_model: deployedModel, request_id: response.request_id }, "json"); return; } @@ -57,18 +56,6 @@ export default defineCommand({ if (deployment.gmt_create) item.created_at = deployment.gmt_create; if (deployment.gmt_modified) item.updated_at = deployment.gmt_modified; - if (format === "json") { - emitResult({ ...item, request_id: response.request_id }, format); - return; - } - - // text / quiet — fixed-width label column for alignment - const label = (key: string) => `${key}:`.padEnd(18); - for (const [key, value] of Object.entries(item)) { - if (value === "" || value === undefined) continue; - const display = typeof value === "string" ? value : JSON.stringify(value); - emitBare(`${label(key)}${display}`); - } - emitRequestId(response.request_id, settings.quiet); + emitResult({ ...item, request_id: response.request_id }, "json"); }, }); diff --git a/packages/commands/src/commands/deploy/list.ts b/packages/commands/src/commands/deploy/list.ts index e7c29173..944038a5 100644 --- a/packages/commands/src/commands/deploy/list.ts +++ b/packages/commands/src/commands/deploy/list.ts @@ -1,10 +1,5 @@ -import { - defineCommand, - detectOutputFormat, - listDeployments, - type FlagsDef, -} from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId, formatTable } from "bailian-cli-runtime"; +import { defineCommand, listDeployments, type FlagsDef } from "bailian-cli-core"; +import { emitResult } from "bailian-cli-runtime"; const LIST_FLAGS = { page: { type: "number", valueHint: "", description: "Page number (default: 1)" }, @@ -28,13 +23,12 @@ export default defineCommand({ exampleArgs: ["", "--status RUNNING", "--page-size 20 --output json"], async run(ctx) { const { settings, flags } = ctx; - const format = detectOutputFormat(settings.output); const status = flags.status || undefined; if (settings.dryRun) { emitResult( { action: "deploy.list", page: flags.page, page_size: flags.pageSize, status }, - format, + "json", ); return; } @@ -57,27 +51,6 @@ export default defineCommand({ created_at: item.gmt_create ?? "", })); - if (format === "json") { - emitResult({ items, total, request_id: response.request_id }, format); - return; - } - - // text / quiet - if (items.length === 0) { - emitBare("No deployments found."); - return; - } - const headers = ["DEPLOYED_MODEL", "MODEL_NAME", "STATUS", "PLAN", "CAPACITY", "CREATED_AT"]; - const rows = items.map((item) => [ - item.deployed_model, - item.model_name, - item.status, - item.plan, - item.capacity, - item.created_at, - ]); - for (const line of formatTable(headers, rows)) emitBare(line); - if (total !== undefined) emitBare(`\nTotal: ${total}`); - emitRequestId(response.request_id, settings.quiet); + emitResult({ items, total, request_id: response.request_id }, "json"); }, }); diff --git a/packages/commands/src/commands/deploy/models.ts b/packages/commands/src/commands/deploy/models.ts index d1ea972e..b2c81d4c 100644 --- a/packages/commands/src/commands/deploy/models.ts +++ b/packages/commands/src/commands/deploy/models.ts @@ -1,10 +1,5 @@ -import { - defineCommand, - detectOutputFormat, - listDeployableModels, - type FlagsDef, -} from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId, formatTable } from "bailian-cli-runtime"; +import { defineCommand, listDeployableModels, type FlagsDef } from "bailian-cli-core"; +import { emitResult } from "bailian-cli-runtime"; const MODELS_FLAGS = { page: { type: "number", valueHint: "", description: "Page number (default: 1)" }, @@ -39,7 +34,6 @@ export default defineCommand({ ], async run(ctx) { const { settings, flags } = ctx; - const format = detectOutputFormat(settings.output); // Default version to v1.0 — without it, the API returns the legacy catalog // (only old fine-tune outputs). Pass --catalog-version "" to opt out. const version = flags.catalogVersion === "" ? undefined : (flags.catalogVersion ?? "v1.0"); @@ -54,7 +48,7 @@ export default defineCommand({ version, model_source: modelSource, }, - format, + "json", ); return; } @@ -72,102 +66,55 @@ export default defineCommand({ // Two response shapes: // - custom (fine-tuned): top-level supported_plans: string[] // - base (catalog): plans: [{plan, templates?, cu_specs?}] - // For json: surface the deployment-relevant fields preserved as a tree, so + // Surface the deployment-relevant fields preserved as a tree, so // downstream tooling can drive `bl deploy create --deploy-spec <…>` - // without a second round-trip. For text: keep the compact one-line summary. - if (format === "json") { - const items = models.map((model) => { - const out: Record = { - model_name: model.model_name ?? "", - }; - if (model.base_model) out.base_model = model.base_model; - if (model.model_source) out.model_source = model.model_source; - if (model.supported_plans && model.supported_plans.length > 0) { - out.supported_plans = model.supported_plans; - } - if (model.plans && model.plans.length > 0) { - out.plans = model.plans.map((plan) => { - const planEntry: Record = { plan: plan.plan ?? "" }; - if (plan.cu_specs && plan.cu_specs.length > 0) { - planEntry.cu_specs = plan.cu_specs; - } - if (plan.templates && plan.templates.length > 0) { - // Pull the top 6 fields most useful for `bl deploy create`. - // Drop noisy/redundant: template_source, template_type, - // template_version, deploy_spec (typically == template_id). - planEntry.templates = plan.templates.map((template) => { - const tpl: Record = {}; - if (template.template_id) tpl.template_id = template.template_id; - if (template.template_name) tpl.template_name = template.template_name; - if (template.charge_type) tpl.charge_type = template.charge_type; - // Flatten roles.unified for the common COUPLED case. - const unified = template.roles?.unified; - if (unified?.model_unit_spec) tpl.model_unit_spec = unified.model_unit_spec; - if (unified?.capacity_unit_per_instance !== undefined) - tpl.capacity_unit_per_instance = unified.capacity_unit_per_instance; - // Preserve split-role configs (SEPERATED) as-is so callers - // can still drive prefill/decode sizing. - if (template.roles?.prefill || template.roles?.decode) { - tpl.roles = { - prefill: template.roles?.prefill, - decode: template.roles?.decode, - }; - } - if (template.template_desc) tpl.template_desc = template.template_desc; - return tpl; - }); - } - return planEntry; - }); - } - return out; - }); - emitResult({ items, total, request_id: response.request_id }, format); - return; - } - - // text / quiet — keep the compact single-line summary table. - const textItems = models.map((model) => { - let plansSummary = ""; - if (model.supported_plans && model.supported_plans.length > 0) { - plansSummary = model.supported_plans.join(","); - } else if (model.plans && model.plans.length > 0) { - plansSummary = model.plans - .map((plan) => { - const planName = plan.plan ?? "?"; - if (plan.templates && plan.templates.length > 0) { - return `${planName}(${plan.templates.length}t)`; - } - if (plan.cu_specs && plan.cu_specs.length > 0) { - return `${planName}(${plan.cu_specs.join("/")})`; - } - return planName; - }) - .join(","); - } else { - plansSummary = "-"; - } - return { + // without a second round-trip. + const items = models.map((model) => { + const out: Record = { model_name: model.model_name ?? "", - base_model: model.base_model ?? "", - source: model.model_source ?? "", - plans: plansSummary, }; + if (model.base_model) out.base_model = model.base_model; + if (model.model_source) out.model_source = model.model_source; + if (model.supported_plans && model.supported_plans.length > 0) { + out.supported_plans = model.supported_plans; + } + if (model.plans && model.plans.length > 0) { + out.plans = model.plans.map((plan) => { + const planEntry: Record = { plan: plan.plan ?? "" }; + if (plan.cu_specs && plan.cu_specs.length > 0) { + planEntry.cu_specs = plan.cu_specs; + } + if (plan.templates && plan.templates.length > 0) { + // Pull the top 6 fields most useful for `bl deploy create`. + // Drop noisy/redundant: template_source, template_type, + // template_version, deploy_spec (typically == template_id). + planEntry.templates = plan.templates.map((template) => { + const tpl: Record = {}; + if (template.template_id) tpl.template_id = template.template_id; + if (template.template_name) tpl.template_name = template.template_name; + if (template.charge_type) tpl.charge_type = template.charge_type; + // Flatten roles.unified for the common COUPLED case. + const unified = template.roles?.unified; + if (unified?.model_unit_spec) tpl.model_unit_spec = unified.model_unit_spec; + if (unified?.capacity_unit_per_instance !== undefined) + tpl.capacity_unit_per_instance = unified.capacity_unit_per_instance; + // Preserve split-role configs (SEPERATED) as-is so callers + // can still drive prefill/decode sizing. + if (template.roles?.prefill || template.roles?.decode) { + tpl.roles = { + prefill: template.roles?.prefill, + decode: template.roles?.decode, + }; + } + if (template.template_desc) tpl.template_desc = template.template_desc; + return tpl; + }); + } + return planEntry; + }); + } + return out; }); - - if (textItems.length === 0) { - emitBare("No deployable models found."); - return; - } - const headers = ["MODEL_NAME", "BASE_MODEL", "SOURCE", "PLANS"]; - const rows = textItems.map((item) => [ - item.model_name, - item.base_model, - item.source, - item.plans, - ]); - for (const line of formatTable(headers, rows)) emitBare(line); - if (total !== undefined) emitBare(`\nTotal: ${total}`); - emitRequestId(response.request_id, settings.quiet); + emitResult({ items, total, request_id: response.request_id }, "json"); }, }); diff --git a/packages/commands/src/commands/deploy/pause.ts b/packages/commands/src/commands/deploy/pause.ts new file mode 100644 index 00000000..23d36c69 --- /dev/null +++ b/packages/commands/src/commands/deploy/pause.ts @@ -0,0 +1,85 @@ +import { + defineCommand, + stopModelService, + listIndependentDeployedModels, + findDeploymentEntry, + BailianError, + ExitCode, + type FlagsDef, +} from "bailian-cli-core"; +import { emitResult, emitBare } from "bailian-cli-runtime"; + +const PAUSE_FLAGS = { + deployedModel: { + type: "string", + valueHint: "", + description: "Deployed model identifier (required)", + required: true, + }, + skipPrecheck: { + type: "switch", + description: "Skip the local RUNNING/PENDING status precheck", + }, +} satisfies FlagsDef; + +/** + * `bl deploy pause` — pause a running deployment. + * + * Takes the model service offline so it no longer serves inference requests. + * For mu/ptu plans, billing stops while paused. + * Precheck: status must be RUNNING or PENDING. + */ +export default defineCommand({ + description: "Pause a running model deployment (stops billing for mu/ptu)", + auth: "console", + usageArgs: "--deployed-model [--skip-precheck]", + flags: PAUSE_FLAGS, + exampleArgs: [ + "--deployed-model dep-...", + "--deployed-model dep-... --skip-precheck", + "--deployed-model dep-... --dry-run", + ], + notes: [ + "While paused, billing ceases for mu/ptu plans. Use `deploy resume` to bring it back online or `deploy delete` to remove.", + "Precheck verifies status is RUNNING/PENDING before issuing the pause; pass --skip-precheck to bypass.", + ], + async run(ctx) { + const { settings, flags } = ctx; + const deployedModel = flags.deployedModel; + + if (settings.dryRun) { + emitResult({ action: "deploy.pause", deployed_model: deployedModel }, "json"); + return; + } + + // Precheck: verify the deployment is in a pausable state. + if (!flags.skipPrecheck) { + try { + const entries = await listIndependentDeployedModels(ctx.client); + const entry = findDeploymentEntry(entries, deployedModel); + if (entry) { + const status = (entry.status ?? "").toUpperCase(); + if (status && status !== "RUNNING" && status !== "PENDING") { + throw new BailianError( + `Deployment ${deployedModel} is ${status}. Only RUNNING / PENDING deployments can be paused. ` + + `Pass --skip-precheck to attempt the pause anyway.`, + ExitCode.USAGE, + ); + } + } + // If entry not found in list, proceed — the server will surface the real error. + } catch (error) { + if (error instanceof BailianError) throw error; + // If the list call itself failed, proceed and let the API call surface the error. + } + } + + const response = await stopModelService(ctx.client, deployedModel); + + if (settings.quiet) { + emitBare(deployedModel); + } else { + emitResult({ deployed_model: deployedModel, action: "pause", ...response }, "json"); + } + }, +}); diff --git a/packages/commands/src/commands/deploy/resume.ts b/packages/commands/src/commands/deploy/resume.ts new file mode 100644 index 00000000..7e0cb287 --- /dev/null +++ b/packages/commands/src/commands/deploy/resume.ts @@ -0,0 +1,84 @@ +import { + defineCommand, + startModelService, + listIndependentDeployedModels, + findDeploymentEntry, + BailianError, + ExitCode, + type FlagsDef, +} from "bailian-cli-core"; +import { emitResult, emitBare } from "bailian-cli-runtime"; + +const RESUME_FLAGS = { + deployedModel: { + type: "string", + valueHint: "", + description: "Deployed model identifier (required)", + required: true, + }, + skipPrecheck: { + type: "switch", + description: "Skip the local STOPPED status precheck", + }, +} satisfies FlagsDef; + +/** + * `bl deploy resume` — resume a paused deployment. + * + * Brings the model service back online so it can serve inference requests. + * Precheck: status must be STOPPED. + */ +export default defineCommand({ + description: "Resume a paused model deployment (brings service back online)", + auth: "console", + usageArgs: "--deployed-model [--skip-precheck]", + flags: RESUME_FLAGS, + exampleArgs: [ + "--deployed-model dep-...", + "--deployed-model dep-... --skip-precheck", + "--deployed-model dep-... --dry-run", + ], + notes: [ + "Precheck verifies status is STOPPED before issuing the resume; pass --skip-precheck to bypass.", + "For mu/ptu plans, billing resumes once the service is back online.", + ], + async run(ctx) { + const { settings, flags } = ctx; + const deployedModel = flags.deployedModel; + + if (settings.dryRun) { + emitResult({ action: "deploy.resume", deployed_model: deployedModel }, "json"); + return; + } + + // Precheck: verify the deployment is in a resumable state. + if (!flags.skipPrecheck) { + try { + const entries = await listIndependentDeployedModels(ctx.client); + const entry = findDeploymentEntry(entries, deployedModel); + if (entry) { + const status = (entry.status ?? "").toUpperCase(); + if (status && status !== "STOPPED") { + throw new BailianError( + `Deployment ${deployedModel} is ${status}. Only STOPPED deployments can be resumed. ` + + `Pass --skip-precheck to attempt the resume anyway.`, + ExitCode.USAGE, + ); + } + } + // If entry not found in list, proceed — the server will surface the real error. + } catch (error) { + if (error instanceof BailianError) throw error; + // If the list call itself failed, proceed and let the API call surface the error. + } + } + + const response = await startModelService(ctx.client, deployedModel); + + if (settings.quiet) { + emitBare(deployedModel); + } else { + emitResult({ deployed_model: deployedModel, action: "resume", ...response }, "json"); + } + }, +}); diff --git a/packages/commands/src/commands/deploy/scale.ts b/packages/commands/src/commands/deploy/scale.ts index 805f9c57..c6400a4a 100644 --- a/packages/commands/src/commands/deploy/scale.ts +++ b/packages/commands/src/commands/deploy/scale.ts @@ -1,10 +1,5 @@ -import { - defineCommand, - detectOutputFormat, - scaleDeployment, - type FlagsDef, -} from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { defineCommand, scaleDeployment, type FlagsDef } from "bailian-cli-core"; +import { emitResult, emitBare } from "bailian-cli-runtime"; const SCALE_FLAGS = { deployedModel: { @@ -52,7 +47,6 @@ export default defineCommand({ async run(ctx) { const { settings, flags } = ctx; const deployedModel = flags.deployedModel; - const format = detectOutputFormat(settings.output); const body: Record = {}; if (flags.capacity !== undefined) body.capacity = flags.capacity; @@ -60,21 +54,16 @@ export default defineCommand({ if (flags.outputTpm !== undefined) body.output_tpm = flags.outputTpm; if (settings.dryRun) { - emitResult({ action: "deploy.scale", deployed_model: deployedModel, body }, format); + emitResult({ action: "deploy.scale", deployed_model: deployedModel, body }, "json"); return; } const response = await scaleDeployment(ctx.client, deployedModel, body); - const deployment = response.output ?? response.data; if (settings.quiet) { emitBare(deployedModel); - } else if (format === "text") { - const cap = deployment?.capacity !== undefined ? ` (capacity=${deployment.capacity})` : ""; - emitBare(`Scaled ${deployedModel}${cap}.`); - emitRequestId(response.request_id, settings.quiet); } else { - emitResult(response, format); + emitResult(response, "json"); } }, }); diff --git a/packages/commands/src/commands/deploy/update.ts b/packages/commands/src/commands/deploy/update.ts index 0bbb29d2..6c3ab896 100644 --- a/packages/commands/src/commands/deploy/update.ts +++ b/packages/commands/src/commands/deploy/update.ts @@ -1,10 +1,5 @@ -import { - defineCommand, - detectOutputFormat, - updateDeployment, - type FlagsDef, -} from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { defineCommand, updateDeployment, type FlagsDef } from "bailian-cli-core"; +import { emitResult, emitBare } from "bailian-cli-runtime"; const UPDATE_FLAGS = { deployedModel: { @@ -48,31 +43,22 @@ export default defineCommand({ async run(ctx) { const { settings, flags } = ctx; const deployedModel = flags.deployedModel; - const format = detectOutputFormat(settings.output); const body: Record = {}; if (flags.rpmLimit !== undefined) body.rpm_limit = flags.rpmLimit; if (flags.tpmLimit !== undefined) body.tpm_limit = flags.tpmLimit; if (settings.dryRun) { - emitResult({ action: "deploy.update", deployed_model: deployedModel, body }, format); + emitResult({ action: "deploy.update", deployed_model: deployedModel, body }, "json"); return; } const response = await updateDeployment(ctx.client, deployedModel, body); - const deployment = response.output ?? response.data; if (settings.quiet) { emitBare(deployedModel); - } else if (format === "text") { - const parts: string[] = []; - if (deployment?.rpm_limit !== undefined) parts.push(`rpm_limit=${deployment.rpm_limit}`); - if (deployment?.tpm_limit !== undefined) parts.push(`tpm_limit=${deployment.tpm_limit}`); - const summary = parts.length ? ` (${parts.join(", ")})` : ""; - emitBare(`Updated ${deployedModel}${summary}.`); - emitRequestId(response.request_id, settings.quiet); } else { - emitResult(response, format); + emitResult(response, "json"); } }, }); diff --git a/packages/commands/src/commands/finetune/cancel.ts b/packages/commands/src/commands/finetune/cancel.ts index 789a3b03..966ce3e7 100644 --- a/packages/commands/src/commands/finetune/cancel.ts +++ b/packages/commands/src/commands/finetune/cancel.ts @@ -1,5 +1,5 @@ -import { defineCommand, detectOutputFormat, cancelFineTune, type FlagsDef } from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { defineCommand, cancelFineTune, type FlagsDef } from "bailian-cli-core"; +import { emitResult, emitBare } from "bailian-cli-runtime"; const CANCEL_FLAGS = { jobId: { @@ -23,24 +23,18 @@ export default defineCommand({ async run(ctx) { const { settings, flags } = ctx; const jobId = flags.jobId; - const format = detectOutputFormat(settings.output); if (settings.dryRun) { - emitResult({ action: "finetune.cancel", job_id: jobId }, format); + emitResult({ action: "finetune.cancel", job_id: jobId }, "json"); return; } const response = await cancelFineTune(ctx.client, jobId); - const job = response.output ?? response.data; if (settings.quiet) { emitBare(jobId); - } else if (format === "text") { - const status = job?.status ? ` (status=${job.status})` : ""; - emitBare(`Cancelled ${jobId}${status}.`); - emitRequestId(response.request_id, settings.quiet); } else { - emitResult(response, format); + emitResult(response, "json"); } }, }); diff --git a/packages/commands/src/commands/finetune/capability.ts b/packages/commands/src/commands/finetune/capability.ts index 1bf72034..e9091a93 100644 --- a/packages/commands/src/commands/finetune/capability.ts +++ b/packages/commands/src/commands/finetune/capability.ts @@ -1,6 +1,5 @@ import { defineCommand, - detectOutputFormat, fetchModelList, fetchModelCapability, listSupportedTrainingTypes, @@ -43,17 +42,6 @@ async function fetchAllFoundationModels(settings: Settings): Promise = { - full: "full-parameter", - lora: "LoRA", -}; - -function describeTrainingType(value: string): string { - if (!isTrainingTypeCli(value)) return value; - const { method, variant } = trainingTypeMethodVariant(value); - return `${VARIANT_LABEL[variant] ?? variant} ${method.toUpperCase()}`; -} - const CAPABILITY_FLAGS = { model: { type: "string", @@ -95,7 +83,6 @@ export default defineCommand({ const { settings, flags } = ctx; const model = flags.model || undefined; const trainingType = flags.trainingType || undefined; - const format = detectOutputFormat(settings.output); if (settings.dryRun) { emitResult( @@ -104,7 +91,7 @@ export default defineCommand({ model, training_type: trainingType, }, - format, + "json", ); return; } @@ -113,7 +100,7 @@ export default defineCommand({ if (model) { const capability = await fetchModelCapability(settings, model); if (!capability) { - emitBare(`No foundation model found matching "${model}".`); + emitResult({ model, error: `No foundation model found matching "${model}".` }, "json"); return; } const supported = listSupportedTrainingTypes(capability); @@ -121,23 +108,15 @@ export default defineCommand({ for (const value of supported) emitBare(value); return; } - if (format !== "text") { - emitResult( - { - model: capability.model ?? model, - supported, - supports: capability.supports, - trainingTypes: capability.trainingTypes, - }, - format, - ); - return; - } - emitBare(`${capability.model ?? model}`); - emitBare(supported.length ? "Supported training types:" : "No supported training types."); - for (const value of supported) { - emitBare(` ${value.padEnd(10)} ${describeTrainingType(value)}`); - } + emitResult( + { + model: capability.model ?? model, + supported, + supports: capability.supports, + trainingTypes: capability.trainingTypes, + }, + "json", + ); return; } @@ -162,20 +141,15 @@ export default defineCommand({ for (const entry of matched) emitBare(entry.model); return; } - if (format !== "text") { - emitResult( - { - training_type: trainingType, - method, - variant, - count: matched.length, - models: matched, - }, - format, - ); - return; - } - emitBare(`Models supporting ${trainingType} (${method} / ${variant}): ${matched.length}`); - for (const entry of matched) emitBare(` ${entry.model}`); + emitResult( + { + training_type: trainingType, + method, + variant, + count: matched.length, + models: matched, + }, + "json", + ); }, }); diff --git a/packages/commands/src/commands/finetune/checkpoints.ts b/packages/commands/src/commands/finetune/checkpoints.ts index 7c05a363..769cf72d 100644 --- a/packages/commands/src/commands/finetune/checkpoints.ts +++ b/packages/commands/src/commands/finetune/checkpoints.ts @@ -1,10 +1,5 @@ -import { - defineCommand, - detectOutputFormat, - listCheckpoints, - type FlagsDef, -} from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId, formatTable } from "bailian-cli-runtime"; +import { defineCommand, listCheckpoints, type FlagsDef } from "bailian-cli-core"; +import { emitResult } from "bailian-cli-runtime"; const CHECKPOINTS_FLAGS = { jobId: { @@ -15,6 +10,8 @@ const CHECKPOINTS_FLAGS = { }, } satisfies FlagsDef; +const EXPIRY_WARN_THRESHOLD_MS = 72 * 60 * 60 * 1000; // 72 hours + export default defineCommand({ description: "List checkpoints produced by a fine-tune job", auth: "apiKey", @@ -22,16 +19,15 @@ export default defineCommand({ flags: CHECKPOINTS_FLAGS, exampleArgs: ["--job-id ft-xxx", "--job-id ft-xxx --output json"], notes: [ - "Use the returned `checkpoint` value with `finetune export` to publish", - "a deployable model.", + "`model_name` (shown for SUCCEEDED checkpoints) is the direct input for `deploy create --model`.", + "Checkpoints expire ~15 days after creation; `expire_time` shows the deadline. Export or deploy before expiry.", ], async run(ctx) { const { settings, flags } = ctx; const jobId = flags.jobId; - const format = detectOutputFormat(settings.output); if (settings.dryRun) { - emitResult({ action: "finetune.checkpoints", job_id: jobId }, format); + emitResult({ action: "finetune.checkpoints", job_id: jobId }, "json"); return; } @@ -44,22 +40,26 @@ export default defineCommand({ checkpoint: item.checkpoint ?? item.checkpoint_id ?? "", step: item.step !== undefined ? String(item.step) : "", status: item.status ?? "", + model_name: item.model_name ?? "", + expire_time: item.expire_time ?? "", })); - if (format === "json") { - emitResult({ items, total, request_id: response.request_id }, format); - return; - } + emitResult({ items, total, request_id: response.request_id }, "json"); - // text / quiet - if (items.length === 0) { - emitBare("No checkpoints found."); - return; + // Near-expiry warning: check if any non-expired checkpoint is within 72h of expiry. + const now = Date.now(); + const expiringSoon = items.filter((item) => { + if (!item.expire_time) return false; + const deadline = new Date(item.expire_time).getTime(); + if (Number.isNaN(deadline)) return false; + const remaining = deadline - now; + return remaining > 0 && remaining < EXPIRY_WARN_THRESHOLD_MS; + }); + if (expiringSoon.length > 0) { + process.stderr.write( + `\n[warning] ${expiringSoon.length} checkpoint(s) will expire within 72 hours. ` + + "Export or deploy before expiry to avoid losing the model artifact.\n", + ); } - const headers = ["CHECKPOINT", "STEP", "STATUS"]; - const rows = items.map((i) => [i.checkpoint, i.step, i.status]); - for (const line of formatTable(headers, rows)) emitBare(line); - emitBare(`\nTotal: ${total}`); - emitRequestId(response.request_id, settings.quiet); }, }); diff --git a/packages/commands/src/commands/finetune/create.ts b/packages/commands/src/commands/finetune/create.ts index 6fbb54f0..f9bbaf0e 100644 --- a/packages/commands/src/commands/finetune/create.ts +++ b/packages/commands/src/commands/finetune/create.ts @@ -1,6 +1,5 @@ import { defineCommand, - detectOutputFormat, createFineTune, getDataset, uploadDataset, @@ -27,7 +26,7 @@ import { } from "bailian-cli-core"; import { existsSync, statSync } from "fs"; import { basename } from "path"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { emitResult, emitBare } from "bailian-cli-runtime"; /** * A `--datasets` / `--validations` token is treated as a local file to upload @@ -606,8 +605,6 @@ async function runCreate( if (modelName) body.model_name = modelName; if (suffix) body.finetuned_output_suffix = suffix; - const format = detectOutputFormat(settings.output); - if (settings.dryRun) { const pending = [ ...training.localPaths.map((path) => ({ field: "datasets", path })), @@ -617,7 +614,7 @@ async function runCreate( pending.length > 0 ? { action: "finetune.create", body, pending_uploads: pending } : { action: "finetune.create", body }, - format, + "json", ); return; } @@ -627,16 +624,8 @@ async function runCreate( if (settings.quiet) { if (job?.job_id) emitBare(job.job_id); - } else if (format === "text") { - if (job?.job_id) { - emitBare(`Created fine-tune job: ${job.job_id}`); - if (job.status) emitBare(`Status: ${job.status}`); - emitRequestId(response.request_id, settings.quiet); - } else { - emitResult(response, format); - } } else { - emitResult(response, format); + emitResult(response, "json"); } } diff --git a/packages/commands/src/commands/finetune/delete.ts b/packages/commands/src/commands/finetune/delete.ts index 102fda61..f18a13f7 100644 --- a/packages/commands/src/commands/finetune/delete.ts +++ b/packages/commands/src/commands/finetune/delete.ts @@ -1,5 +1,5 @@ -import { defineCommand, detectOutputFormat, deleteFineTune, type FlagsDef } from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { defineCommand, deleteFineTune, type FlagsDef } from "bailian-cli-core"; +import { emitResult, emitBare } from "bailian-cli-runtime"; const DELETE_FLAGS = { jobId: { @@ -23,10 +23,9 @@ export default defineCommand({ async run(ctx) { const { settings, flags } = ctx; const jobId = flags.jobId; - const format = detectOutputFormat(settings.output); if (settings.dryRun) { - emitResult({ action: "finetune.delete", job_id: jobId }, format); + emitResult({ action: "finetune.delete", job_id: jobId }, "json"); return; } @@ -34,11 +33,8 @@ export default defineCommand({ if (settings.quiet) { emitBare(jobId); - } else if (format === "text") { - emitBare(`Deleted ${jobId}.`); - emitRequestId(response.request_id, settings.quiet); } else { - emitResult(response, format); + emitResult(response, "json"); } }, }); diff --git a/packages/commands/src/commands/finetune/export.ts b/packages/commands/src/commands/finetune/export.ts index 7e52de0b..97bb0aa7 100644 --- a/packages/commands/src/commands/finetune/export.ts +++ b/packages/commands/src/commands/finetune/export.ts @@ -1,10 +1,5 @@ -import { - defineCommand, - detectOutputFormat, - exportCheckpoint, - type FlagsDef, -} from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { defineCommand, exportCheckpoint, type FlagsDef } from "bailian-cli-core"; +import { emitResult, emitBare } from "bailian-cli-runtime"; const EXPORT_FLAGS = { jobId: { @@ -39,11 +34,10 @@ export default defineCommand({ "explicit export is the canonical path for non-best checkpoints.", ], async run(ctx) { - const { identity, settings, flags } = ctx; + const { settings, flags } = ctx; const jobId = flags.jobId; const checkpoint = flags.checkpoint; const modelName = flags.modelName; - const format = detectOutputFormat(settings.output); if (settings.dryRun) { emitResult( @@ -53,7 +47,7 @@ export default defineCommand({ checkpoint, model_name: modelName, }, - format, + "json", ); return; } @@ -64,14 +58,8 @@ export default defineCommand({ if (settings.quiet) { emitBare(exported); - } else if (format === "text") { - emitBare(`Exported ${jobId} / ${checkpoint} → model_name=${exported}`); - emitBare( - `Next: ${identity.binName} deploy text create --model ${exported} --name `, - ); - emitRequestId(response.request_id, settings.quiet); } else { - emitResult(response, format); + emitResult(response, "json"); } }, }); diff --git a/packages/commands/src/commands/finetune/get.ts b/packages/commands/src/commands/finetune/get.ts index 0ba01147..05d96fb7 100644 --- a/packages/commands/src/commands/finetune/get.ts +++ b/packages/commands/src/commands/finetune/get.ts @@ -1,5 +1,5 @@ -import { defineCommand, detectOutputFormat, getFineTune, type FlagsDef } from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { defineCommand, getFineTune, type FlagsDef } from "bailian-cli-core"; +import { emitResult } from "bailian-cli-runtime"; const GET_FLAGS = { jobId: { @@ -17,12 +17,11 @@ export default defineCommand({ flags: GET_FLAGS, exampleArgs: ["--job-id ft-xxx", "--job-id ft-xxx --output json"], async run(ctx) { - const { identity, settings, flags } = ctx; + const { settings, flags } = ctx; const jobId = flags.jobId; - const format = detectOutputFormat(settings.output); if (settings.dryRun) { - emitResult({ action: "finetune.get", job_id: jobId }, format); + emitResult({ action: "finetune.get", job_id: jobId }, "json"); return; } @@ -30,16 +29,20 @@ export default defineCommand({ const job = response.output ?? response.data; if (!job) { - emitBare(`No data returned for ${jobId}`); + emitResult({ job_id: jobId, error: "No data returned" }, "json"); return; } - const hp = job.hyper_parameters; + const hyperParameters = job.hyper_parameters; const hyperParts: string[] = []; - if (hp?.n_epochs !== undefined) hyperParts.push(`n_epochs=${hp.n_epochs}`); - if (hp?.batch_size !== undefined) hyperParts.push(`batch_size=${hp.batch_size}`); - if (hp?.learning_rate !== undefined) hyperParts.push(`learning_rate=${hp.learning_rate}`); - if (hp?.max_length !== undefined) hyperParts.push(`max_length=${hp.max_length}`); + if (hyperParameters?.n_epochs !== undefined) + hyperParts.push(`n_epochs=${hyperParameters.n_epochs}`); + if (hyperParameters?.batch_size !== undefined) + hyperParts.push(`batch_size=${hyperParameters.batch_size}`); + if (hyperParameters?.learning_rate !== undefined) + hyperParts.push(`learning_rate=${hyperParameters.learning_rate}`); + if (hyperParameters?.max_length !== undefined) + hyperParts.push(`max_length=${hyperParameters.max_length}`); const item = { job_id: job.job_id ?? jobId, @@ -53,29 +56,10 @@ export default defineCommand({ model_name: job.model_name ?? "", created_at: job.create_time ?? job.gmt_create ?? "", updated_at: job.end_time ?? job.gmt_modified ?? "", + usage: typeof job.usage === "number" ? String(job.usage) : "", + charge_type: typeof job.charge_type === "string" ? job.charge_type : "", }; - if (format === "json") { - emitResult({ ...item, request_id: response.request_id }, format); - return; - } - - // text / quiet - emitBare(`job_id: ${item.job_id}`); - if (item.base_model) emitBare(`base_model: ${item.base_model}`); - if (item.status) emitBare(`status: ${item.status}`); - if (item.training_type) emitBare(`training_type: ${item.training_type}`); - if (item.training_files.length) emitBare(`training_files: ${item.training_files.join(", ")}`); - if (item.validation_files.length) - emitBare(`validation_files: ${item.validation_files.join(", ")}`); - if (item.hyper_params) emitBare(`hyper_params: ${item.hyper_params}`); - if (item.output_model) - emitBare( - `output_model: ${item.output_model} (→ ${identity.binName} deploy text create --model)`, - ); - if (item.model_name) emitBare(`model_name: ${item.model_name}`); - if (item.created_at) emitBare(`created_at: ${item.created_at}`); - if (item.updated_at) emitBare(`updated_at: ${item.updated_at}`); - emitRequestId(response.request_id, settings.quiet); + emitResult({ ...item, request_id: response.request_id }, "json"); }, }); diff --git a/packages/commands/src/commands/finetune/list.ts b/packages/commands/src/commands/finetune/list.ts index ca4edddf..529a5fd0 100644 --- a/packages/commands/src/commands/finetune/list.ts +++ b/packages/commands/src/commands/finetune/list.ts @@ -1,5 +1,5 @@ -import { defineCommand, detectOutputFormat, listFineTunes, type FlagsDef } from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId, formatTable } from "bailian-cli-runtime"; +import { defineCommand, listFineTunes, type FlagsDef } from "bailian-cli-core"; +import { emitResult } from "bailian-cli-runtime"; const LIST_FLAGS = { page: { type: "number", valueHint: "", description: "Page number (default: 1)" }, @@ -22,14 +22,13 @@ export default defineCommand({ flags: LIST_FLAGS, exampleArgs: ["", "--status RUNNING", "--page-size 20 --output json"], async run(ctx) { - const { identity, settings, flags } = ctx; - const format = detectOutputFormat(settings.output); + const { settings, flags } = ctx; const pageNo = flags.page; const pageSize = flags.pageSize; const status = flags.status || undefined; if (settings.dryRun) { - emitResult({ action: "finetune.list", page: pageNo, page_size: pageSize, status }, format); + emitResult({ action: "finetune.list", page: pageNo, page_size: pageSize, status }, "json"); return; } @@ -38,46 +37,15 @@ export default defineCommand({ const jobs = payload?.jobs ?? []; const total = payload?.total; - const items = jobs.map((item) => ({ - job_id: item.job_id ?? "", - base_model: item.model ?? "", - status: item.status ?? "", - training_type: item.training_type ?? "", - output_model: item.finetuned_output ?? "", - created_at: item.create_time ?? item.gmt_create ?? "", + const items = jobs.map((job) => ({ + job_id: job.job_id ?? "", + base_model: job.model ?? "", + status: job.status ?? "", + training_type: job.training_type ?? "", + output_model: job.finetuned_output ?? "", + created_at: job.create_time ?? job.gmt_create ?? "", })); - if (format === "json") { - emitResult({ items, total, request_id: response.request_id }, format); - return; - } - - // text / quiet - if (items.length === 0) { - emitBare("No fine-tune jobs found."); - return; - } - const headers = [ - "JOB_ID", - "BASE_MODEL", - "STATUS", - "TRAINING_TYPE", - "OUTPUT_MODEL", - "CREATED_AT", - ]; - const rows = items.map((i) => [ - i.job_id, - i.base_model, - i.status, - i.training_type, - i.output_model, - i.created_at, - ]); - for (const line of formatTable(headers, rows)) emitBare(line); - if (total !== undefined) emitBare(`\nTotal: ${total}`); - emitBare( - `Tip: OUTPUT_MODEL is the input for \`${identity.binName} deploy text create --model\``, - ); - emitRequestId(response.request_id, settings.quiet); + emitResult({ items, total, request_id: response.request_id }, "json"); }, }); diff --git a/packages/commands/src/commands/finetune/logs.ts b/packages/commands/src/commands/finetune/logs.ts index bd0802ed..320c07cb 100644 --- a/packages/commands/src/commands/finetune/logs.ts +++ b/packages/commands/src/commands/finetune/logs.ts @@ -1,25 +1,24 @@ import { defineCommand, - detectOutputFormat, getFineTuneLogs, type Client, type FineTuneLogEntry, type FlagsDef, } from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { emitResult } from "bailian-cli-runtime"; /** - * Render a single log entry as a single line (mirrors the flatten logic used - * for non-search text output: prefer common fields, fall back to JSON). + * Render a single log entry as a single line (used for search matching: + * prefer common fields, fall back to JSON). */ function renderEntry(entry: FineTuneLogEntry | string): string { if (typeof entry === "string") return entry; const record = entry as Record; - const ts = (record.timestamp ?? record.time ?? record.create_time ?? "") as string; + const timestamp = (record.timestamp ?? record.time ?? record.create_time ?? "") as string; const level = (record.level ?? "") as string; - const msg = (record.message ?? record.msg ?? record.log ?? "") as string; - if (msg || ts || level) { - return [ts, level, msg].filter(Boolean).join("\t"); + const message = (record.message ?? record.msg ?? record.log ?? "") as string; + if (message || timestamp || level) { + return [timestamp, level, message].filter(Boolean).join("\t"); } return JSON.stringify(entry); } @@ -48,16 +47,16 @@ async function fetchAllLogs( let total = 0; // Hard cap to avoid an unbounded loop if the server misreports `total`. const maxPages = 200; - for (let i = 0; i < maxPages; i++) { + for (let page = 0; page < maxPages; page++) { const response = await getFineTuneLogs(client, jobId, { pageNo, pageSize }); const payload = response.output ?? response.data; - const page = payload?.logs ?? []; + const logs = payload?.logs ?? []; total = payload?.total ?? total; - if (page.length === 0) break; - entries.push(...page); + if (logs.length === 0) break; + entries.push(...logs); // Stop once we've collected everything the server claims exists. if (total && entries.length >= total) break; - if (page.length < pageSize) break; + if (logs.length < pageSize) break; pageNo++; } return { entries, total }; @@ -110,7 +109,6 @@ export default defineCommand({ const pageSize = flags.pageSize; const search = flags.search || undefined; const tail = flags.tail; - const format = detectOutputFormat(settings.output); if (settings.dryRun) { emitResult( @@ -122,7 +120,7 @@ export default defineCommand({ search, tail, }, - format, + "json", ); return; } @@ -147,18 +145,6 @@ export default defineCommand({ const result = tailApplied !== undefined ? scanned.slice(scanned.length - tailApplied) : scanned; - if (settings.quiet || format === "text") { - if (result.length === 0) { - emitBare(search ? `No logs matched "${search}".` : "No logs returned."); - return; - } - for (const entry of result) emitBare(renderEntry(entry)); - const parts: string[] = [`${result.length} shown`]; - if (matched !== undefined) parts.push(`matched ${matched}`); - parts.push(`of ${entries.length}` + (total ? ` (total ${total})` : "")); - emitBare(`\n${parts.join(", ")}`); - return; - } emitResult( { ...(matched !== undefined ? { matched } : {}), @@ -168,28 +154,13 @@ export default defineCommand({ ...(tailApplied !== undefined ? { tail: tailApplied } : {}), logs: result, }, - format, + "json", ); return; } // Default: single page, verbatim response. const response = await getFineTuneLogs(ctx.client, jobId, { pageNo, pageSize }); - const payload = response.output ?? response.data; - const logs = payload?.logs ?? []; - - if (settings.quiet || format === "text") { - if (logs.length === 0) { - emitBare("No logs returned."); - return; - } - for (const entry of logs) { - emitBare(renderEntry(entry)); - } - if (payload?.total !== undefined) emitBare(`\nTotal: ${payload.total}`); - emitRequestId(response.request_id, settings.quiet); - } else { - emitResult(response, format); - } + emitResult(response, "json"); }, }); diff --git a/packages/commands/src/commands/finetune/watch.ts b/packages/commands/src/commands/finetune/watch.ts index c02b2079..8baf7380 100644 --- a/packages/commands/src/commands/finetune/watch.ts +++ b/packages/commands/src/commands/finetune/watch.ts @@ -1,12 +1,11 @@ import { defineCommand, - detectOutputFormat, getFineTune, BailianError, ExitCode, type FlagsDef, } from "bailian-cli-core"; -import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime"; +import { emitResult, emitBare } from "bailian-cli-runtime"; const DEFAULT_INTERVAL_SEC = 10; const MIN_INTERVAL_SEC = 1; @@ -103,7 +102,6 @@ export default defineCommand({ const follow = flags.follow; const intervalSec = Math.max(MIN_INTERVAL_SEC, flags.interval ?? DEFAULT_INTERVAL_SEC); const pollTimeoutSec = flags.pollTimeout; - const format = detectOutputFormat(settings.output); if (settings.dryRun) { emitResult( @@ -114,7 +112,7 @@ export default defineCommand({ interval: intervalSec, timeout: pollTimeoutSec, }, - format, + "json", ); return; } @@ -132,15 +130,10 @@ export default defineCommand({ if (settings.quiet) { // Just the status word — ideal for `status=$(... finetune watch ... --quiet)`. emitBare(status || "UNKNOWN"); - } else if (format === "text") { - emitBare(`${nowStamp()} ${jobId} ${status || "UNKNOWN"}`); - if (status === "SUCCEEDED") emitBare(`✓ ${jobId} ${status}`); - emitRequestId(response.request_id, settings.quiet); } else { - // json: a compact, purpose-built status probe. emitResult( { job_id: jobId, status: status || "UNKNOWN", terminal, request_id: response.request_id }, - format, + "json", ); } @@ -168,18 +161,17 @@ export default defineCommand({ const job = response.output ?? response.data; const status = String(job?.status ?? "").toUpperCase(); - if (format === "text" && !settings.quiet && status !== lastStatus) { - emitBare(`${nowStamp()} ${jobId} ${status || "UNKNOWN"}`); + if (!settings.quiet && status !== lastStatus) { + process.stderr.write(`${nowStamp()} ${jobId} ${status || "UNKNOWN"}\n`); lastStatus = status; } if (TERMINAL_STATUSES.has(status)) { const elapsed = Date.now() - startedAt; - if (format !== "text" || settings.quiet) { - emitResult(response, format); - } else if (status === "SUCCEEDED") { - emitBare(`\n✓ ${jobId} ${status} (elapsed ${formatElapsed(elapsed)})`); - emitRequestId(response.request_id, settings.quiet); + if (settings.quiet) { + emitBare(status || "UNKNOWN"); + } else { + emitResult(response, "json"); } if (status !== "SUCCEEDED") { throw new BailianError( @@ -205,7 +197,7 @@ export default defineCommand({ // Any other error (including the BailianError thrown above) propagates to // the central handler. if (controller.signal.aborted) { - emitBare("\nInterrupted."); + process.stderr.write("\nInterrupted.\n"); return; } throw error; diff --git a/packages/commands/src/index.ts b/packages/commands/src/index.ts index cba8b321..63a13163 100644 --- a/packages/commands/src/index.ts +++ b/packages/commands/src/index.ts @@ -87,6 +87,8 @@ export { default as deployModels } from "./commands/deploy/models.ts"; export { default as deployScale } from "./commands/deploy/scale.ts"; export { default as deployUpdate } from "./commands/deploy/update.ts"; export { default as deployDelete } from "./commands/deploy/delete.ts"; +export { default as deployPause } from "./commands/deploy/pause.ts"; +export { default as deployResume } from "./commands/deploy/resume.ts"; export { default as tokenPlanListSeats } from "./commands/token-plan/list-seats.ts"; export { default as tokenPlanCreateKey } from "./commands/token-plan/create-key.ts"; export { default as tokenPlanAssignSeats } from "./commands/token-plan/assign-seats.ts"; diff --git a/packages/core/src/deploy/index.ts b/packages/core/src/deploy/index.ts index e0515c13..ce3909ac 100644 --- a/packages/core/src/deploy/index.ts +++ b/packages/core/src/deploy/index.ts @@ -2,3 +2,4 @@ export * from "./api.ts"; export * from "./types.ts"; export * from "./constants.ts"; export * from "./plans.ts"; +export * from "./lifecycle.ts"; diff --git a/packages/core/src/deploy/lifecycle.ts b/packages/core/src/deploy/lifecycle.ts new file mode 100644 index 00000000..9f647d53 --- /dev/null +++ b/packages/core/src/deploy/lifecycle.ts @@ -0,0 +1,99 @@ +/** + * Deployment lifecycle operations via the **console gateway**. + * + * Unlike the DashScope REST endpoints in `api.ts`, start/stop/list-independent + * are console-domain APIs (`zeldaEasy.broadscope-platform.modelInstance.*`). + * Commands using these must declare `auth: "console"`. + */ +import type { Client } from "../client/client.ts"; +import { unwrapResponse } from "../console/models.ts"; + +// --------------------------------------------------------------------------- +// API names +// --------------------------------------------------------------------------- + +export const DEPLOY_START_API = "zeldaEasy.broadscope-platform.modelInstance.startModelService"; +export const DEPLOY_STOP_API = "zeldaEasy.broadscope-platform.modelInstance.stopModelService"; +export const DEPLOY_LIST_INDEPENDENT_API = + "zeldaEasy.broadscope-platform.modelInstance.listIndependentDeployedModel"; + +// --------------------------------------------------------------------------- +// Types +// --------------------------------------------------------------------------- + +export interface ModelServiceEntry { + modelServiceId?: string; + deployedModel?: string; + deployed_model?: string; + status?: string; + modelName?: string; + model_name?: string; + plan?: string; + [key: string]: unknown; +} + +// --------------------------------------------------------------------------- +// API wrappers +// --------------------------------------------------------------------------- + +/** Start (bring online) a stopped deployment. */ +export async function startModelService( + client: Client, + modelServiceId: string, +): Promise> { + const raw = await client.console>(DEPLOY_START_API, { + input: { modelServiceId }, + }); + return unwrapResponse(raw); +} + +/** Stop (take offline) a running deployment. Stops billing for mu/ptu plans. */ +export async function stopModelService( + client: Client, + modelServiceId: string, +): Promise> { + const raw = await client.console>(DEPLOY_STOP_API, { + input: { modelServiceId }, + }); + return unwrapResponse(raw); +} + +/** + * List independently deployed models (console domain). + * Used for precheck status verification and ID mapping. + * Paginates internally to return all entries. + */ +export async function listIndependentDeployedModels(client: Client): Promise { + const allEntries: ModelServiceEntry[] = []; + let page = 1; + + while (true) { + const raw = await client.console>(DEPLOY_LIST_INDEPENDENT_API, { + input: { pageNo: page, pageSize: 50 }, + }); + const resp = unwrapResponse(raw); + const records = (resp.records ?? []) as ModelServiceEntry[]; + allEntries.push(...records); + const pageCount = (resp.pageCount as number) ?? 1; + if (page >= pageCount || records.length === 0) break; + page++; + } + + return allEntries; +} + +/** + * Find a deployment entry by its identifier in the console-domain list. + * Matches against `modelServiceId`, `deployedModel`, or `deployed_model`. + */ +export function findDeploymentEntry( + entries: ModelServiceEntry[], + deployedModel: string, +): ModelServiceEntry | undefined { + return entries.find( + (entry) => + entry.modelServiceId === deployedModel || + entry.deployedModel === deployedModel || + entry.deployed_model === deployedModel, + ); +} diff --git a/skills/bailian-finetune/reference/deploy.md b/skills/bailian-finetune/reference/deploy.md index a33d2c5e..8d6c7770 100644 --- a/skills/bailian-finetune/reference/deploy.md +++ b/skills/bailian-finetune/reference/deploy.md @@ -7,17 +7,19 @@ Index: [index.md](index.md) ## Commands in this group -| Command | Description | -| ------------------------ | --------------------------------------------------------- | -| `bl deploy audio create` | Create an audio (TTS) model deployment | -| `bl deploy delete` | Delete a model deployment (must be STOPPED or FAILED) | -| `bl deploy get` | Get details of a single model deployment | -| `bl deploy image create` | Create an image generation model deployment | -| `bl deploy list` | List model deployments | -| `bl deploy models` | List models available for deployment | -| `bl deploy scale` | Scale a deployment's capacity | -| `bl deploy text create` | Create a text model deployment | -| `bl deploy update` | Update a deployment's rate limits (rpm_limit / tpm_limit) | +| Command | Description | +| ------------------------ | ------------------------------------------------------------- | +| `bl deploy audio create` | Create an audio (TTS) model deployment | +| `bl deploy delete` | Delete a model deployment (must be STOPPED or FAILED) | +| `bl deploy get` | Get details of a single model deployment | +| `bl deploy image create` | Create an image generation model deployment | +| `bl deploy list` | List model deployments | +| `bl deploy models` | List models available for deployment | +| `bl deploy pause` | Pause a running model deployment (stops billing for mu/ptu) | +| `bl deploy resume` | Resume a paused model deployment (brings service back online) | +| `bl deploy scale` | Scale a deployment's capacity | +| `bl deploy text create` | Create a text model deployment | +| `bl deploy update` | Update a deployment's rate limits (rpm_limit / tpm_limit) | ## Command details @@ -263,6 +265,82 @@ bl deploy models --source custom --page-size 50 bl deploy models --catalog-version v1.0 --output json ``` +### `bl deploy pause` + +| Field | Value | +| --------------- | ----------------------------------------------------------- | +| **Name** | `deploy pause` | +| **Description** | Pause a running model deployment (stops billing for mu/ptu) | +| **Usage** | `bl deploy pause --deployed-model [--skip-precheck]` | + +#### Flags + +| Flag | Type | Required | Description | +| ------------------------------ | ------ | -------- | -------------------------------------------------------- | +| `--deployed-model ` | string | yes | Deployed model identifier (required) | +| `--skip-precheck` | switch | no | Skip the local RUNNING/PENDING status precheck | +| `--console-region ` | string | no | Console gateway region (e.g. cn-beijing, ap-southeast-1) | +| `--console-site ` | string | no | Console site: domestic, international | +| `--console-switch-agent ` | number | no | Switch agent UID for delegated access | +| `--workspace-id ` | string | no | Workspace ID (env: BAILIAN_WORKSPACE_ID) | + +#### Notes + +- While paused, billing ceases for mu/ptu plans. Use `deploy resume` to bring it back online or `deploy delete` to remove. +- Precheck verifies status is RUNNING/PENDING before issuing the pause; pass --skip-precheck to bypass. + +#### Examples + +```bash +bl deploy pause --deployed-model dep-... +``` + +```bash +bl deploy pause --deployed-model dep-... --skip-precheck +``` + +```bash +bl deploy pause --deployed-model dep-... --dry-run +``` + +### `bl deploy resume` + +| Field | Value | +| --------------- | ------------------------------------------------------------- | +| **Name** | `deploy resume` | +| **Description** | Resume a paused model deployment (brings service back online) | +| **Usage** | `bl deploy resume --deployed-model [--skip-precheck]` | + +#### Flags + +| Flag | Type | Required | Description | +| ------------------------------ | ------ | -------- | -------------------------------------------------------- | +| `--deployed-model ` | string | yes | Deployed model identifier (required) | +| `--skip-precheck` | switch | no | Skip the local STOPPED status precheck | +| `--console-region ` | string | no | Console gateway region (e.g. cn-beijing, ap-southeast-1) | +| `--console-site ` | string | no | Console site: domestic, international | +| `--console-switch-agent ` | number | no | Switch agent UID for delegated access | +| `--workspace-id ` | string | no | Workspace ID (env: BAILIAN_WORKSPACE_ID) | + +#### Notes + +- Precheck verifies status is STOPPED before issuing the resume; pass --skip-precheck to bypass. +- For mu/ptu plans, billing resumes once the service is back online. + +#### Examples + +```bash +bl deploy resume --deployed-model dep-... +``` + +```bash +bl deploy resume --deployed-model dep-... --skip-precheck +``` + +```bash +bl deploy resume --deployed-model dep-... --dry-run +``` + ### `bl deploy scale` | Field | Value | diff --git a/skills/bailian-finetune/reference/finetune.md b/skills/bailian-finetune/reference/finetune.md index d269dfc0..0b05ca09 100644 --- a/skills/bailian-finetune/reference/finetune.md +++ b/skills/bailian-finetune/reference/finetune.md @@ -166,8 +166,8 @@ bl finetune capability --training-type sft --quiet #### Notes -- Use the returned `checkpoint` value with `finetune export` to publish -- a deployable model. +- `model_name` (shown for SUCCEEDED checkpoints) is the direct input for `deploy create --model`. +- Checkpoints expire ~15 days after creation; `expire_time` shows the deadline. Export or deploy before expiry. #### Examples diff --git a/skills/bailian-finetune/reference/index.md b/skills/bailian-finetune/reference/index.md index 4877e27a..07cf63f7 100644 --- a/skills/bailian-finetune/reference/index.md +++ b/skills/bailian-finetune/reference/index.md @@ -22,6 +22,8 @@ Use this index for the skill-scoped quick index and global flags. | `bl deploy image create` | Create an image generation model deployment | [deploy.md](deploy.md) | | `bl deploy list` | List model deployments | [deploy.md](deploy.md) | | `bl deploy models` | List models available for deployment | [deploy.md](deploy.md) | +| `bl deploy pause` | Pause a running model deployment (stops billing for mu/ptu) | [deploy.md](deploy.md) | +| `bl deploy resume` | Resume a paused model deployment (brings service back online) | [deploy.md](deploy.md) | | `bl deploy scale` | Scale a deployment's capacity | [deploy.md](deploy.md) | | `bl deploy text create` | Create a text model deployment | [deploy.md](deploy.md) | | `bl deploy update` | Update a deployment's rate limits (rpm_limit / tpm_limit) | [deploy.md](deploy.md) | @@ -43,7 +45,7 @@ Use this index for the skill-scoped quick index and global flags. | Group | Commands | Reference | | ---------- | ---------------------------------------------------------------------------------------------------------------------------------------- | -------------------------- | | `dataset` | `delete`, `get`, `list`, `upload`, `validate` | [dataset.md](dataset.md) | -| `deploy` | `audio create`, `delete`, `get`, `image create`, `list`, `models`, `scale`, `text create`, `update` | [deploy.md](deploy.md) | +| `deploy` | `audio create`, `delete`, `get`, `image create`, `list`, `models`, `pause`, `resume`, `scale`, `text create`, `update` | [deploy.md](deploy.md) | | `finetune` | `audio create`, `cancel`, `capability`, `checkpoints`, `delete`, `export`, `get`, `image create`, `list`, `logs`, `text create`, `watch` | [finetune.md](finetune.md) | ## Global flags From f30fff90654de3be3b36260cd29fdb30cf5556e1 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Thu, 6 Aug 2026 15:08:02 +0800 Subject: [PATCH 02/14] feat(finetune): clarify model flags; add price estimate & actual cost MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - Rename for clarity: finetune --model → --base-model (create/price/ capability/list); deploy create --model → --model-name, --name → --display-name - Add `finetune price` (console domain) for pre-training cost estimate (sft/dpo/cpt) - Add actual training cost (fee.ts) enriched into finetune get/watch from catalog price × reported usage --- README.md | 6 +- README.zh.md | 6 +- packages/cli/README.md | 6 +- packages/cli/README.zh.md | 6 +- packages/cli/src/commands.ts | 2 + .../commands/src/commands/deploy/create.ts | 47 +++-- .../src/commands/finetune/capability.ts | 17 +- .../src/commands/finetune/checkpoints.ts | 2 +- .../commands/src/commands/finetune/create.ts | 50 +++--- .../commands/src/commands/finetune/fee.ts | 100 +++++++++++ .../commands/src/commands/finetune/get.ts | 17 +- .../commands/src/commands/finetune/list.ts | 17 +- .../commands/src/commands/finetune/price.ts | 139 +++++++++++++++ .../commands/src/commands/finetune/watch.ts | 35 +++- packages/commands/src/index.ts | 1 + .../commands/tests/e2e/deploy.e2e.test.ts | 14 +- .../commands/tests/e2e/finetune.e2e.test.ts | 28 +-- packages/core/src/finetune/api.ts | 3 + packages/core/src/finetune/index.ts | 1 + packages/core/src/finetune/price.ts | 121 +++++++++++++ skills/bailian-finetune/SKILL.md | 10 +- skills/bailian-finetune/reference/deploy.md | 167 +++++++++--------- skills/bailian-finetune/reference/finetune.md | 157 ++++++++++------ skills/bailian-finetune/reference/index.md | 11 +- 24 files changed, 707 insertions(+), 256 deletions(-) create mode 100644 packages/commands/src/commands/finetune/fee.ts create mode 100644 packages/commands/src/commands/finetune/price.ts create mode 100644 packages/core/src/finetune/price.ts diff --git a/README.md b/README.md index 1e869d13..719e94fe 100644 --- a/README.md +++ b/README.md @@ -130,10 +130,10 @@ bl auth login --console # Fine-tune & deploy — a one-shot train-to-serve workflow bl dataset upload --file ./train.jsonl # Upload a .jsonl dataset (validated first) -bl finetune text create --model qwen3-8b --datasets ./train.jsonl --training-type sft-lora # Local paths auto-upload +bl finetune text create --base-model qwen3-8b --datasets ./train.jsonl --training-type sft-lora # Local paths auto-upload bl finetune watch --job-id ft-xxx --output json # Non-blocking probe (running/succeeded return 0; failed/canceled report an error) -bl finetune capability --model qwen3-8b # Which training types a model supports -bl deploy text create --model qwen3-8b --name my-svc --plan mu # Deploy the trained model as an endpoint +bl finetune capability --base-model qwen3-8b # Which training types a model supports +bl deploy text create --model-name qwen3-8b --display-name my-svc --plan mu # Deploy the trained model as an endpoint # Browse models / apps / free-tier quota / usage statistics / workspaces bl model list # Browse model families and pricing diff --git a/README.zh.md b/README.zh.md index b5f2b41d..df8a1de3 100644 --- a/README.zh.md +++ b/README.zh.md @@ -128,10 +128,10 @@ bl auth login --console # 微调与部署 — 从训练到服务的一站式流程 bl dataset upload --file ./train.jsonl # 上传 .jsonl 数据集（先校验） -bl finetune text create --model qwen3-8b --datasets ./train.jsonl --training-type sft-lora # 本地路径自动上传 +bl finetune text create --base-model qwen3-8b --datasets ./train.jsonl --training-type sft-lora # 本地路径自动上传 bl finetune watch --job-id ft-xxx --output json # 非阻塞探测（运行中/成功返回 0；失败/取消报错） -bl finetune capability --model qwen3-8b # 查询模型支持哪些训练方式 -bl deploy text create --model qwen3-8b --name my-svc --plan mu # 把训练好的模型部署为推理服务 +bl finetune capability --base-model qwen3-8b # 查询模型支持哪些训练方式 +bl deploy text create --model-name qwen3-8b --display-name my-svc --plan mu # 把训练好的模型部署为推理服务 # 浏览模型 / 应用 / 免费额度 / 用量统计 / 业务空间 bl model list # 浏览模型系列与价格信息 diff --git a/packages/cli/README.md b/packages/cli/README.md index 1e869d13..719e94fe 100644 --- a/packages/cli/README.md +++ b/packages/cli/README.md @@ -130,10 +130,10 @@ bl auth login --console # Fine-tune & deploy — a one-shot train-to-serve workflow bl dataset upload --file ./train.jsonl # Upload a .jsonl dataset (validated first) -bl finetune text create --model qwen3-8b --datasets ./train.jsonl --training-type sft-lora # Local paths auto-upload +bl finetune text create --base-model qwen3-8b --datasets ./train.jsonl --training-type sft-lora # Local paths auto-upload bl finetune watch --job-id ft-xxx --output json # Non-blocking probe (running/succeeded return 0; failed/canceled report an error) -bl finetune capability --model qwen3-8b # Which training types a model supports -bl deploy text create --model qwen3-8b --name my-svc --plan mu # Deploy the trained model as an endpoint +bl finetune capability --base-model qwen3-8b # Which training types a model supports +bl deploy text create --model-name qwen3-8b --display-name my-svc --plan mu # Deploy the trained model as an endpoint # Browse models / apps / free-tier quota / usage statistics / workspaces bl model list # Browse model families and pricing diff --git a/packages/cli/README.zh.md b/packages/cli/README.zh.md index b5f2b41d..df8a1de3 100644 --- a/packages/cli/README.zh.md +++ b/packages/cli/README.zh.md @@ -128,10 +128,10 @@ bl auth login --console # 微调与部署 — 从训练到服务的一站式流程 bl dataset upload --file ./train.jsonl # 上传 .jsonl 数据集（先校验） -bl finetune text create --model qwen3-8b --datasets ./train.jsonl --training-type sft-lora # 本地路径自动上传 +bl finetune text create --base-model qwen3-8b --datasets ./train.jsonl --training-type sft-lora # 本地路径自动上传 bl finetune watch --job-id ft-xxx --output json # 非阻塞探测（运行中/成功返回 0；失败/取消报错） -bl finetune capability --model qwen3-8b # 查询模型支持哪些训练方式 -bl deploy text create --model qwen3-8b --name my-svc --plan mu # 把训练好的模型部署为推理服务 +bl finetune capability --base-model qwen3-8b # 查询模型支持哪些训练方式 +bl deploy text create --model-name qwen3-8b --display-name my-svc --plan mu # 把训练好的模型部署为推理服务 # 浏览模型 / 应用 / 免费额度 / 用量统计 / 业务空间 bl model list # 浏览模型系列与价格信息 diff --git a/packages/cli/src/commands.ts b/packages/cli/src/commands.ts index bd18f4d5..ebd8a629 100644 --- a/packages/cli/src/commands.ts +++ b/packages/cli/src/commands.ts @@ -71,6 +71,7 @@ import { finetuneExport, finetuneWatch, finetuneCapability, + finetunePrice, deployTextCreate, deployAudioCreate, deployImageCreate, @@ -191,6 +192,7 @@ export const commands: Record = { "finetune export": finetuneExport, "finetune watch": finetuneWatch, "finetune capability": finetuneCapability, + "finetune price": finetunePrice, "deploy text create": deployTextCreate, "deploy audio create": deployAudioCreate, "deploy image create": deployImageCreate, diff --git a/packages/commands/src/commands/deploy/create.ts b/packages/commands/src/commands/deploy/create.ts index 15f6b2a0..de468fcb 100644 --- a/packages/commands/src/commands/deploy/create.ts +++ b/packages/commands/src/commands/deploy/create.ts @@ -13,13 +13,13 @@ import { import { emitResult, emitBare } from "bailian-cli-runtime"; const CREATE_FLAGS = { - model: { + modelName: { type: "string", - valueHint: "", - description: "Model name (catalog model or fine-tuned output) (required)", + valueHint: "", + description: "Model to deploy — fine-tuned output name or catalog model (required)", required: true, }, - name: { + displayName: { type: "string", valueHint: "", description: "Console display name for the deployment (required)", @@ -63,7 +63,7 @@ const CREATE_FLAGS = { } satisfies FlagsDef; const CREATE_USAGE = - "--model --name [--plan ] [--deploy-spec ] [--capacity ] [--billing-method ] [--input-tpm ] [--output-tpm ] [--thinking-output-tpm ]"; + "--model-name --display-name [--plan ] [--deploy-spec ] [--capacity ] [--billing-method ] [--input-tpm ] [--output-tpm ] [--thinking-output-tpm ]"; const CREATE_NOTES = [ "Plan defaults to `lora` (Token-billed) for text/image and `mu` (model-unit-", @@ -77,14 +77,11 @@ const CREATE_NOTES = [ "Use `bl deploy models --source base` to inspect available templates.", "After creation, status starts at PENDING and transitions to RUNNING.", "Invoke the deployed model with: bl text chat --model ", - "WARNING: --model is overloaded across commands and refers to DIFFERENT", - "values. `bl deploy create --model` takes the exported model_name", - "(e.g. `qwen3-8b-ft-...`), but the create response also returns a", - "`deployed_model` field (the deployment instance id, e.g.", - "`qwen3-8b-5ecb5f068d79`). The inference call `bl text chat --model` must use", - "the `deployed_model` from the create response — NOT the `model_name` you", - "passed to `deploy create`. Do not reuse the value across the two", - "commands.", + "NOTE: --model-name is the model being deployed (e.g. `qwen3-8b-ft-...`).", + "The create response also returns a `deployed_model` field — the deployment", + "instance id (e.g. `qwen3-8b-5ecb5f068d79`). Use that id for inference", + "(`bl text chat --model `) and lifecycle commands", + "(`deploy get/scale/pause/resume/delete --deployed-model `).", ]; /** @@ -118,8 +115,8 @@ async function runCreate( ctx: CommandContext, ): Promise { const { identity, settings, flags } = ctx; - const model = flags.model as string; - const name = flags.name as string; + const model = flags.modelName as string; + const name = flags.displayName as string; const plan = (flags.plan as string | undefined) || defaultDeployPlan(modality); // Plan-specific behaviour is owned by core `plans.ts`. The strategy resolves @@ -165,10 +162,10 @@ export const deployTextCreate = defineCommand({ usageArgs: CREATE_USAGE, flags: CREATE_FLAGS, exampleArgs: [ - "--model my-qwen-sft --name my-sft-test", - "--model qwen3.6-flash-2026-04-16 --name my-flash --plan ptu --input-tpm 10000 --output-tpm 1000", - "--model qwen3-8b --name my-qwen3-mu --plan mu", - "--model qwen3-8b --name my-qwen3 --plan mu --deploy-spec MU1 --capacity 2", + "--model-name my-qwen-sft --display-name my-sft-test", + "--model-name qwen3.6-flash-2026-04-16 --display-name my-flash --plan ptu --input-tpm 10000 --output-tpm 1000", + "--model-name qwen3-8b --display-name my-qwen3-mu --plan mu", + "--model-name qwen3-8b --display-name my-qwen3 --plan mu --deploy-spec MU1 --capacity 2", ], notes: CREATE_NOTES, validate: (flags) => validateCreate("text", flags), @@ -182,9 +179,9 @@ export const deployAudioCreate = defineCommand({ usageArgs: CREATE_USAGE, flags: CREATE_FLAGS, exampleArgs: [ - "--model my-cosyvoice-ft --name my-tts", - "--model my-cosyvoice-ft --name my-tts --deploy-spec dps-xxxx --capacity 1", - "--model my-cosyvoice-ft --name my-tts --dry-run", + "--model-name my-cosyvoice-ft --display-name my-tts", + "--model-name my-cosyvoice-ft --display-name my-tts --deploy-spec dps-xxxx --capacity 1", + "--model-name my-cosyvoice-ft --display-name my-tts --dry-run", ], notes: CREATE_NOTES, validate: (flags) => validateCreate("audio", flags), @@ -198,9 +195,9 @@ export const deployImageCreate = defineCommand({ usageArgs: CREATE_USAGE, flags: CREATE_FLAGS, exampleArgs: [ - "--model my-wan-ft --name my-wan", - "--model my-wan-ft --name my-wan-mu --plan mu", - "--model my-wan-ft --name my-wan --dry-run", + "--model-name my-wan-ft --display-name my-wan", + "--model-name my-wan-ft --display-name my-wan-mu --plan mu", + "--model-name my-wan-ft --display-name my-wan --dry-run", ], notes: CREATE_NOTES, validate: (flags) => validateCreate("image", flags), diff --git a/packages/commands/src/commands/finetune/capability.ts b/packages/commands/src/commands/finetune/capability.ts index e9091a93..7ab81c75 100644 --- a/packages/commands/src/commands/finetune/capability.ts +++ b/packages/commands/src/commands/finetune/capability.ts @@ -43,7 +43,7 @@ async function fetchAllFoundationModels(settings: Settings): Promise", description: "List training types supported by this base model.", @@ -59,29 +59,30 @@ export default defineCommand({ description: "Query fine-tune training capability — by model (which training types it supports) or by training type (which models support it)", auth: "none", - usageArgs: "--model | --training-type ", + usageArgs: "--base-model | --training-type ", flags: CAPABILITY_FLAGS, exampleArgs: [ - "--model qwen3-8b", + "--base-model qwen3-8b", "--training-type sft-lora", "--training-type cpt --output json", "--training-type sft --quiet", ], notes: [ - "Exactly one of --model / --training-type is required.", + "Exactly one of --base-model / --training-type is required.", "Training-type values use the `` / `-lora` convention:", "sft | sft-lora | dpo | dpo-lora | cpt. (cpt has no -lora variant server-side.)", "Queries listFoundationModels, a public API — no console login needed.", ], validate: (f) => { - if (f.model && f.trainingType) - return "--model and --training-type are mutually exclusive; pass one."; - if (!f.model && !f.trainingType) return "one of --model / --training-type is required."; + if (f.baseModel && f.trainingType) + return "--base-model and --training-type are mutually exclusive; pass one."; + if (!f.baseModel && !f.trainingType) + return "one of --base-model / --training-type is required."; return undefined; }, async run(ctx) { const { settings, flags } = ctx; - const model = flags.model || undefined; + const model = flags.baseModel || undefined; const trainingType = flags.trainingType || undefined; if (settings.dryRun) { diff --git a/packages/commands/src/commands/finetune/checkpoints.ts b/packages/commands/src/commands/finetune/checkpoints.ts index 769cf72d..190110d6 100644 --- a/packages/commands/src/commands/finetune/checkpoints.ts +++ b/packages/commands/src/commands/finetune/checkpoints.ts @@ -19,7 +19,7 @@ export default defineCommand({ flags: CHECKPOINTS_FLAGS, exampleArgs: ["--job-id ft-xxx", "--job-id ft-xxx --output json"], notes: [ - "`model_name` (shown for SUCCEEDED checkpoints) is the direct input for `deploy create --model`.", + "`model_name` (shown for SUCCEEDED checkpoints) is the direct input for `deploy create --model-name`.", "Checkpoints expire ~15 days after creation; `expire_time` shows the deadline. Export or deploy before expiry.", ], async run(ctx) { diff --git a/packages/commands/src/commands/finetune/create.ts b/packages/commands/src/commands/finetune/create.ts index f9bbaf0e..d6496f6b 100644 --- a/packages/commands/src/commands/finetune/create.ts +++ b/packages/commands/src/commands/finetune/create.ts @@ -215,10 +215,10 @@ type CommandModality = "text" | "audio" | "image"; * output. Every modality's model consumes these. */ const COMMON_FLAGS = { - model: { + baseModel: { type: "string", valueHint: "", - description: "Base model to fine-tune", + description: "Base model to fine-tune (e.g. qwen3-8b; not the output model name)", required: true, }, datasets: { @@ -316,13 +316,13 @@ const IMAGE_FLAGS = { } satisfies FlagsDef; const TEXT_USAGE = - "--model --datasets [--validations ] [--model-name ] [--suffix ] [--n-epochs ] [--batch-size ] [--learning-rate ] [--max-length ] [--training-type ]"; + "--base-model --datasets [--validations ] [--model-name ] [--suffix ] [--n-epochs ] [--batch-size ] [--learning-rate ] [--max-length ] [--training-type ]"; const AUDIO_USAGE = - "--model --datasets [--validations ] [--model-name ] [--suffix ]"; + "--base-model --datasets [--validations ] [--model-name ] [--suffix ]"; const IMAGE_USAGE = - "--model --datasets [--validations ] [--model-name ] [--suffix ] [--generation-type ] [--learning-rate ]"; + "--base-model --datasets [--validations ] [--model-name ] [--suffix ] [--generation-type ] [--learning-rate ]"; const COMMON_NOTES = [ "Creating a job uploads any local datasets and consumes training quota.", @@ -382,7 +382,7 @@ async function runCreate( ): Promise { const { identity, settings } = ctx; const flags = ctx.flags as Record; - const model = flags.model as string; + const model = flags.baseModel as string; const datasetsRaw = flags.datasets as string; // CosyVoice audio fine-tuning accepts exactly one training file @@ -636,14 +636,14 @@ export const finetuneTextCreate = defineCommand({ usageArgs: TEXT_USAGE, flags: TEXT_FLAGS, exampleArgs: [ - "--model qwen3-8b --datasets file-xxx", - "--model qwen3-8b --datasets ./train.jsonl", - "--model qwen3-8b --datasets ./train.jsonl --validations ./eval.jsonl", - "--model qwen3-8b --datasets file-aaa,./extra.jsonl", - "--model qwen3-8b --datasets ./train.jsonl --training-type sft", - '--model qwen3-8b --datasets file-xxx --learning-rate "1.6e-5" --n-epochs 4', - "--model qwen3-8b --datasets file-xxx --output json", - "--model qwen3-8b --datasets file-xxx --dry-run", + "--base-model qwen3-8b --datasets file-xxx", + "--base-model qwen3-8b --datasets ./train.jsonl", + "--base-model qwen3-8b --datasets ./train.jsonl --validations ./eval.jsonl", + "--base-model qwen3-8b --datasets file-aaa,./extra.jsonl", + "--base-model qwen3-8b --datasets ./train.jsonl --training-type sft", + '--base-model qwen3-8b --datasets file-xxx --learning-rate "1.6e-5" --n-epochs 4', + "--base-model qwen3-8b --datasets file-xxx --output json", + "--base-model qwen3-8b --datasets file-xxx --dry-run", ], notes: TEXT_NOTES, run: (ctx) => runCreate("text", ctx), @@ -656,11 +656,11 @@ export const finetuneAudioCreate = defineCommand({ usageArgs: AUDIO_USAGE, flags: AUDIO_FLAGS, exampleArgs: [ - "--model cosyvoice-v3-flash --datasets ./audio.zip", - "--model cosyvoice-v3-flash --datasets file-xxx", - "--model cosyvoice-v3-flash --datasets ./audio.zip --model-name my-tts", - "--model cosyvoice-v3-flash --datasets file-xxx --output json", - "--model cosyvoice-v3-flash --datasets ./audio.zip --dry-run", + "--base-model cosyvoice-v3-flash --datasets ./audio.zip", + "--base-model cosyvoice-v3-flash --datasets file-xxx", + "--base-model cosyvoice-v3-flash --datasets ./audio.zip --model-name my-tts", + "--base-model cosyvoice-v3-flash --datasets file-xxx --output json", + "--base-model cosyvoice-v3-flash --datasets ./audio.zip --dry-run", ], notes: AUDIO_NOTES, run: (ctx) => runCreate("audio", ctx), @@ -673,12 +673,12 @@ export const finetuneImageCreate = defineCommand({ usageArgs: IMAGE_USAGE, flags: IMAGE_FLAGS, exampleArgs: [ - "--model wan2.7-image-pro --datasets ./images.zip", - "--model wan2.7-image-pro --datasets file-xxx", - "--model wan2.7-image-pro --datasets file-xxx --generation-type i2i", - "--model wan2.7-image-pro --datasets ./images.zip --model-name my-wan", - "--model wan2.7-image-pro --datasets file-xxx --output json", - "--model wan2.7-image-pro --datasets ./images.zip --dry-run", + "--base-model wan2.7-image-pro --datasets ./images.zip", + "--base-model wan2.7-image-pro --datasets file-xxx", + "--base-model wan2.7-image-pro --datasets file-xxx --generation-type i2i", + "--base-model wan2.7-image-pro --datasets ./images.zip --model-name my-wan", + "--base-model wan2.7-image-pro --datasets file-xxx --output json", + "--base-model wan2.7-image-pro --datasets ./images.zip --dry-run", ], notes: IMAGE_NOTES, run: (ctx) => runCreate("image", ctx), diff --git a/packages/commands/src/commands/finetune/fee.ts b/packages/commands/src/commands/finetune/fee.ts new file mode 100644 index 00000000..7559524a --- /dev/null +++ b/packages/commands/src/commands/finetune/fee.ts @@ -0,0 +1,100 @@ +/** + * Best-effort actual training fee calculation using the model catalog's + * "ft" (fine-tune) price entry. Pure API-key domain — no console auth needed. + * + * The model catalog (`listFoundationModels` via public gateway) returns a + * `prices[]` array **only when `queryPrice: true` is passed** (the same flag + * `fetchModelDetail` uses). Combined with the job's `output.usage` (actual + * consumed tokens, present on SUCCEEDED / CANCELED), this gives the exact + * training cost without any console-domain login. + */ +import { + callConsoleGateway, + effectiveConsoleGatewayConfig, + unwrapResponse, + MODEL_LIST_API, + type Settings, + type ModelPriceInfo, +} from "bailian-cli-core"; + +export interface ActualFee { + cost: number; + unitPrice: number; + priceUnit: string; +} + +/** + * Fetch the model's training price from the public catalog gateway. + * Uses the same anonymous gateway path as `fetchModelCapability` (no console + * token required), but adds `queryPrice: true` to include the prices array. + */ +async function fetchTrainingPrice( + settings: Settings, + model: string, +): Promise { + const eff = effectiveConsoleGatewayConfig(settings); + const result = await callConsoleGateway( + { region: eff.consoleRegion, site: eff.consoleSite, switchAgent: eff.consoleSwitchAgent }, + settings.timeout, + { + api: MODEL_LIST_API, + data: { + input: { + pageNo: 1, + pageSize: 10, + group: true, + model, + queryPrice: true, + querySampleCode: false, + queryGroupByModel: true, + queryQuota: false, + queryQpmInfo: false, + queryApplyStatus: false, + queryPermissions: false, + queryActivationStatus: false, + }, + }, + }, + ); + const responseData = unwrapResponse(result as Record); + const list = (responseData.list as Record[]) ?? []; + // The response is grouped; find the exact model in items. + for (const group of list) { + const items = (group.items as Record[]) ?? []; + for (const item of items) { + if (item.model === model) { + const prices = (item.prices as ModelPriceInfo[]) ?? []; + return prices.find((entry) => entry.type === "ft") ?? null; + } + } + // Flat response fallback (no items nesting). + if (group.model === model) { + const prices = (group.prices as ModelPriceInfo[]) ?? []; + return prices.find((entry) => entry.type === "ft") ?? null; + } + } + return null; +} + +/** + * Compute the actual training fee from the model catalog. + * Returns null when the price is unavailable (network error, model not in + * catalog, or no "ft" entry in the prices array). Never throws. + */ +export async function computeActualFee( + settings: Settings, + model: string, + usageTokens: number, +): Promise { + try { + const ftEntry = await fetchTrainingPrice(settings, model); + const unitPrice = Number(ftEntry?.price); + if (!Number.isFinite(unitPrice) || unitPrice <= 0) return null; + const priceUnit = ftEntry?.priceUnit ?? "每百万tokens"; + // price is yuan per million tokens. + const cost = (usageTokens / 1_000_000) * unitPrice; + return { cost: Number(cost.toFixed(4)), unitPrice, priceUnit }; + } catch { + return null; + } +} diff --git a/packages/commands/src/commands/finetune/get.ts b/packages/commands/src/commands/finetune/get.ts index 05d96fb7..0ec4d5a9 100644 --- a/packages/commands/src/commands/finetune/get.ts +++ b/packages/commands/src/commands/finetune/get.ts @@ -1,5 +1,6 @@ import { defineCommand, getFineTune, type FlagsDef } from "bailian-cli-core"; import { emitResult } from "bailian-cli-runtime"; +import { computeActualFee } from "./fee.ts"; const GET_FLAGS = { jobId: { @@ -44,7 +45,9 @@ export default defineCommand({ if (hyperParameters?.max_length !== undefined) hyperParts.push(`max_length=${hyperParameters.max_length}`); - const item = { + const usageTokens = typeof job.usage === "number" ? job.usage : undefined; + + const item: Record = { job_id: job.job_id ?? jobId, base_model: job.model ?? "", status: job.status ?? "", @@ -56,10 +59,20 @@ export default defineCommand({ model_name: job.model_name ?? "", created_at: job.create_time ?? job.gmt_create ?? "", updated_at: job.end_time ?? job.gmt_modified ?? "", - usage: typeof job.usage === "number" ? String(job.usage) : "", + usage_tokens: usageTokens ?? "", charge_type: typeof job.charge_type === "string" ? job.charge_type : "", }; + // Actual fee: only when the platform reports a concrete token count + // (SUCCEEDED / CANCELED). Best-effort — silently omitted on lookup failure. + if (usageTokens !== undefined && usageTokens > 0 && job.model) { + const fee = await computeActualFee(settings, job.model, usageTokens); + if (fee) { + item.training_cost = fee.cost; + item.cost_basis = `${fee.unitPrice} 元/${fee.priceUnit}`; + } + } + emitResult({ ...item, request_id: response.request_id }, "json"); }, }); diff --git a/packages/commands/src/commands/finetune/list.ts b/packages/commands/src/commands/finetune/list.ts index 529a5fd0..d37fc5be 100644 --- a/packages/commands/src/commands/finetune/list.ts +++ b/packages/commands/src/commands/finetune/list.ts @@ -13,26 +13,35 @@ const LIST_FLAGS = { valueHint: "", description: "Filter by status (PENDING / RUNNING / SUCCEEDED / FAILED / CANCELED)", }, + baseModel: { + type: "string", + valueHint: "", + description: "Filter by base model ID (server-side)", + }, } satisfies FlagsDef; export default defineCommand({ description: "List fine-tune jobs", auth: "apiKey", - usageArgs: "[--page ] [--page-size ] [--status ]", + usageArgs: "[--page ] [--page-size ] [--status ] [--base-model ]", flags: LIST_FLAGS, - exampleArgs: ["", "--status RUNNING", "--page-size 20 --output json"], + exampleArgs: ["", "--status RUNNING", "--base-model qwen3-8b", "--page-size 20"], async run(ctx) { const { settings, flags } = ctx; const pageNo = flags.page; const pageSize = flags.pageSize; const status = flags.status || undefined; + const model = flags.baseModel || undefined; if (settings.dryRun) { - emitResult({ action: "finetune.list", page: pageNo, page_size: pageSize, status }, "json"); + emitResult( + { action: "finetune.list", page: pageNo, page_size: pageSize, status, model }, + "json", + ); return; } - const response = await listFineTunes(ctx.client, { pageNo, pageSize, status }); + const response = await listFineTunes(ctx.client, { pageNo, pageSize, status, model }); const payload = response.output ?? response.data; const jobs = payload?.jobs ?? []; const total = payload?.total; diff --git a/packages/commands/src/commands/finetune/price.ts b/packages/commands/src/commands/finetune/price.ts new file mode 100644 index 00000000..32dd7562 --- /dev/null +++ b/packages/commands/src/commands/finetune/price.ts @@ -0,0 +1,139 @@ +import { + defineCommand, + fetchTrainingModelPrice, + estimateSftDpoTokens, + estimateCptTokens, + BailianError, + ExitCode, + type FlagsDef, +} from "bailian-cli-core"; +import { emitResult } from "bailian-cli-runtime"; + +const PRICE_FLAGS = { + baseModel: { + type: "string", + valueHint: "", + description: "Base model to fine-tune (e.g. qwen3-8b; not the output model name)", + required: true, + }, + datasets: { + type: "string", + valueHint: "", + description: "Training dataset file IDs, comma-separated (required)", + required: true, + }, + trainingType: { + type: "string", + valueHint: "", + description: "Training type: sft | dpo | cpt (default: sft)", + }, + nEpochs: { + type: "number", + valueHint: "", + description: "Number of training epochs (default: 3)", + }, +} satisfies FlagsDef; + +const SUPPORTED_TRAINING_TYPES = ["sft", "dpo", "cpt"]; + +// Fixed hyper-parameters used for estimation. Only n_epochs materially affects +// the estimate; the rest are held at representative defaults (not exposed as +// flags to keep the command surface minimal). +const ESTIMATE_BATCH_SIZE = 16; +const ESTIMATE_MAX_LENGTH = 8192; +const DEFAULT_N_EPOCHS = 3; + +export default defineCommand({ + description: "Estimate the training cost for a fine-tune job (token billing)", + auth: "console", + usageArgs: "--base-model --datasets [--training-type ] [--n-epochs ]", + flags: PRICE_FLAGS, + exampleArgs: [ + "--base-model qwen3-8b --datasets file-ft-xxx", + "--base-model qwen3-8b --datasets file-ft-xxx,file-ft-yyy --n-epochs 2", + "--base-model qwen3-8b --datasets file-ft-xxx --training-type cpt", + ], + notes: [ + "Estimate only — the server computes token usage from the datasets; final cost is subject to the bill.", + "Covers token billing for sft / dpo / cpt. Training-unit (MTU) billing is not supported by this command.", + "Hyper-parameters other than --n-epochs are fixed at representative defaults for estimation.", + ], + async run(ctx) { + const { settings, flags } = ctx; + const model = flags.baseModel; + const datasetIds = flags.datasets + .split(",") + .map((datasetId) => datasetId.trim()) + .filter(Boolean); + const trainingType = (flags.trainingType ?? "sft").toLowerCase(); + const nEpochs = flags.nEpochs ?? DEFAULT_N_EPOCHS; + + if (!SUPPORTED_TRAINING_TYPES.includes(trainingType)) { + throw new BailianError( + `Unsupported training type "${trainingType}". Supported: ${SUPPORTED_TRAINING_TYPES.join(", ")}.`, + ExitCode.USAGE, + ); + } + if (datasetIds.length === 0) { + throw new BailianError("--datasets must contain at least one file ID.", ExitCode.USAGE); + } + + if (settings.dryRun) { + emitResult( + { action: "finetune.price", model, datasets: datasetIds, trainingType, nEpochs }, + "json", + ); + return; + } + + // Unit price (yuan per 千Token). + const priceInfo = await fetchTrainingModelPrice(ctx.client, model); + const unitPrice = Number(priceInfo.price); + if (!Number.isFinite(unitPrice)) { + throw new BailianError( + `No training price found for model "${model}".`, + ExitCode.GENERAL, + undefined, + { rawResponse: JSON.stringify(priceInfo) }, + ); + } + + // Per-epoch token estimate (min/max range). + const estimate = + trainingType === "cpt" + ? await estimateCptTokens(ctx.client, model, datasetIds.join(","), nEpochs) + : await estimateSftDpoTokens(ctx.client, datasetIds, { + nEpochs, + batchSize: ESTIMATE_BATCH_SIZE, + maxLength: ESTIMATE_MAX_LENGTH, + }); + + const minPerEpoch = estimate.estimatedDatasetConsumedTokensMinPerEpoch ?? 0; + const maxPerEpoch = estimate.estimatedDatasetConsumedTokensMaxPerEpoch ?? 0; + const mixedMinPerEpoch = estimate.estimatedMixedConsumedTokensMinPerEpoch ?? 0; + const mixedMaxPerEpoch = estimate.estimatedMixedConsumedTokensMaxPerEpoch ?? 0; + + const minTokens = (minPerEpoch + mixedMinPerEpoch) * nEpochs; + const maxTokens = (maxPerEpoch + mixedMaxPerEpoch) * nEpochs; + // price is yuan per 1000 tokens. + const minFee = (minTokens / 1000) * unitPrice; + const maxFee = (maxTokens / 1000) * unitPrice; + + emitResult( + { + model, + training_type: trainingType, + n_epochs: nEpochs, + unit_price: unitPrice, + price_unit: priceInfo.priceUnit ?? "千Token", + estimated_tokens: { min: minTokens, max: maxTokens }, + estimated_fee_yuan: { + min: Number(minFee.toFixed(4)), + max: Number(maxFee.toFixed(4)), + }, + disclaimer: "Server-side estimate; final cost is subject to the bill.", + }, + "json", + ); + }, +}); diff --git a/packages/commands/src/commands/finetune/watch.ts b/packages/commands/src/commands/finetune/watch.ts index 8baf7380..eb802a5d 100644 --- a/packages/commands/src/commands/finetune/watch.ts +++ b/packages/commands/src/commands/finetune/watch.ts @@ -6,6 +6,7 @@ import { type FlagsDef, } from "bailian-cli-core"; import { emitResult, emitBare } from "bailian-cli-runtime"; +import { computeActualFee } from "./fee.ts"; const DEFAULT_INTERVAL_SEC = 10; const MIN_INTERVAL_SEC = 1; @@ -131,10 +132,23 @@ export default defineCommand({ // Just the status word — ideal for `status=$(... finetune watch ... --quiet)`. emitBare(status || "UNKNOWN"); } else { - emitResult( - { job_id: jobId, status: status || "UNKNOWN", terminal, request_id: response.request_id }, - "json", - ); + const output: Record = { + job_id: jobId, + status: status || "UNKNOWN", + terminal, + request_id: response.request_id, + }; + // Enrich terminal output with actual fee when usage is reported. + const usageTokens = typeof job?.usage === "number" ? job.usage : undefined; + if (terminal && usageTokens && usageTokens > 0 && job?.model) { + output.usage_tokens = usageTokens; + const fee = await computeActualFee(settings, job.model as string, usageTokens); + if (fee) { + output.training_cost = fee.cost; + output.cost_basis = `${fee.unitPrice} 元/${fee.priceUnit}`; + } + } + emitResult(output, "json"); } if (terminal && status !== "SUCCEEDED") { @@ -171,7 +185,18 @@ export default defineCommand({ if (settings.quiet) { emitBare(status || "UNKNOWN"); } else { - emitResult(response, "json"); + // Enrich the raw response with actual fee when usage is available. + const usageTokens = typeof job?.usage === "number" ? job.usage : undefined; + const enriched: Record = { ...response }; + if (usageTokens && usageTokens > 0 && job?.model) { + const fee = await computeActualFee(settings, job.model as string, usageTokens); + if (fee) { + enriched.training_cost = fee.cost; + enriched.usage_tokens = usageTokens; + enriched.cost_basis = `${fee.unitPrice} 元/${fee.priceUnit}`; + } + } + emitResult(enriched, "json"); } if (status !== "SUCCEEDED") { throw new BailianError( diff --git a/packages/commands/src/index.ts b/packages/commands/src/index.ts index 63a13163..848fcb72 100644 --- a/packages/commands/src/index.ts +++ b/packages/commands/src/index.ts @@ -76,6 +76,7 @@ export { default as finetuneCheckpoints } from "./commands/finetune/checkpoints. export { default as finetuneExport } from "./commands/finetune/export.ts"; export { default as finetuneWatch } from "./commands/finetune/watch.ts"; export { default as finetuneCapability } from "./commands/finetune/capability.ts"; +export { default as finetunePrice } from "./commands/finetune/price.ts"; export { deployTextCreate, deployAudioCreate, diff --git a/packages/commands/tests/e2e/deploy.e2e.test.ts b/packages/commands/tests/e2e/deploy.e2e.test.ts index 6fb5deda..47d5b814 100644 --- a/packages/commands/tests/e2e/deploy.e2e.test.ts +++ b/packages/commands/tests/e2e/deploy.e2e.test.ts @@ -30,7 +30,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: deploy (offline)", () => { "--help", ]); expect(exitCode, stderr).toBe(0); - expect(stderr).toMatch(/--model|--name/i); + expect(stderr).toMatch(/--model-name|--display-name/i); }); test("deploy create --dry-run 构造 lora 部署请求体", async () => { @@ -38,9 +38,9 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: deploy (offline)", () => { "deploy", "text", "create", - "--model", + "--model-name", "qwen-plus-2025-12-01", - "--name", + "--display-name", "my-qwen-plus", "--dry-run", "--output", @@ -68,9 +68,9 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: deploy (offline)", () => { "deploy", "text", "create", - "--model", + "--model-name", "qwen3-8b", - "--name", + "--display-name", "my-qwen3-mu", "--plan", "mu", @@ -102,9 +102,9 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: deploy (offline)", () => { "deploy", "audio", "create", - "--model", + "--model-name", "my-cosyvoice-ft", - "--name", + "--display-name", "my-tts", "--dry-run", "--output", diff --git a/packages/commands/tests/e2e/finetune.e2e.test.ts b/packages/commands/tests/e2e/finetune.e2e.test.ts index 59e9374d..a045b2ec 100644 --- a/packages/commands/tests/e2e/finetune.e2e.test.ts +++ b/packages/commands/tests/e2e/finetune.e2e.test.ts @@ -31,7 +31,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { "--help", ]); expect(exitCode, stderr).toBe(0); - expect(stderr).toMatch(/--model|--datasets/i); + expect(stderr).toMatch(/--base-model|--datasets/i); }); test("finetune create --dry-run 构造 SFT 默认请求体", async () => { @@ -39,7 +39,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { "finetune", "text", "create", - "--model", + "--base-model", "qwen3-8b", "--datasets", "file-aaa,file-bbb", @@ -73,7 +73,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { "finetune", "text", "create", - "--model", + "--base-model", "qwen3-8b", "--datasets", "file-aaa", @@ -135,7 +135,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { "finetune", "text", "create", - "--model", + "--base-model", "qwen3-8b", "--datasets", "file-aaa", @@ -157,7 +157,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { "finetune", "text", "create", - "--model", + "--base-model", "qwen3-8b", "--datasets", "file-aaa", @@ -176,7 +176,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { "finetune", "text", "create", - "--model", + "--base-model", "qwen3-8b", "--datasets", `${localPath},file-bbb`, @@ -207,7 +207,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { "finetune", "text", "create", - "--model", + "--base-model", "qwen3-8b", "--datasets", " , ", @@ -228,7 +228,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { "finetune", "text", "create", - "--model", + "--base-model", "qwen3-8b", "--datasets", localPath, @@ -250,7 +250,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { "finetune", "text", "create", - "--model", + "--base-model", "qwen3-8b", "--datasets", localPath, @@ -272,7 +272,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { ["cancel", ["--job-id", "ft-xxx"]], ["delete", ["--job-id", "ft-xxx"]], ["watch", ["--job-id", "ft-xxx"]], - ["capability", ["--model", "qwen3-8b"]], + ["capability", ["--base-model", "qwen3-8b"]], ])("finetune %s --dry-run 发出结构化动作", async (sub, extra) => { const { stdout, stderr, exitCode } = await runCommandE2e(FINETUNE_ROUTES, [ "finetune", @@ -292,7 +292,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { "finetune", "text", "create", - "--model", + "--base-model", "qwen3-8b", "--datasets", " file-a , ,file-b ", @@ -314,7 +314,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { "finetune", "audio", "create", - "--model", + "--base-model", "cosyvoice-v3-flash", "--datasets", "file-audio", @@ -343,7 +343,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { "--help", ]); expect(exitCode, stderr).toBe(0); - expect(stderr).toMatch(/--model|--datasets/i); + expect(stderr).toMatch(/--base-model|--datasets/i); expect(stderr).not.toMatch(/--training-type|--n-epochs|--batch-size|--max-length/); }); @@ -352,7 +352,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { "finetune", "image", "create", - "--model", + "--base-model", "wan2.7-image-pro", "--datasets", "file-image", diff --git a/packages/core/src/finetune/api.ts b/packages/core/src/finetune/api.ts index 1126c9dd..757c0440 100644 --- a/packages/core/src/finetune/api.ts +++ b/packages/core/src/finetune/api.ts @@ -43,6 +43,8 @@ export interface ListFineTunesParams { pageNo?: number; pageSize?: number; status?: string; + /** Filter by base model ID (server-side). */ + model?: string; signal?: AbortSignal; } @@ -55,6 +57,7 @@ export async function listFineTunes( if (params.pageNo !== undefined) qs.set("page_no", String(params.pageNo)); if (params.pageSize !== undefined) qs.set("page_size", String(params.pageSize)); if (params.status) qs.set("status", params.status); + if (params.model) qs.set("model", params.model); const base = finetuneJobsPath(); const path = qs.toString() ? `${base}?${qs.toString()}` : base; return client.requestJson({ diff --git a/packages/core/src/finetune/index.ts b/packages/core/src/finetune/index.ts index e0b064b6..3b82e583 100644 --- a/packages/core/src/finetune/index.ts +++ b/packages/core/src/finetune/index.ts @@ -3,3 +3,4 @@ export * from "./api.ts"; export * from "./capability.ts"; export * from "./preflight.ts"; export * from "./profiles/index.ts"; +export * from "./price.ts"; diff --git a/packages/core/src/finetune/price.ts b/packages/core/src/finetune/price.ts new file mode 100644 index 00000000..64651ae5 --- /dev/null +++ b/packages/core/src/finetune/price.ts @@ -0,0 +1,121 @@ +/** + * Training price estimation via the **console gateway**. + * + * These are console-domain APIs (`zeldaEasy.broadscope-platform.*`); commands + * using them must declare `auth: "console"`. Two different argument wrappers + * exist: `getModelPrice` takes a top-level `query`, while the token-estimation + * APIs take a top-level `input`. + */ +import type { Client } from "../client/client.ts"; +import { unwrapResponse } from "../console/models.ts"; + +// --------------------------------------------------------------------------- +// API names +// --------------------------------------------------------------------------- + +export const TRAINING_MODEL_PRICE_API = "zeldaEasy.broadscope-platform.modelCenter.getModelPrice"; +export const CALC_DATASETS_TOKENS_API = + "zeldaEasy.broadscope-platform.modelInstance.calculateDatasetsTotalTokens"; +export const ESTIMATE_FINETUNE_TOKENS_API = + "zeldaEasy.broadscope-platform.modelInstance.estimateFinetuneTokens"; + +// --------------------------------------------------------------------------- +// Types +// --------------------------------------------------------------------------- + +export interface TrainingModelPrice { + price?: string; + priceUnit?: string; + modelId?: string; + [key: string]: unknown; +} + +export interface TokenEstimate { + estimatedDatasetConsumedTokensMinPerEpoch?: number; + estimatedDatasetConsumedTokensMaxPerEpoch?: number; + estimatedMixedConsumedTokensMinPerEpoch?: number; + estimatedMixedConsumedTokensMaxPerEpoch?: number; + [key: string]: unknown; +} + +// --------------------------------------------------------------------------- +// API wrappers +// --------------------------------------------------------------------------- + +/** + * Training unit price for a model. `price` is denominated in `priceUnit` + * (typically "千Token" — yuan per 1000 tokens). + */ +export async function fetchTrainingModelPrice( + client: Client, + modelId: string, +): Promise { + const raw = await client.console>(TRAINING_MODEL_PRICE_API, { + query: { type: 0, modelId }, + }); + return unwrapResponse(raw) as TrainingModelPrice; +} + +/** + * Estimate training tokens for SFT / DPO jobs. + * Returns a per-epoch min/max range; multiply by `n_epochs` for the total. + */ +export async function estimateSftDpoTokens( + client: Client, + datasetIds: string[], + hyperParams: { nEpochs: number; batchSize: number; maxLength: number }, +): Promise { + const raw = await client.console>(CALC_DATASETS_TOKENS_API, { + input: { trainDatasetIds: datasetIds, hyperParams }, + }); + return unwrapResponse(raw) as TokenEstimate; +} + +/** + * Estimate training tokens for CPT jobs. + * + * The console API requires `hyperParams` as a **JSON string** with a full + * `userDefinedObj` payload (captured from the console frontend), plus several + * top-level fields (`algorithmType`, `bizType`, `priority`, …). Only + * `n_epochs` / `max_length` materially affect the estimate; the remaining + * hyper-parameters are fixed defaults. + */ +export async function estimateCptTokens( + client: Client, + model: string, + datasetIdsCsv: string, + nEpochs: number, +): Promise { + const userDefinedObj = { + batch_size: 16, + eval_steps: 50, + learning_rate: "7e-6", + lr_scheduler_type: "linear", + max_length: 8192, + n_epochs: nEpochs, + split: 0.9, + save_total_limit: "3", + resume_from_checkpoint: false, + save_strategy: "epoch", + }; + const hyperParams = JSON.stringify({ + useDefault: false, + userDefinedObj, + useQwenMixedStrategy: false, + }); + const raw = await client.console>(ESTIMATE_FINETUNE_TOKENS_API, { + input: { + trainingType: "cpt", + instanceName: `${model}_cli_estimate`, + algorithmType: 100, + bizType: 100, + trainDatasetIds: datasetIdsCsv, + hyperParams, + bailianTrainModel: model, + validationDatasetIds: "", + jobName: `${model}_cli_estimate`, + priority: "L0", + }, + }); + return unwrapResponse(raw) as TokenEstimate; +} diff --git a/skills/bailian-finetune/SKILL.md b/skills/bailian-finetune/SKILL.md index b647c8d9..d2187020 100644 --- a/skills/bailian-finetune/SKILL.md +++ b/skills/bailian-finetune/SKILL.md @@ -23,14 +23,14 @@ description: >- ``` 1. Validate data bl dataset validate --file train.jsonl [--schema chatml|dpo|cpt|tts|image] 2. Upload data bl dataset upload --file train.jsonl # returns a file-id -3. Create job bl finetune text|audio|image create --model --datasets +3. Create job bl finetune text|audio|image create --base-model --datasets 4. Watch progress bl finetune watch --job-id ft-xxx # or get / logs 5. Pick artifact bl finetune checkpoints --job-id ft-xxx 6. Export model bl finetune export --job-id ft-xxx --checkpoint ckpt-N --model-name my-model -7. Deploy service bl deploy text|audio|image create --model my-model --name my-svc +7. Deploy service bl deploy text|audio|image create --model-name my-model --display-name my-svc ``` -- Unsure which training methods a base model supports → `bl finetune capability --model ` or `--training-type sft|sft-lora|dpo|cpt`. +- Unsure which training methods a base model supports → `bl finetune capability --base-model ` or `--training-type sft|sft-lora|dpo|cpt`. - Text `--training-type` values: `sft` / `sft-lora` / `dpo` / `dpo-lora` / `cpt`. Audio bases include `cosyvoice-v3-flash`; image bases include `wan2.7-image-pro`. - Deployment plans: audio defaults to `--plan mu`; text/image default to `lora`. - Preview write operations (create / delete / cancel / scale) with `--dry-run` first, and confirm with the user before deleting a job or dataset. @@ -55,10 +55,10 @@ Flags, usage, and examples: see [`reference/`](reference/index.md) or `bl --name [--plan ] [--deploy-spec ] [--capacity ] [--billing-method ] [--input-tpm ] [--output-tpm ] [--thinking-output-tpm ]` | +| Field | Value | +| --------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | +| **Name** | `deploy audio create` | +| **Description** | Create an audio (TTS) model deployment | +| **Usage** | `bl deploy audio create --model-name --display-name [--plan ] [--deploy-spec ] [--capacity ] [--billing-method ] [--input-tpm ] [--output-tpm ] [--thinking-output-tpm ]` | #### Flags -| Flag | Type | Required | Description | -| --------------------------- | ------ | -------- | ------------------------------------------------------------------------------- | -| `--model ` | string | yes | Model name (catalog model or fine-tuned output) (required) | -| `--name ` | string | yes | Console display name for the deployment (required) | -| `--plan ` | string | no | Billing plan: lora (default, Token-billed) \| ptu (Token-billed) \| mu | -| `--deploy-spec ` | string | no | Deploy spec (only used by plan=mu; auto-picked if omitted) | -| `--capacity ` | number | no | Resource units (plan=mu only; required by API; defaults to the template's unit) | -| `--billing-method ` | string | no | Billing method (plan=mu only; default "POST_PAY", the only supported value) | -| `--input-tpm ` | number | no | PTU max input tokens/min (required for plan=ptu) | -| `--output-tpm ` | number | no | PTU max output tokens/min (required for plan=ptu) | -| `--thinking-output-tpm ` | number | no | PTU max thinking-output tokens/min (optional, some models) | -| `--api-key ` | string | no | API key | -| `--base-url ` | string | no | API base URL | +| Flag | Type | Required | Description | +| ------------------------------- | ------ | -------- | ------------------------------------------------------------------------------- | +| `--model-name ` | string | yes | Model to deploy — fine-tuned output name or catalog model (required) | +| `--display-name ` | string | yes | Console display name for the deployment (required) | +| `--plan ` | string | no | Billing plan: lora (default, Token-billed) \| ptu (Token-billed) \| mu | +| `--deploy-spec ` | string | no | Deploy spec (only used by plan=mu; auto-picked if omitted) | +| `--capacity ` | number | no | Resource units (plan=mu only; required by API; defaults to the template's unit) | +| `--billing-method ` | string | no | Billing method (plan=mu only; default "POST_PAY", the only supported value) | +| `--input-tpm ` | number | no | PTU max input tokens/min (required for plan=ptu) | +| `--output-tpm ` | number | no | PTU max output tokens/min (required for plan=ptu) | +| `--thinking-output-tpm ` | number | no | PTU max thinking-output tokens/min (optional, some models) | +| `--api-key ` | string | no | API key | +| `--base-url ` | string | no | API base URL | #### Notes @@ -60,27 +60,24 @@ Index: [index.md](index.md) - Use `bl deploy models --source base` to inspect available templates. - After creation, status starts at PENDING and transitions to RUNNING. - Invoke the deployed model with: bl text chat --model -- WARNING: --model is overloaded across commands and refers to DIFFERENT -- values. `bl deploy create --model` takes the exported model_name -- (e.g. `qwen3-8b-ft-...`), but the create response also returns a -- `deployed_model` field (the deployment instance id, e.g. -- `qwen3-8b-5ecb5f068d79`). The inference call `bl text chat --model` must use -- the `deployed_model` from the create response — NOT the `model_name` you -- passed to `deploy create`. Do not reuse the value across the two -- commands. +- NOTE: --model-name is the model being deployed (e.g. `qwen3-8b-ft-...`). +- The create response also returns a `deployed_model` field — the deployment +- instance id (e.g. `qwen3-8b-5ecb5f068d79`). Use that id for inference +- (`bl text chat --model `) and lifecycle commands +- (`deploy get/scale/pause/resume/delete --deployed-model `). #### Examples ```bash -bl deploy audio create --model my-cosyvoice-ft --name my-tts +bl deploy audio create --model-name my-cosyvoice-ft --display-name my-tts ``` ```bash -bl deploy audio create --model my-cosyvoice-ft --name my-tts --deploy-spec dps-xxxx --capacity 1 +bl deploy audio create --model-name my-cosyvoice-ft --display-name my-tts --deploy-spec dps-xxxx --capacity 1 ``` ```bash -bl deploy audio create --model my-cosyvoice-ft --name my-tts --dry-run +bl deploy audio create --model-name my-cosyvoice-ft --display-name my-tts --dry-run ``` ### `bl deploy delete` @@ -138,27 +135,27 @@ bl deploy get --deployed-model qwen-plus-2025-12-01-b6d61c71 --output json ### `bl deploy image create` -| Field | Value | -| --------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -| **Name** | `deploy image create` | -| **Description** | Create an image generation model deployment | -| **Usage** | `bl deploy image create --model --name [--plan ] [--deploy-spec ] [--capacity ] [--billing-method ] [--input-tpm ] [--output-tpm ] [--thinking-output-tpm ]` | +| Field | Value | +| --------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | +| **Name** | `deploy image create` | +| **Description** | Create an image generation model deployment | +| **Usage** | `bl deploy image create --model-name --display-name [--plan ] [--deploy-spec ] [--capacity ] [--billing-method ] [--input-tpm ] [--output-tpm ] [--thinking-output-tpm ]` | #### Flags -| Flag | Type | Required | Description | -| --------------------------- | ------ | -------- | ------------------------------------------------------------------------------- | -| `--model ` | string | yes | Model name (catalog model or fine-tuned output) (required) | -| `--name ` | string | yes | Console display name for the deployment (required) | -| `--plan ` | string | no | Billing plan: lora (default, Token-billed) \| ptu (Token-billed) \| mu | -| `--deploy-spec ` | string | no | Deploy spec (only used by plan=mu; auto-picked if omitted) | -| `--capacity ` | number | no | Resource units (plan=mu only; required by API; defaults to the template's unit) | -| `--billing-method ` | string | no | Billing method (plan=mu only; default "POST_PAY", the only supported value) | -| `--input-tpm ` | number | no | PTU max input tokens/min (required for plan=ptu) | -| `--output-tpm ` | number | no | PTU max output tokens/min (required for plan=ptu) | -| `--thinking-output-tpm ` | number | no | PTU max thinking-output tokens/min (optional, some models) | -| `--api-key ` | string | no | API key | -| `--base-url ` | string | no | API base URL | +| Flag | Type | Required | Description | +| ------------------------------- | ------ | -------- | ------------------------------------------------------------------------------- | +| `--model-name ` | string | yes | Model to deploy — fine-tuned output name or catalog model (required) | +| `--display-name ` | string | yes | Console display name for the deployment (required) | +| `--plan ` | string | no | Billing plan: lora (default, Token-billed) \| ptu (Token-billed) \| mu | +| `--deploy-spec ` | string | no | Deploy spec (only used by plan=mu; auto-picked if omitted) | +| `--capacity ` | number | no | Resource units (plan=mu only; required by API; defaults to the template's unit) | +| `--billing-method ` | string | no | Billing method (plan=mu only; default "POST_PAY", the only supported value) | +| `--input-tpm ` | number | no | PTU max input tokens/min (required for plan=ptu) | +| `--output-tpm ` | number | no | PTU max output tokens/min (required for plan=ptu) | +| `--thinking-output-tpm ` | number | no | PTU max thinking-output tokens/min (optional, some models) | +| `--api-key ` | string | no | API key | +| `--base-url ` | string | no | API base URL | #### Notes @@ -173,27 +170,24 @@ bl deploy get --deployed-model qwen-plus-2025-12-01-b6d61c71 --output json - Use `bl deploy models --source base` to inspect available templates. - After creation, status starts at PENDING and transitions to RUNNING. - Invoke the deployed model with: bl text chat --model -- WARNING: --model is overloaded across commands and refers to DIFFERENT -- values. `bl deploy create --model` takes the exported model_name -- (e.g. `qwen3-8b-ft-...`), but the create response also returns a -- `deployed_model` field (the deployment instance id, e.g. -- `qwen3-8b-5ecb5f068d79`). The inference call `bl text chat --model` must use -- the `deployed_model` from the create response — NOT the `model_name` you -- passed to `deploy create`. Do not reuse the value across the two -- commands. +- NOTE: --model-name is the model being deployed (e.g. `qwen3-8b-ft-...`). +- The create response also returns a `deployed_model` field — the deployment +- instance id (e.g. `qwen3-8b-5ecb5f068d79`). Use that id for inference +- (`bl text chat --model `) and lifecycle commands +- (`deploy get/scale/pause/resume/delete --deployed-model `). #### Examples ```bash -bl deploy image create --model my-wan-ft --name my-wan +bl deploy image create --model-name my-wan-ft --display-name my-wan ``` ```bash -bl deploy image create --model my-wan-ft --name my-wan-mu --plan mu +bl deploy image create --model-name my-wan-ft --display-name my-wan-mu --plan mu ``` ```bash -bl deploy image create --model my-wan-ft --name my-wan --dry-run +bl deploy image create --model-name my-wan-ft --display-name my-wan --dry-run ``` ### `bl deploy list` @@ -372,27 +366,27 @@ bl deploy scale --deployed-model dep-... --capacity 2 ### `bl deploy text create` -| Field | Value | -| --------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -| **Name** | `deploy text create` | -| **Description** | Create a text model deployment | -| **Usage** | `bl deploy text create --model --name [--plan ] [--deploy-spec ] [--capacity ] [--billing-method ] [--input-tpm ] [--output-tpm ] [--thinking-output-tpm ]` | +| Field | Value | +| --------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | +| **Name** | `deploy text create` | +| **Description** | Create a text model deployment | +| **Usage** | `bl deploy text create --model-name --display-name [--plan ] [--deploy-spec ] [--capacity ] [--billing-method ] [--input-tpm ] [--output-tpm ] [--thinking-output-tpm ]` | #### Flags -| Flag | Type | Required | Description | -| --------------------------- | ------ | -------- | ------------------------------------------------------------------------------- | -| `--model ` | string | yes | Model name (catalog model or fine-tuned output) (required) | -| `--name ` | string | yes | Console display name for the deployment (required) | -| `--plan ` | string | no | Billing plan: lora (default, Token-billed) \| ptu (Token-billed) \| mu | -| `--deploy-spec ` | string | no | Deploy spec (only used by plan=mu; auto-picked if omitted) | -| `--capacity ` | number | no | Resource units (plan=mu only; required by API; defaults to the template's unit) | -| `--billing-method ` | string | no | Billing method (plan=mu only; default "POST_PAY", the only supported value) | -| `--input-tpm ` | number | no | PTU max input tokens/min (required for plan=ptu) | -| `--output-tpm ` | number | no | PTU max output tokens/min (required for plan=ptu) | -| `--thinking-output-tpm ` | number | no | PTU max thinking-output tokens/min (optional, some models) | -| `--api-key ` | string | no | API key | -| `--base-url ` | string | no | API base URL | +| Flag | Type | Required | Description | +| ------------------------------- | ------ | -------- | ------------------------------------------------------------------------------- | +| `--model-name ` | string | yes | Model to deploy — fine-tuned output name or catalog model (required) | +| `--display-name ` | string | yes | Console display name for the deployment (required) | +| `--plan ` | string | no | Billing plan: lora (default, Token-billed) \| ptu (Token-billed) \| mu | +| `--deploy-spec ` | string | no | Deploy spec (only used by plan=mu; auto-picked if omitted) | +| `--capacity ` | number | no | Resource units (plan=mu only; required by API; defaults to the template's unit) | +| `--billing-method ` | string | no | Billing method (plan=mu only; default "POST_PAY", the only supported value) | +| `--input-tpm ` | number | no | PTU max input tokens/min (required for plan=ptu) | +| `--output-tpm ` | number | no | PTU max output tokens/min (required for plan=ptu) | +| `--thinking-output-tpm ` | number | no | PTU max thinking-output tokens/min (optional, some models) | +| `--api-key ` | string | no | API key | +| `--base-url ` | string | no | API base URL | #### Notes @@ -407,31 +401,28 @@ bl deploy scale --deployed-model dep-... --capacity 2 - Use `bl deploy models --source base` to inspect available templates. - After creation, status starts at PENDING and transitions to RUNNING. - Invoke the deployed model with: bl text chat --model -- WARNING: --model is overloaded across commands and refers to DIFFERENT -- values. `bl deploy create --model` takes the exported model_name -- (e.g. `qwen3-8b-ft-...`), but the create response also returns a -- `deployed_model` field (the deployment instance id, e.g. -- `qwen3-8b-5ecb5f068d79`). The inference call `bl text chat --model` must use -- the `deployed_model` from the create response — NOT the `model_name` you -- passed to `deploy create`. Do not reuse the value across the two -- commands. +- NOTE: --model-name is the model being deployed (e.g. `qwen3-8b-ft-...`). +- The create response also returns a `deployed_model` field — the deployment +- instance id (e.g. `qwen3-8b-5ecb5f068d79`). Use that id for inference +- (`bl text chat --model `) and lifecycle commands +- (`deploy get/scale/pause/resume/delete --deployed-model `). #### Examples ```bash -bl deploy text create --model my-qwen-sft --name my-sft-test +bl deploy text create --model-name my-qwen-sft --display-name my-sft-test ``` ```bash -bl deploy text create --model qwen3.6-flash-2026-04-16 --name my-flash --plan ptu --input-tpm 10000 --output-tpm 1000 +bl deploy text create --model-name qwen3.6-flash-2026-04-16 --display-name my-flash --plan ptu --input-tpm 10000 --output-tpm 1000 ``` ```bash -bl deploy text create --model qwen3-8b --name my-qwen3-mu --plan mu +bl deploy text create --model-name qwen3-8b --display-name my-qwen3-mu --plan mu ``` ```bash -bl deploy text create --model qwen3-8b --name my-qwen3 --plan mu --deploy-spec MU1 --capacity 2 +bl deploy text create --model-name qwen3-8b --display-name my-qwen3 --plan mu --deploy-spec MU1 --capacity 2 ``` ### `bl deploy update` diff --git a/skills/bailian-finetune/reference/finetune.md b/skills/bailian-finetune/reference/finetune.md index 0b05ca09..178adb27 100644 --- a/skills/bailian-finetune/reference/finetune.md +++ b/skills/bailian-finetune/reference/finetune.md @@ -19,6 +19,7 @@ Index: [index.md](index.md) | `bl finetune image create` | Create an image generation model fine-tune job (sft-lora) | | `bl finetune list` | List fine-tune jobs | | `bl finetune logs` | Fetch training logs for a fine-tune job | +| `bl finetune price` | Estimate the training cost for a fine-tune job (token billing) | | `bl finetune text create` | Create a text model fine-tune job (sft \| sft-lora \| dpo \| dpo-lora \| cpt) | | `bl finetune watch` | Probe a fine-tune job's status (default: single non-blocking fetch). Pass --follow to poll until terminal. | @@ -26,17 +27,17 @@ Index: [index.md](index.md) ### `bl finetune audio create` -| Field | Value | -| --------------- | ----------------------------------------------------------------------------------------------------------------------------------- | -| **Name** | `finetune audio create` | -| **Description** | Create an audio TTS model fine-tune job (sft-lora) | -| **Usage** | `bl finetune audio create --model --datasets [--validations ] [--model-name ] [--suffix ]` | +| Field | Value | +| --------------- | ---------------------------------------------------------------------------------------------------------------------------------------- | +| **Name** | `finetune audio create` | +| **Description** | Create an audio TTS model fine-tune job (sft-lora) | +| **Usage** | `bl finetune audio create --base-model --datasets [--validations ] [--model-name ] [--suffix ]` | #### Flags | Flag | Type | Required | Description | | ---------------------------- | ------ | -------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------ | -| `--model ` | string | yes | Base model to fine-tune | +| `--base-model ` | string | yes | Base model to fine-tune (e.g. qwen3-8b; not the output model name) | | `--datasets ` | string | yes | Comma-separated dataset file IDs or local paths (.jsonl for text, .zip for audio/image). Local paths are uploaded (validated) first, then their file-ids are used. | | `--validations ` | string | no | Comma-separated validation dataset file IDs or local paths (auto-uploaded like --datasets). | | `--model-name ` | string | no | Output model name (after training) | @@ -58,23 +59,23 @@ Index: [index.md](index.md) #### Examples ```bash -bl finetune audio create --model cosyvoice-v3-flash --datasets ./audio.zip +bl finetune audio create --base-model cosyvoice-v3-flash --datasets ./audio.zip ``` ```bash -bl finetune audio create --model cosyvoice-v3-flash --datasets file-xxx +bl finetune audio create --base-model cosyvoice-v3-flash --datasets file-xxx ``` ```bash -bl finetune audio create --model cosyvoice-v3-flash --datasets ./audio.zip --model-name my-tts +bl finetune audio create --base-model cosyvoice-v3-flash --datasets ./audio.zip --model-name my-tts ``` ```bash -bl finetune audio create --model cosyvoice-v3-flash --datasets file-xxx --output json +bl finetune audio create --base-model cosyvoice-v3-flash --datasets file-xxx --output json ``` ```bash -bl finetune audio create --model cosyvoice-v3-flash --datasets ./audio.zip --dry-run +bl finetune audio create --base-model cosyvoice-v3-flash --datasets ./audio.zip --dry-run ``` ### `bl finetune cancel` @@ -114,18 +115,18 @@ bl finetune cancel --job-id ft-xxx --dry-run | --------------- | ------------------------------------------------------------------------------------------------------------------------------- | | **Name** | `finetune capability` | | **Description** | Query fine-tune training capability — by model (which training types it supports) or by training type (which models support it) | -| **Usage** | `bl finetune capability --model \| --training-type ` | +| **Usage** | `bl finetune capability --base-model \| --training-type ` | #### Flags | Flag | Type | Required | Description | | --------------------- | ------ | -------- | ------------------------------------------------------------------------------------- | -| `--model ` | string | no | List training types supported by this base model. | +| `--base-model ` | string | no | List training types supported by this base model. | | `--training-type ` | string | no | List models supporting this training type: sft \| sft-lora \| dpo \| dpo-lora \| cpt. | #### Notes -- Exactly one of --model / --training-type is required. +- Exactly one of --base-model / --training-type is required. - Training-type values use the `` / `-lora` convention: - sft | sft-lora | dpo | dpo-lora | cpt. (cpt has no -lora variant server-side.) - Queries listFoundationModels, a public API — no console login needed. @@ -133,7 +134,7 @@ bl finetune cancel --job-id ft-xxx --dry-run #### Examples ```bash -bl finetune capability --model qwen3-8b +bl finetune capability --base-model qwen3-8b ``` ```bash @@ -166,7 +167,7 @@ bl finetune capability --training-type sft --quiet #### Notes -- `model_name` (shown for SUCCEEDED checkpoints) is the direct input for `deploy create --model`. +- `model_name` (shown for SUCCEEDED checkpoints) is the direct input for `deploy create --model-name`. - Checkpoints expire ~15 days after creation; `expire_time` shows the deadline. Export or deploy before expiry. #### Examples @@ -268,17 +269,17 @@ bl finetune get --job-id ft-xxx --output json ### `bl finetune image create` -| Field | Value | -| --------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | -| **Name** | `finetune image create` | -| **Description** | Create an image generation model fine-tune job (sft-lora) | -| **Usage** | `bl finetune image create --model --datasets [--validations ] [--model-name ] [--suffix ] [--generation-type ] [--learning-rate ]` | +| Field | Value | +| --------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | +| **Name** | `finetune image create` | +| **Description** | Create an image generation model fine-tune job (sft-lora) | +| **Usage** | `bl finetune image create --base-model --datasets [--validations ] [--model-name ] [--suffix ] [--generation-type ] [--learning-rate ]` | #### Flags | Flag | Type | Required | Description | | ------------------------------ | ------ | -------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -| `--model ` | string | yes | Base model to fine-tune | +| `--base-model ` | string | yes | Base model to fine-tune (e.g. qwen3-8b; not the output model name) | | `--datasets ` | string | yes | Comma-separated dataset file IDs or local paths (.jsonl for text, .zip for audio/image). Local paths are uploaded (validated) first, then their file-ids are used. | | `--validations ` | string | no | Comma-separated validation dataset file IDs or local paths (auto-uploaded like --datasets). | | `--model-name ` | string | no | Output model name (after training) | @@ -304,46 +305,47 @@ bl finetune get --job-id ft-xxx --output json #### Examples ```bash -bl finetune image create --model wan2.7-image-pro --datasets ./images.zip +bl finetune image create --base-model wan2.7-image-pro --datasets ./images.zip ``` ```bash -bl finetune image create --model wan2.7-image-pro --datasets file-xxx +bl finetune image create --base-model wan2.7-image-pro --datasets file-xxx ``` ```bash -bl finetune image create --model wan2.7-image-pro --datasets file-xxx --generation-type i2i +bl finetune image create --base-model wan2.7-image-pro --datasets file-xxx --generation-type i2i ``` ```bash -bl finetune image create --model wan2.7-image-pro --datasets ./images.zip --model-name my-wan +bl finetune image create --base-model wan2.7-image-pro --datasets ./images.zip --model-name my-wan ``` ```bash -bl finetune image create --model wan2.7-image-pro --datasets file-xxx --output json +bl finetune image create --base-model wan2.7-image-pro --datasets file-xxx --output json ``` ```bash -bl finetune image create --model wan2.7-image-pro --datasets ./images.zip --dry-run +bl finetune image create --base-model wan2.7-image-pro --datasets ./images.zip --dry-run ``` ### `bl finetune list` -| Field | Value | -| --------------- | ---------------------------------------------------------------- | -| **Name** | `finetune list` | -| **Description** | List fine-tune jobs | -| **Usage** | `bl finetune list [--page ] [--page-size ] [--status ]` | +| Field | Value | +| --------------- | --------------------------------------------------------------------------------------- | +| **Name** | `finetune list` | +| **Description** | List fine-tune jobs | +| **Usage** | `bl finetune list [--page ] [--page-size ] [--status ] [--base-model ]` | #### Flags -| Flag | Type | Required | Description | -| ------------------ | ------ | -------- | -------------------------------------------------------------------- | -| `--page ` | number | no | Page number (default: 1) | -| `--page-size ` | number | no | Results per page (default: 10, max 100) | -| `--status ` | string | no | Filter by status (PENDING / RUNNING / SUCCEEDED / FAILED / CANCELED) | -| `--api-key ` | string | no | API key | -| `--base-url ` | string | no | API base URL | +| Flag | Type | Required | Description | +| ---------------------- | ------ | -------- | -------------------------------------------------------------------- | +| `--page ` | number | no | Page number (default: 1) | +| `--page-size ` | number | no | Results per page (default: 10, max 100) | +| `--status ` | string | no | Filter by status (PENDING / RUNNING / SUCCEEDED / FAILED / CANCELED) | +| `--base-model ` | string | no | Filter by base model ID (server-side) | +| `--api-key ` | string | no | API key | +| `--base-url ` | string | no | API base URL | #### Examples @@ -356,7 +358,11 @@ bl finetune list --status RUNNING ``` ```bash -bl finetune list --page-size 20 --output json +bl finetune list --base-model qwen3-8b +``` + +```bash +bl finetune list --page-size 20 ``` ### `bl finetune logs` @@ -405,19 +411,60 @@ bl finetune logs --job-id ft-xxx --tail 20 bl finetune logs --job-id ft-xxx --search checkpoint --tail 5 ``` +### `bl finetune price` + +| Field | Value | +| --------------- | --------------------------------------------------------------------------------------------------- | +| **Name** | `finetune price` | +| **Description** | Estimate the training cost for a fine-tune job (token billing) | +| **Usage** | `bl finetune price --base-model --datasets [--training-type ] [--n-epochs ]` | + +#### Flags + +| Flag | Type | Required | Description | +| ------------------------------ | ------ | -------- | ------------------------------------------------------------------ | +| `--base-model ` | string | yes | Base model to fine-tune (e.g. qwen3-8b; not the output model name) | +| `--datasets ` | string | yes | Training dataset file IDs, comma-separated (required) | +| `--training-type ` | string | no | Training type: sft \| dpo \| cpt (default: sft) | +| `--n-epochs ` | number | no | Number of training epochs (default: 3) | +| `--console-region ` | string | no | Console gateway region (e.g. cn-beijing, ap-southeast-1) | +| `--console-site ` | string | no | Console site: domestic, international | +| `--console-switch-agent ` | number | no | Switch agent UID for delegated access | +| `--workspace-id ` | string | no | Workspace ID (env: BAILIAN_WORKSPACE_ID) | + +#### Notes + +- Estimate only — the server computes token usage from the datasets; final cost is subject to the bill. +- Covers token billing for sft / dpo / cpt. Training-unit (MTU) billing is not supported by this command. +- Hyper-parameters other than --n-epochs are fixed at representative defaults for estimation. + +#### Examples + +```bash +bl finetune price --base-model qwen3-8b --datasets file-ft-xxx +``` + +```bash +bl finetune price --base-model qwen3-8b --datasets file-ft-xxx,file-ft-yyy --n-epochs 2 +``` + +```bash +bl finetune price --base-model qwen3-8b --datasets file-ft-xxx --training-type cpt +``` + ### `bl finetune text create` -| Field | Value | -| --------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -| **Name** | `finetune text create` | -| **Description** | Create a text model fine-tune job (sft \| sft-lora \| dpo \| dpo-lora \| cpt) | -| **Usage** | `bl finetune text create --model --datasets [--validations ] [--model-name ] [--suffix ] [--n-epochs ] [--batch-size ] [--learning-rate ] [--max-length ] [--training-type ]` | +| Field | Value | +| --------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | +| **Name** | `finetune text create` | +| **Description** | Create a text model fine-tune job (sft \| sft-lora \| dpo \| dpo-lora \| cpt) | +| **Usage** | `bl finetune text create --base-model --datasets [--validations ] [--model-name ] [--suffix ] [--n-epochs ] [--batch-size ] [--learning-rate ] [--max-length ] [--training-type ]` | #### Flags | Flag | Type | Required | Description | | ---------------------------- | ------ | -------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -| `--model ` | string | yes | Base model to fine-tune | +| `--base-model ` | string | yes | Base model to fine-tune (e.g. qwen3-8b; not the output model name) | | `--datasets ` | string | yes | Comma-separated dataset file IDs or local paths (.jsonl for text, .zip for audio/image). Local paths are uploaded (validated) first, then their file-ids are used. | | `--validations ` | string | no | Comma-separated validation dataset file IDs or local paths (auto-uploaded like --datasets). | | `--model-name ` | string | no | Output model name (after training) | @@ -453,35 +500,35 @@ bl finetune logs --job-id ft-xxx --search checkpoint --tail 5 #### Examples ```bash -bl finetune text create --model qwen3-8b --datasets file-xxx +bl finetune text create --base-model qwen3-8b --datasets file-xxx ``` ```bash -bl finetune text create --model qwen3-8b --datasets ./train.jsonl +bl finetune text create --base-model qwen3-8b --datasets ./train.jsonl ``` ```bash -bl finetune text create --model qwen3-8b --datasets ./train.jsonl --validations ./eval.jsonl +bl finetune text create --base-model qwen3-8b --datasets ./train.jsonl --validations ./eval.jsonl ``` ```bash -bl finetune text create --model qwen3-8b --datasets file-aaa,./extra.jsonl +bl finetune text create --base-model qwen3-8b --datasets file-aaa,./extra.jsonl ``` ```bash -bl finetune text create --model qwen3-8b --datasets ./train.jsonl --training-type sft +bl finetune text create --base-model qwen3-8b --datasets ./train.jsonl --training-type sft ``` ```bash -bl finetune text create --model qwen3-8b --datasets file-xxx --learning-rate "1.6e-5" --n-epochs 4 +bl finetune text create --base-model qwen3-8b --datasets file-xxx --learning-rate "1.6e-5" --n-epochs 4 ``` ```bash -bl finetune text create --model qwen3-8b --datasets file-xxx --output json +bl finetune text create --base-model qwen3-8b --datasets file-xxx --output json ``` ```bash -bl finetune text create --model qwen3-8b --datasets file-xxx --dry-run +bl finetune text create --base-model qwen3-8b --datasets file-xxx --dry-run ``` ### `bl finetune watch` diff --git a/skills/bailian-finetune/reference/index.md b/skills/bailian-finetune/reference/index.md index 07cf63f7..3d163eb3 100644 --- a/skills/bailian-finetune/reference/index.md +++ b/skills/bailian-finetune/reference/index.md @@ -37,16 +37,17 @@ Use this index for the skill-scoped quick index and global flags. | `bl finetune image create` | Create an image generation model fine-tune job (sft-lora) | [finetune.md](finetune.md) | | `bl finetune list` | List fine-tune jobs | [finetune.md](finetune.md) | | `bl finetune logs` | Fetch training logs for a fine-tune job | [finetune.md](finetune.md) | +| `bl finetune price` | Estimate the training cost for a fine-tune job (token billing) | [finetune.md](finetune.md) | | `bl finetune text create` | Create a text model fine-tune job (sft \| sft-lora \| dpo \| dpo-lora \| cpt) | [finetune.md](finetune.md) | | `bl finetune watch` | Probe a fine-tune job's status (default: single non-blocking fetch). Pass --follow to poll until terminal. | [finetune.md](finetune.md) | ## By group -| Group | Commands | Reference | -| ---------- | ---------------------------------------------------------------------------------------------------------------------------------------- | -------------------------- | -| `dataset` | `delete`, `get`, `list`, `upload`, `validate` | [dataset.md](dataset.md) | -| `deploy` | `audio create`, `delete`, `get`, `image create`, `list`, `models`, `pause`, `resume`, `scale`, `text create`, `update` | [deploy.md](deploy.md) | -| `finetune` | `audio create`, `cancel`, `capability`, `checkpoints`, `delete`, `export`, `get`, `image create`, `list`, `logs`, `text create`, `watch` | [finetune.md](finetune.md) | +| Group | Commands | Reference | +| ---------- | ------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------- | +| `dataset` | `delete`, `get`, `list`, `upload`, `validate` | [dataset.md](dataset.md) | +| `deploy` | `audio create`, `delete`, `get`, `image create`, `list`, `models`, `pause`, `resume`, `scale`, `text create`, `update` | [deploy.md](deploy.md) | +| `finetune` | `audio create`, `cancel`, `capability`, `checkpoints`, `delete`, `export`, `get`, `image create`, `list`, `logs`, `price`, `text create`, `watch` | [finetune.md](finetune.md) | ## Global flags From 3c64461cca55b8865a795442d7673239b3c026d1 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Fri, 7 Aug 2026 07:03:01 +0800 Subject: [PATCH 03/14] feat: video finetune/deploy/invoke full pipeline + training cost calculation - Add finetune video create subcommand (wan2.7/2.5/2.2 i2v + kf2v) - Align video hyperparams with official docs (n_epochs=50, per-model batch_size/max_pixels) - Add --last-frame flag to video generate for kf2v (image2video endpoint) - Fix wan2.1-2.6 i2v input format (img_url instead of media[]) - Add training_cost field to finetune get/watch (catalog ft price, API-key domain only) - Add --aigc-* flags to deploy create (optional, for video LoRA prompt config) --- packages/cli/src/commands.ts | 2 + .../commands/src/commands/finetune/create.ts | 55 ++++++++++- .../commands/src/commands/finetune/fee.ts | 11 ++- .../commands/src/commands/video/generate.ts | 61 +++++++++--- packages/commands/src/index.ts | 1 + .../commands/tests/e2e/finetune.e2e.test.ts | 98 +++++++++++++++++++ packages/commands/tests/e2e/topic-routes.ts | 1 + .../tests/e2e/video-generate-i2v.e2e.test.ts | 56 +++++++++++ packages/core/src/client/endpoints.ts | 5 + packages/core/src/client/index.ts | 1 + .../core/src/finetune/profiles/sft-lora.ts | 30 ++++-- packages/core/src/types/api.ts | 2 + skills/bailian-finetune/reference/finetune.md | 55 +++++++++++ skills/bailian-finetune/reference/index.md | 11 ++- skills/bailian-gen/reference/video.md | 1 + 15 files changed, 358 insertions(+), 32 deletions(-) diff --git a/packages/cli/src/commands.ts b/packages/cli/src/commands.ts index ebd8a629..6f61eb39 100644 --- a/packages/cli/src/commands.ts +++ b/packages/cli/src/commands.ts @@ -62,6 +62,7 @@ import { finetuneTextCreate, finetuneAudioCreate, finetuneImageCreate, + finetuneVideoCreate, finetuneList, finetuneGet, finetuneCancel, @@ -183,6 +184,7 @@ export const commands: Record = { "finetune text create": finetuneTextCreate, "finetune audio create": finetuneAudioCreate, "finetune image create": finetuneImageCreate, + "finetune video create": finetuneVideoCreate, "finetune list": finetuneList, "finetune get": finetuneGet, "finetune cancel": finetuneCancel, diff --git a/packages/commands/src/commands/finetune/create.ts b/packages/commands/src/commands/finetune/create.ts index d6496f6b..0f83f980 100644 --- a/packages/commands/src/commands/finetune/create.ts +++ b/packages/commands/src/commands/finetune/create.ts @@ -207,7 +207,7 @@ async function uploadResolvedLocal( } /** The modality a `finetune create` subcommand is bound to. */ -type CommandModality = "text" | "audio" | "image"; +type CommandModality = "text" | "audio" | "image" | "video"; /** * Flags shared by every `finetune create` subcommand: what to train @@ -324,6 +324,34 @@ const AUDIO_USAGE = const IMAGE_USAGE = "--base-model --datasets [--validations ] [--model-name ] [--suffix ] [--generation-type ] [--learning-rate ]"; +/** + * Video (Wan i2v/kf2v) flags: exposes the three hyper-parameters that the + * video API supports and users may want to override. Defaults are model-specific + * (resolved by the sft-lora profile: wan2.7 → batch_size 1 / max_pixels 102400, + * wan2.5 → 4 / 36864, wan2.2 → 4 / 262144). + */ +const VIDEO_FLAGS = { + ...COMMON_FLAGS, + nEpochs: { + type: "number", + valueHint: "", + description: "Training epochs (default: 50)", + }, + batchSize: { + type: "number", + valueHint: "", + description: "Batch size (default: model-specific, 1 for wan2.7, 4 for wan2.5/2.2)", + }, + learningRate: { + type: "string", + valueHint: "", + description: 'Learning rate as a string to preserve precision (default: "2e-5")', + }, +} satisfies FlagsDef; + +const VIDEO_USAGE = + "--base-model --datasets [--validations ] [--model-name ] [--suffix ] [--n-epochs ] [--batch-size ] [--learning-rate ]"; + const COMMON_NOTES = [ "Creating a job uploads any local datasets and consumes training quota.", "Use --dry-run to preview the request body without submitting.", @@ -683,3 +711,28 @@ export const finetuneImageCreate = defineCommand({ notes: IMAGE_NOTES, run: (ctx) => runCreate("image", ctx), }); + +const VIDEO_NOTES = [ + ...COMMON_NOTES, + "Video generation training (Wan i2v/kf2v) runs efficient_sft with model-", + "specific defaults: wan2.7 (batch_size=1, max_pixels=102400), wan2.5/2.2", + "(batch_size=4, max_pixels per model). Override with --batch-size/--n-epochs.", + "Datasets are .zip archives with data.jsonl + frame images + videos.", + "Recommended: ≥10 training samples, 20-100 for stable results.", +]; + +/** `bl finetune video create` — fine-tune a video generation model. Datasets are `.zip`. */ +export const finetuneVideoCreate = defineCommand({ + description: "Create a video generation model fine-tune job (Wan i2v/kf2v, efficient_sft)", + auth: "apiKey", + usageArgs: VIDEO_USAGE, + flags: VIDEO_FLAGS, + exampleArgs: [ + "--base-model wan2.7-i2v --datasets file-xxx", + "--base-model wan2.7-i2v --datasets ./i2v-data.zip", + "--base-model wan2.2-kf2v-flash --datasets file-xxx --n-epochs 100", + "--base-model wan2.7-i2v --datasets file-xxx --dry-run", + ], + notes: VIDEO_NOTES, + run: (ctx) => runCreate("video", ctx), +}); diff --git a/packages/commands/src/commands/finetune/fee.ts b/packages/commands/src/commands/finetune/fee.ts index 7559524a..f1325686 100644 --- a/packages/commands/src/commands/finetune/fee.ts +++ b/packages/commands/src/commands/finetune/fee.ts @@ -77,9 +77,14 @@ async function fetchTrainingPrice( } /** - * Compute the actual training fee from the model catalog. + * Compute the actual training fee from the model catalog's "ft" price entry. * Returns null when the price is unavailable (network error, model not in - * catalog, or no "ft" entry in the prices array). Never throws. + * catalog, or no "ft" entry). Never throws. + * + * Only uses the public model catalog (model metadata) — does NOT call + * console-domain pricing APIs (modelCenter.getModelPrice). Models whose + * catalog entry lacks a "ft" price (e.g. CosyVoice) will simply omit the + * training_cost field until the platform adds it to the catalog. */ export async function computeActualFee( settings: Settings, @@ -91,7 +96,7 @@ export async function computeActualFee( const unitPrice = Number(ftEntry?.price); if (!Number.isFinite(unitPrice) || unitPrice <= 0) return null; const priceUnit = ftEntry?.priceUnit ?? "每百万tokens"; - // price is yuan per million tokens. + // Catalog price is yuan per million tokens. const cost = (usageTokens / 1_000_000) * unitPrice; return { cost: Number(cost.toFixed(4)), unitPrice, priceUnit }; } catch { diff --git a/packages/commands/src/commands/video/generate.ts b/packages/commands/src/commands/video/generate.ts index be3f89af..c58cc4e9 100644 --- a/packages/commands/src/commands/video/generate.ts +++ b/packages/commands/src/commands/video/generate.ts @@ -1,6 +1,7 @@ import { defineCommand, videoGeneratePath, + image2videoPath, taskPath, detectOutputFormat, type DashScopeVideoRequest, @@ -43,6 +44,11 @@ export default defineCommand({ valueHint: "", description: "Input image URL for image-to-video generation", }, + lastFrame: { + type: "string", + valueHint: "", + description: "Last frame image URL (with --image, enables kf2v first+last frame mode)", + }, negativePrompt: { type: "string", valueHint: "", @@ -110,12 +116,20 @@ export default defineCommand({ const format = detectOutputFormat(settings.output); const imageUrl = flags.image; + const lastFrameUrl = flags.lastFrame as string | undefined; // Auto-upload local image file for i2v let resolvedImageUrl: string | undefined; if (imageUrl) { resolvedImageUrl = await ctx.client.resolveImageInput(imageUrl, model); } + let resolvedLastFrameUrl: string | undefined; + if (lastFrameUrl) { + resolvedLastFrameUrl = await ctx.client.resolveImageInput(lastFrameUrl, model); + } + + // kf2v mode: both --image and --last-frame provided. + const isKf2v = Boolean(resolvedImageUrl && resolvedLastFrameUrl); const watermark = resolveWatermark(flags.watermark); const promptExtend = resolveBooleanFlag(flags.promptExtend, undefined, "prompt-extend"); @@ -125,10 +139,16 @@ export default defineCommand({ input: { prompt: prompt, negative_prompt: flags.negativePrompt || undefined, - // i2v models (happyhorse-1.1-i2v) require input.media with type 'first_frame' - ...(resolvedImageUrl - ? { media: [{ type: "first_frame" as const, url: resolvedImageUrl }] } - : {}), + // kf2v: first+last frame flat fields via image2video endpoint. + // wan2.1~2.6 i2v: flat img_url via video-generation endpoint. + // wan2.7+ / happyhorse i2v: media[] via video-generation endpoint. + ...(isKf2v + ? { first_frame_url: resolvedImageUrl, last_frame_url: resolvedLastFrameUrl } + : resolvedImageUrl + ? /wan[x]?2\.[1-6]/i.test(model) + ? { img_url: resolvedImageUrl } + : { media: [{ type: "first_frame" as const, url: resolvedImageUrl }] } + : {}), }, parameters: { resolution: flags.resolution || undefined, @@ -141,15 +161,28 @@ export default defineCommand({ }; if (settings.dryRun) { - const previewBody = resolvedImageUrl - ? { - ...body, - input: { - ...body.input, - media: [{ type: "first_frame" as const, url: redactDataUri(resolvedImageUrl) }], - }, - } - : body; + let previewBody = body; + if (isKf2v) { + previewBody = { + ...body, + input: { + ...body.input, + first_frame_url: redactDataUri(resolvedImageUrl ?? ""), + last_frame_url: redactDataUri(resolvedLastFrameUrl ?? ""), + }, + }; + } else if (resolvedImageUrl) { + const redactedUrl = redactDataUri(resolvedImageUrl); + previewBody = { + ...body, + input: { + ...body.input, + ...(/wan[x]?2\.[1-6]/i.test(model) + ? { img_url: redactedUrl } + : { media: [{ type: "first_frame" as const, url: redactedUrl }] }), + }, + }; + } emitResult({ request: previewBody }, format); return; } @@ -162,7 +195,7 @@ export default defineCommand({ settings, () => ctx.client.requestJson({ - path: videoGeneratePath(), + path: isKf2v ? image2videoPath() : videoGeneratePath(), method: "POST", body, async: true, diff --git a/packages/commands/src/index.ts b/packages/commands/src/index.ts index 848fcb72..ea04d7a3 100644 --- a/packages/commands/src/index.ts +++ b/packages/commands/src/index.ts @@ -66,6 +66,7 @@ export { finetuneTextCreate, finetuneAudioCreate, finetuneImageCreate, + finetuneVideoCreate, } from "./commands/finetune/create.ts"; export { default as finetuneList } from "./commands/finetune/list.ts"; export { default as finetuneGet } from "./commands/finetune/get.ts"; diff --git a/packages/commands/tests/e2e/finetune.e2e.test.ts b/packages/commands/tests/e2e/finetune.e2e.test.ts index a045b2ec..53ee29a6 100644 --- a/packages/commands/tests/e2e/finetune.e2e.test.ts +++ b/packages/commands/tests/e2e/finetune.e2e.test.ts @@ -365,6 +365,104 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => { expect(data.action).toBe("finetune.create"); expect(data.body.training_type).toBe("efficient_sft"); }); + + test("finetune video create --help 暴露视频超参且不含文本超参", async () => { + // Video exposes --n-epochs / --batch-size / --learning-rate; the text-only + // --training-type / --max-length surface is not offered. + const { stderr, exitCode } = await runCommandE2e(FINETUNE_ROUTES, [ + "finetune", + "video", + "create", + "--help", + ]); + expect(exitCode, stderr).toBe(0); + expect(stderr).toMatch(/--base-model/); + expect(stderr).toMatch(/--n-epochs/); + expect(stderr).toMatch(/--batch-size/); + expect(stderr).toMatch(/--learning-rate/); + expect(stderr).not.toMatch(/--training-type|--max-length/); + }); + + test("finetune video create --datasets 缺失时退出为用法错误 (2)", async () => { + const { stderr, exitCode } = await runCommandE2e(FINETUNE_ROUTES, [ + "finetune", + "video", + "create", + "--base-model", + "wan2.7-i2v", + "--quiet", + ]); + expect(exitCode).toBe(2); + expect(stderr).toMatch(/--datasets|Missing required/i); + }); + + test.each([ + // Model-family-specific defaults resolved by the sft-lora video profile. + ["wan2.7-i2v", 1, 102400], + ["wan2.5-i2v-preview", 4, 36864], + ["wan2.2-kf2v-flash", 4, 262144], + ])( + "finetune video create --dry-run %s 解析 batch_size=%i / max_pixels=%i", + async (baseModel, batchSize, maxPixels) => { + const { stdout, stderr, exitCode } = await runCommandE2e(FINETUNE_ROUTES, [ + "finetune", + "video", + "create", + "--base-model", + baseModel, + "--datasets", + "file-video", + "--dry-run", + "--output", + "json", + ]); + expect(exitCode, stderr).toBe(0); + const data = parseStdoutJson<{ + action: string; + body: { + model: string; + training_type: string; + hyper_parameters: Record; + }; + }>(stdout); + expect(data.action).toBe("finetune.create"); + expect(data.body.model).toBe(baseModel); + expect(data.body.training_type).toBe("efficient_sft"); + expect(data.body.hyper_parameters.batch_size).toBe(batchSize); + expect(data.body.hyper_parameters.max_pixels).toBe(maxPixels); + expect(data.body.hyper_parameters.learning_rate).toBe("2e-5"); + expect(data.body.hyper_parameters.lora_rank).toBe(32); + }, + ); + + test("finetune video create --dry-run 转发超参覆盖且不做 clamp", async () => { + const { stdout, stderr, exitCode } = await runCommandE2e(FINETUNE_ROUTES, [ + "finetune", + "video", + "create", + "--base-model", + "wan2.7-i2v", + "--datasets", + "file-video", + "--n-epochs", + "100", + "--batch-size", + "2", + "--learning-rate", + "1e-5", + "--dry-run", + "--output", + "json", + ]); + expect(exitCode, stderr).toBe(0); + const data = parseStdoutJson<{ + body: { hyper_parameters: Record }; + }>(stdout); + // Video overrides are forwarded verbatim (no [8, 1024] text clamp). + expect(data.body.hyper_parameters.n_epochs).toBe(100); + expect(data.body.hyper_parameters.batch_size).toBe(2); + expect(data.body.hyper_parameters.learning_rate).toBe("1e-5"); + }); }); describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (DashScope)", () => { diff --git a/packages/commands/tests/e2e/topic-routes.ts b/packages/commands/tests/e2e/topic-routes.ts index 35497843..860e6bcf 100644 --- a/packages/commands/tests/e2e/topic-routes.ts +++ b/packages/commands/tests/e2e/topic-routes.ts @@ -135,6 +135,7 @@ export const FINETUNE_ROUTES: E2eRouteExports = { "finetune text create": "finetuneTextCreate", "finetune audio create": "finetuneAudioCreate", "finetune image create": "finetuneImageCreate", + "finetune video create": "finetuneVideoCreate", "finetune list": "finetuneList", "finetune get": "finetuneGet", "finetune cancel": "finetuneCancel", diff --git a/packages/commands/tests/e2e/video-generate-i2v.e2e.test.ts b/packages/commands/tests/e2e/video-generate-i2v.e2e.test.ts index 1a26e2aa..aa932fdd 100644 --- a/packages/commands/tests/e2e/video-generate-i2v.e2e.test.ts +++ b/packages/commands/tests/e2e/video-generate-i2v.e2e.test.ts @@ -112,6 +112,62 @@ describe("e2e: video generate (i2v)", () => { }>(stdout); expect(data.request?.input?.media?.[0]?.url).toBe("data:image/png;base64,"); }); + + test.each([ + // wan2.1~2.6 (legacy) use flat img_url; wan2.7+ and happyhorse use media[]. + ["wan2.5-i2v-preview", "img_url"], + ["wan2.6-i2v", "img_url"], + ["wan2.7-i2v", "media"], + ["happyhorse-1.1-i2v", "media"], + ])("video generate --dry-run %s 首帧走 %s 字段", async (model, field) => { + const configDir = makeE2eOutputDir(`video-i2v-input-shape-${model}`); + writeFileSync( + join(configDir, "config.json"), + JSON.stringify({ + "token-plan": { + api_key: "sk-sp-e2e-placeholder", + base_url: "https://token-plan.cn-beijing.maas.aliyuncs.com", + }, + }), + ); + + const { stdout, stderr, exitCode } = await runCommandE2e( + VIDEO_ROUTES, + [ + "video", + "generate", + "--config", + "token-plan", + "--dry-run", + "--model", + model, + "--image", + "https://example.com/placeholder.png", + "--prompt", + "干跑校验", + "--output", + "json", + ], + { + BAILIAN_CONFIG_DIR: configDir, + DASHSCOPE_API_KEY: "", + DASHSCOPE_BASE_URL: "", + }, + ); + expect(exitCode, stderr).toBe(0); + const data = parseStdoutJson<{ + request?: { + input?: { img_url?: string; media?: Array<{ type?: string; url?: string }> }; + }; + }>(stdout); + if (field === "img_url") { + expect(data.request?.input?.img_url).toBe("https://example.com/placeholder.png"); + expect(data.request?.input?.media).toBeUndefined(); + } else { + expect(data.request?.input?.media?.[0]?.type).toBe("first_frame"); + expect(data.request?.input?.img_url).toBeUndefined(); + } + }); }); describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())( diff --git a/packages/core/src/client/endpoints.ts b/packages/core/src/client/endpoints.ts index 119c7bda..93706e60 100644 --- a/packages/core/src/client/endpoints.ts +++ b/packages/core/src/client/endpoints.ts @@ -32,6 +32,11 @@ export function videoGeneratePath(): string { return "/api/v1/services/aigc/video-generation/video-synthesis"; } +/** POST /api/v1/services/aigc/image2video/video-synthesis — kf2v (first+last frame). */ +export function image2videoPath(): string { + return "/api/v1/services/aigc/image2video/video-synthesis"; +} + // ---- Async Task Query ---- export function taskPath(taskId: string): string { return `/api/v1/tasks/${encodeURIComponent(taskId)}`; diff --git a/packages/core/src/client/index.ts b/packages/core/src/client/index.ts index a10e28d8..3632f2ec 100644 --- a/packages/core/src/client/index.ts +++ b/packages/core/src/client/index.ts @@ -19,6 +19,7 @@ export { taskPath, userProfilePath, videoGeneratePath, + image2videoPath, } from "./endpoints.ts"; export { isLegacyImage2ImageModel, diff --git a/packages/core/src/finetune/profiles/sft-lora.ts b/packages/core/src/finetune/profiles/sft-lora.ts index d442dc5e..538543bf 100644 --- a/packages/core/src/finetune/profiles/sft-lora.ts +++ b/packages/core/src/finetune/profiles/sft-lora.ts @@ -75,20 +75,24 @@ const IMAGE_HYPER_PARAMS_I2I: Record = { * * Shared across all video models; `batch_size` and `max_pixels` differ by model * family (resolved per model in `resolveHyperParameters`): - * - wan2.5 (e.g. wan2.5-i2v-preview): batch_size 2, max_pixels 36864 - * - wan2.2 (i2v-flash / kf2v-flash): batch_size 4, max_pixels 262144 + * - wan2.7 (e.g. wan2.7-i2v): batch_size 1, max_pixels 102400 + * - wan2.5 (e.g. wan2.5-i2v-preview): batch_size 4, max_pixels 36864 + * - wan2.2 (i2v-flash / kf2v-flash): batch_size 4, max_pixels 262144 * * `learning_rate` is a string to avoid JSON-number precision loss (consistent * with the image defaults). `split` (0.9) + `max_split_val_dataset_sample` * (5) drive the automatic train/validation split when no explicit * validation_file_ids are provided. + * + * Values aligned with the official user guide (2026-07): + * n_epochs 50, eval_epochs 20 (≥ n_epochs/10). */ const VIDEO_HYPER_PARAMS_BASE: Record = { - n_epochs: 400, + n_epochs: 50, learning_rate: "2e-5", - split: 0.9, + split: 0.5, max_split_val_dataset_sample: 5, - eval_epochs: 50, + eval_epochs: 20, save_total_limit: 10, lora_rank: 32, lora_alpha: 32, @@ -109,6 +113,11 @@ function isWan25(model: string | undefined): boolean { return typeof model === "string" && /wan2\.5/i.test(model); } +/** wan2.7 family uses batch_size 1 and max_pixels 102400. */ +function isWan27(model: string | undefined): boolean { + return typeof model === "string" && /wan2\.7/i.test(model); +} + export const sftLoraProfile: TrainingProfile = { clientTrainingType: "sft-lora", serverTrainingType: "efficient_sft", @@ -195,15 +204,18 @@ export const sftLoraProfile: TrainingProfile = { } if (isVideo(modality)) { // Video: shared defaults + model-family-specific batch_size / max_pixels. - // wan2.5 uses batch_size 2 / max_pixels 36864; wan2.2 uses 4 / 262144. - const wan25 = isWan25(flags.model as string | undefined); + // wan2.7: batch_size 1, max_pixels 102400 + // wan2.5: batch_size 4, max_pixels 36864 + // wan2.2: batch_size 4, max_pixels 262144 + const model = (flags.model ?? flags.baseModel) as string | undefined; const hp: Record = { ...VIDEO_HYPER_PARAMS_BASE, - batch_size: 4, - max_pixels: wan25 ? 36864 : 262144, + batch_size: isWan27(model) ? 1 : 4, + max_pixels: isWan27(model) ? 102400 : isWan25(model) ? 36864 : 262144, }; // Optional overrides (no clamping — video batch_size is intentionally small). if (flags.nEpochs !== undefined) hp.n_epochs = flags.nEpochs as number; + if (flags.batchSize !== undefined) hp.batch_size = flags.batchSize as number; if (flags.learningRate !== undefined) hp.learning_rate = flags.learningRate as string; return hp; } diff --git a/packages/core/src/types/api.ts b/packages/core/src/types/api.ts index c6c00ec0..a50f2050 100644 --- a/packages/core/src/types/api.ts +++ b/packages/core/src/types/api.ts @@ -168,6 +168,8 @@ export interface DashScopeVideoRequest { prompt: string; negative_prompt?: string; img_url?: string; + first_frame_url?: string; + last_frame_url?: string; media?: Array<{ type: "image" | "video" | "first_frame" | "last_frame" | "driving_audio" | "first_clip"; url: string; diff --git a/skills/bailian-finetune/reference/finetune.md b/skills/bailian-finetune/reference/finetune.md index 178adb27..d243de31 100644 --- a/skills/bailian-finetune/reference/finetune.md +++ b/skills/bailian-finetune/reference/finetune.md @@ -21,6 +21,7 @@ Index: [index.md](index.md) | `bl finetune logs` | Fetch training logs for a fine-tune job | | `bl finetune price` | Estimate the training cost for a fine-tune job (token billing) | | `bl finetune text create` | Create a text model fine-tune job (sft \| sft-lora \| dpo \| dpo-lora \| cpt) | +| `bl finetune video create` | Create a video generation model fine-tune job (Wan i2v/kf2v, efficient_sft) | | `bl finetune watch` | Probe a fine-tune job's status (default: single non-blocking fetch). Pass --follow to poll until terminal. | ## Command details @@ -531,6 +532,60 @@ bl finetune text create --base-model qwen3-8b --datasets file-xxx --output json bl finetune text create --base-model qwen3-8b --datasets file-xxx --dry-run ``` +### `bl finetune video create` + +| Field | Value | +| --------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | +| **Name** | `finetune video create` | +| **Description** | Create a video generation model fine-tune job (Wan i2v/kf2v, efficient_sft) | +| **Usage** | `bl finetune video create --base-model --datasets [--validations ] [--model-name ] [--suffix ] [--n-epochs ] [--batch-size ] [--learning-rate ]` | + +#### Flags + +| Flag | Type | Required | Description | +| ---------------------------- | ------ | -------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------ | +| `--base-model ` | string | yes | Base model to fine-tune (e.g. qwen3-8b; not the output model name) | +| `--datasets ` | string | yes | Comma-separated dataset file IDs or local paths (.jsonl for text, .zip for audio/image). Local paths are uploaded (validated) first, then their file-ids are used. | +| `--validations ` | string | no | Comma-separated validation dataset file IDs or local paths (auto-uploaded like --datasets). | +| `--model-name ` | string | no | Output model name (after training) | +| `--suffix ` | string | no | Output suffix appended by the platform (finetuned_output_suffix) | +| `--n-epochs ` | number | no | Training epochs (default: 50) | +| `--batch-size ` | number | no | Batch size (default: model-specific, 1 for wan2.7, 4 for wan2.5/2.2) | +| `--learning-rate ` | string | no | Learning rate as a string to preserve precision (default: "2e-5") | +| `--api-key ` | string | no | API key | +| `--base-url ` | string | no | API base URL | + +#### Notes + +- Creating a job uploads any local datasets and consumes training quota. +- Use --dry-run to preview the request body without submitting. +- --datasets / --validations accept either file-ids (from `dataset upload`) +- or local paths. Local paths are validated and uploaded first, then their +- file-ids are submitted — a one-step upload-and-train. +- Video generation training (Wan i2v/kf2v) runs efficient_sft with model- +- specific defaults: wan2.7 (batch_size=1, max_pixels=102400), wan2.5/2.2 +- (batch_size=4, max_pixels per model). Override with --batch-size/--n-epochs. +- Datasets are .zip archives with data.jsonl + frame images + videos. +- Recommended: ≥10 training samples, 20-100 for stable results. + +#### Examples + +```bash +bl finetune video create --base-model wan2.7-i2v --datasets file-xxx +``` + +```bash +bl finetune video create --base-model wan2.7-i2v --datasets ./i2v-data.zip +``` + +```bash +bl finetune video create --base-model wan2.2-kf2v-flash --datasets file-xxx --n-epochs 100 +``` + +```bash +bl finetune video create --base-model wan2.7-i2v --datasets file-xxx --dry-run +``` + ### `bl finetune watch` | Field | Value | diff --git a/skills/bailian-finetune/reference/index.md b/skills/bailian-finetune/reference/index.md index 3d163eb3..126621bf 100644 --- a/skills/bailian-finetune/reference/index.md +++ b/skills/bailian-finetune/reference/index.md @@ -39,15 +39,16 @@ Use this index for the skill-scoped quick index and global flags. | `bl finetune logs` | Fetch training logs for a fine-tune job | [finetune.md](finetune.md) | | `bl finetune price` | Estimate the training cost for a fine-tune job (token billing) | [finetune.md](finetune.md) | | `bl finetune text create` | Create a text model fine-tune job (sft \| sft-lora \| dpo \| dpo-lora \| cpt) | [finetune.md](finetune.md) | +| `bl finetune video create` | Create a video generation model fine-tune job (Wan i2v/kf2v, efficient_sft) | [finetune.md](finetune.md) | | `bl finetune watch` | Probe a fine-tune job's status (default: single non-blocking fetch). Pass --follow to poll until terminal. | [finetune.md](finetune.md) | ## By group -| Group | Commands | Reference | -| ---------- | ------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------- | -| `dataset` | `delete`, `get`, `list`, `upload`, `validate` | [dataset.md](dataset.md) | -| `deploy` | `audio create`, `delete`, `get`, `image create`, `list`, `models`, `pause`, `resume`, `scale`, `text create`, `update` | [deploy.md](deploy.md) | -| `finetune` | `audio create`, `cancel`, `capability`, `checkpoints`, `delete`, `export`, `get`, `image create`, `list`, `logs`, `price`, `text create`, `watch` | [finetune.md](finetune.md) | +| Group | Commands | Reference | +| ---------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------- | +| `dataset` | `delete`, `get`, `list`, `upload`, `validate` | [dataset.md](dataset.md) | +| `deploy` | `audio create`, `delete`, `get`, `image create`, `list`, `models`, `pause`, `resume`, `scale`, `text create`, `update` | [deploy.md](deploy.md) | +| `finetune` | `audio create`, `cancel`, `capability`, `checkpoints`, `delete`, `export`, `get`, `image create`, `list`, `logs`, `price`, `text create`, `video create`, `watch` | [finetune.md](finetune.md) | ## Global flags diff --git a/skills/bailian-gen/reference/video.md b/skills/bailian-gen/reference/video.md index 62327050..ee8656a3 100644 --- a/skills/bailian-gen/reference/video.md +++ b/skills/bailian-gen/reference/video.md @@ -108,6 +108,7 @@ bl video edit --video https://example.com/input.mp4 --prompt "Put clothes on the | `--model ` | string | no | Model ID (default: happyhorse-1.1-t2v, or happyhorse-1.1-i2v with --image) | | `--prompt ` | string | yes | Video description | | `--image ` | string | no | Input image URL for image-to-video generation | +| `--last-frame ` | string | no | Last frame image URL (with --image, enables kf2v first+last frame mode) | | `--negative-prompt ` | string | no | Negative prompt to exclude unwanted content | | `--resolution ` | string | no | Resolution: 720P or 1080P (default: 1080P) | | `--ratio ` | string | no | Aspect ratio (e.g. 16:9, 9:16, 1:1) | From e25d3b0b8e413143784bb786d871a83bb1fbd9a8 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Fri, 7 Aug 2026 14:13:47 +0800 Subject: [PATCH 04/14] ci: add workflow to publish skills to OSS --- .github/workflows/publish-skills.yml | 27 +++++++++++++++++++++++++++ 1 file changed, 27 insertions(+) create mode 100644 .github/workflows/publish-skills.yml diff --git a/.github/workflows/publish-skills.yml b/.github/workflows/publish-skills.yml new file mode 100644 index 00000000..775b7c3d --- /dev/null +++ b/.github/workflows/publish-skills.yml @@ -0,0 +1,27 @@ +# Poke the FC publish-skills flow after skills/ changes land. +# The FC side reconciles this repo's skills/ directory against OSS +# (bailian-wiki/skills/) using the repo HEAD snapshot as the only +# source of truth — the request itself carries no content. Both the +# repo and branch params are validated against FC-side whitelists +# (PUBLISH_REPOS / PUBLISH_BRANCHES). +# +# feat/cli-skill-sync is temporary for end-to-end testing; remove it +# (here and from the FC PUBLISH_BRANCHES whitelist) once the sync +# link is verified on main. +name: Publish skills to OSS + +on: + push: + branches: + - main + - feat/cli-skill-sync + paths: + - "skills/**" + +jobs: + poke: + runs-on: ubuntu-latest + steps: + - name: Trigger FC publish-skills + run: | + curl -sf -X POST "${{ secrets.FC_PUBLISH_TRIGGER_URL }}/publish-skills?repo=modelstudioai/cli&branch=${{ github.ref_name }}" From ebd6226a9f917d83c495d13497a3c362654dc74c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Fri, 7 Aug 2026 14:25:49 +0800 Subject: [PATCH 05/14] ci: rename trigger secret to FC_TRIGGER_URL --- .github/workflows/publish-skills.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/workflows/publish-skills.yml b/.github/workflows/publish-skills.yml index 775b7c3d..f53314c3 100644 --- a/.github/workflows/publish-skills.yml +++ b/.github/workflows/publish-skills.yml @@ -24,4 +24,4 @@ jobs: steps: - name: Trigger FC publish-skills run: | - curl -sf -X POST "${{ secrets.FC_PUBLISH_TRIGGER_URL }}/publish-skills?repo=modelstudioai/cli&branch=${{ github.ref_name }}" + curl -sf -X POST "${{ secrets.FC_TRIGGER_URL }}/publish-skills?repo=modelstudioai/cli&branch=${{ github.ref_name }}" From 1d35326c86e39b1a7a41312142e6b9fa5aee66b2 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Fri, 7 Aug 2026 14:36:42 +0800 Subject: [PATCH 06/14] ci: read FC trigger url from variables --- .github/workflows/publish-skills.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/workflows/publish-skills.yml b/.github/workflows/publish-skills.yml index f53314c3..6064c40f 100644 --- a/.github/workflows/publish-skills.yml +++ b/.github/workflows/publish-skills.yml @@ -24,4 +24,4 @@ jobs: steps: - name: Trigger FC publish-skills run: | - curl -sf -X POST "${{ secrets.FC_TRIGGER_URL }}/publish-skills?repo=modelstudioai/cli&branch=${{ github.ref_name }}" + curl -sf -X POST "${{ vars.FC_TRIGGER_URL }}/publish-skills?repo=modelstudioai/cli&branch=${{ github.ref_name }}" From 9eb2acbb65026661b27aa5a2c932072329fd4227 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Fri, 7 Aug 2026 14:38:40 +0800 Subject: [PATCH 07/14] test: trigger skills sync pipeline --- skills/bailian-protocol/README.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/skills/bailian-protocol/README.md b/skills/bailian-protocol/README.md index 224430da..4c02c751 100644 --- a/skills/bailian-protocol/README.md +++ b/skills/bailian-protocol/README.md @@ -15,3 +15,5 @@ For CLI installation and command examples, see the [main README](../../README.md ## License Apache-2.0 + + From 8286a74fb69cf98b7e4ca0cd330086cde9e0f5a2 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Fri, 7 Aug 2026 14:40:33 +0800 Subject: [PATCH 08/14] test: remove sync pipeline verification marker --- skills/bailian-protocol/README.md | 2 -- 1 file changed, 2 deletions(-) diff --git a/skills/bailian-protocol/README.md b/skills/bailian-protocol/README.md index 4c02c751..224430da 100644 --- a/skills/bailian-protocol/README.md +++ b/skills/bailian-protocol/README.md @@ -15,5 +15,3 @@ For CLI installation and command examples, see the [main README](../../README.md ## License Apache-2.0 - - From 5007b9b574cf74dab033e0bf882b2a981581412a Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Fri, 7 Aug 2026 16:35:47 +0800 Subject: [PATCH 09/14] feat: change output to json --- packages/commands/src/commands/skill/add.ts | 3 +-- packages/commands/src/commands/skill/list.ts | 3 +-- packages/commands/src/commands/skill/remove.ts | 3 +-- packages/commands/src/commands/skill/update.ts | 3 +-- 4 files changed, 4 insertions(+), 8 deletions(-) diff --git a/packages/commands/src/commands/skill/add.ts b/packages/commands/src/commands/skill/add.ts index 073a80a9..3182dab3 100644 --- a/packages/commands/src/commands/skill/add.ts +++ b/packages/commands/src/commands/skill/add.ts @@ -2,7 +2,6 @@ import { BailianError, ExitCode, defineCommand, - detectOutputFormat, detectInstalledAgents, fetchSkillsIndex, getSkillRegistryBaseUrl, @@ -39,7 +38,7 @@ export default defineCommand({ }, exampleArgs: ["--name all", "--name spark-video,bailian-model-recommend"], async run(ctx) { - const format = detectOutputFormat(ctx.settings.output); + const format = ctx.settings.outputExplicit ? ctx.settings.output : "json"; const requested = parseSkillNames(ctx.flags.name, false); const index = await fetchSkillsIndex(); const remoteNames = Object.keys(index.skills); diff --git a/packages/commands/src/commands/skill/list.ts b/packages/commands/src/commands/skill/list.ts index aa6552cc..1a21887d 100644 --- a/packages/commands/src/commands/skill/list.ts +++ b/packages/commands/src/commands/skill/list.ts @@ -1,6 +1,5 @@ import { defineCommand, - detectOutputFormat, computeSkillStatuses, fetchSkillsIndex, getSkillRegistryBaseUrl, @@ -24,7 +23,7 @@ export default defineCommand({ "STATUS: installed | outdated | not-installed | missing (lock has it, dir deleted) | untracked (dir exists, not managed)", ], async run(ctx) { - const format = detectOutputFormat(ctx.settings.output); + const format = ctx.settings.outputExplicit ? ctx.settings.output : "json"; // Three-way reconciliation: live remote index × skill-lock.json (installation facts) × disk const index = await fetchSkillsIndex(); const lock = readSkillLock(); diff --git a/packages/commands/src/commands/skill/remove.ts b/packages/commands/src/commands/skill/remove.ts index e51106c2..50749629 100644 --- a/packages/commands/src/commands/skill/remove.ts +++ b/packages/commands/src/commands/skill/remove.ts @@ -2,7 +2,6 @@ import { BailianError, ExitCode, defineCommand, - detectOutputFormat, listSkillDirsOnDisk, parseSkillNames, readSkillLock, @@ -34,7 +33,7 @@ export default defineCommand({ exampleArgs: ["--name spark-video", "--name all"], async run(ctx) { // Purely local operation: no remote access, works offline - const format = detectOutputFormat(ctx.settings.output); + const format = ctx.settings.outputExplicit ? ctx.settings.output : "json"; const requested = parseSkillNames(ctx.flags.name, false); const lock = readSkillLock(); const names = requested === "all" ? Object.keys(lock.skills) : requested; diff --git a/packages/commands/src/commands/skill/update.ts b/packages/commands/src/commands/skill/update.ts index 61e1b1f6..b41ac012 100644 --- a/packages/commands/src/commands/skill/update.ts +++ b/packages/commands/src/commands/skill/update.ts @@ -2,7 +2,6 @@ import { BailianError, ExitCode, defineCommand, - detectOutputFormat, detectInstalledAgents, fanOutSkillToAgents, fetchSkillsIndex, @@ -40,7 +39,7 @@ export default defineCommand({ }, exampleArgs: ["", "--name spark-video"], async run(ctx) { - const format = detectOutputFormat(ctx.settings.output); + const format = ctx.settings.outputExplicit ? ctx.settings.output : "json"; const requested = parseSkillNames(ctx.flags.name, true); const index = await fetchSkillsIndex(); const lock = readSkillLock(); From 1c76749ee59e6f3acf3944e639d2f1a31da2bc8c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Tue, 11 Aug 2026 13:33:31 +0800 Subject: [PATCH 10/14] feat: update datalist validate --- .../commands/src/commands/dataset/upload.ts | 15 ++-- .../commands/src/commands/dataset/validate.ts | 28 ++++---- skills/bailian-finetune/reference/dataset.md | 68 ++++++++++--------- 3 files changed, 53 insertions(+), 58 deletions(-) diff --git a/packages/commands/src/commands/dataset/upload.ts b/packages/commands/src/commands/dataset/upload.ts index 13635095..6888b115 100644 --- a/packages/commands/src/commands/dataset/upload.ts +++ b/packages/commands/src/commands/dataset/upload.ts @@ -28,7 +28,7 @@ const UPLOAD_FLAGS = { type: "string", valueHint: "", description: - 'Record schema: "chatml" (SFT), "dpo" (chosen/rejected), "cpt" (raw text), "tts" (audio), or "image" (image generation). Default auto-detects per record.', + 'Record schema: "chatml" (SFT), "dpo" (chosen/rejected), "cpt" (raw text), "tts" (audio), "image" (image generation), or "video" (video generation). Default auto-detects per record.', }, noValidate: { type: "switch", @@ -44,7 +44,7 @@ export default defineCommand({ description: "Upload a dataset file (.jsonl or .zip) to Bailian", auth: "apiKey", usageArgs: - "--file [--purpose ] [--schema ] [--no-validate] [--full-validate]", + "--file [--purpose ] [--schema ] [--no-validate] [--full-validate]", flags: UPLOAD_FLAGS, exampleArgs: [ "--file train.jsonl", @@ -72,15 +72,8 @@ export default defineCommand({ const filePath = flags.file; const purpose = flags.purpose || "fine-tune"; const schema = parseDatasetSchemaFlag(flags.schema); - if (schema === "video") { - throw new BailianError( - `--schema video is not supported.`, - ExitCode.USAGE, - `Supported schemas: chatml, dpo, cpt, tts, image.`, - ); - } - // Image schema allows larger ZIPs (1 GB vs 300 MB for text). - const isMediaSchema = schema === "image"; + // Image and video schemas allow larger ZIPs (1 GB vs 300 MB for text). + const isMediaSchema = schema === "image" || schema === "video"; if (!flags.noValidate) { const maxBytes = isMediaSchema ? MAX_MEDIA_ZIP_BYTES : MAX_DATASET_BYTES; diff --git a/packages/commands/src/commands/dataset/validate.ts b/packages/commands/src/commands/dataset/validate.ts index 142fbdf5..b9e7f62b 100644 --- a/packages/commands/src/commands/dataset/validate.ts +++ b/packages/commands/src/commands/dataset/validate.ts @@ -23,7 +23,7 @@ const VALIDATE_FLAGS = { type: "string", valueHint: "", description: - 'Record schema: "chatml" (SFT), "dpo" (chosen/rejected), "cpt" (raw text), "tts" (audio), or "image" (image generation). Default auto-detects per record.', + 'Record schema: "chatml" (SFT), "dpo" (chosen/rejected), "cpt" (raw text), "tts" (audio), "image" (image generation), or "video" (video generation). Default auto-detects per record.', }, } satisfies FlagsDef; @@ -31,13 +31,14 @@ export default defineCommand({ description: "Locally validate a dataset file (.jsonl or .zip) without uploading", // 纯本地校验，不触网、不需 API key（与 `pipeline validate` 一致）。 auth: "none", - usageArgs: "--file [--full-validate] [--schema ]", + usageArgs: "--file [--full-validate] [--schema ]", flags: VALIDATE_FLAGS, exampleArgs: [ "--file train.jsonl", "--file dpo.jsonl --schema dpo", "--file cpt.jsonl --schema cpt", "--file audio.zip --schema tts", + "--file wan-i2v-training-dataset.zip --schema video", "--file eval.jsonl --full-validate", "--file train.jsonl --output json", ], @@ -47,25 +48,20 @@ export default defineCommand({ "Schemas: chatml = {messages:[...]} (SFT); dpo = {messages:[...], chosen,", 'rejected}; cpt = {text:"..."} (continual pre-training, raw text);', 'tts = {wav_fn:"train/xxx.wav", text:"..."} (audio fine-tuning);', - 'image = {img_path:"..."} (image generation). With no --schema, a record', - "carrying wav_fn is validated as TTS, img_path as image, chosen/rejected", - "as DPO, text (no messages) as CPT, otherwise ChatML. Pass --schema to", - "require a specific shape on every record. ZIP archives (.zip) are", - "validated structurally (data.jsonl present, media references resolve) in", - "addition to per-record content checks. Use --full-validate to JSON.parse", - "every line.", + 'image = {img_path:"..."} (image generation);', + 'video = {first_frame_path:"...", video_path:"..."} (video generation,', + "i2v first-frame or kf2v first+last-frame with last_frame_path). With no", + "--schema, a record carrying wav_fn is validated as TTS, img_path as image,", + "first_frame_path/video_path as video, chosen/rejected as DPO, text (no", + "messages) as CPT, otherwise ChatML. Pass --schema to require a specific", + "shape on every record. ZIP archives (.zip) are validated structurally", + "(data.jsonl present, media references resolve) in addition to per-record", + "content checks. Use --full-validate to JSON.parse every line.", ], async run(ctx) { const { settings, flags } = ctx; const filePath = flags.file; const schema = parseDatasetSchemaFlag(flags.schema); - if (schema === "video") { - throw new BailianError( - `--schema video is not supported.`, - ExitCode.USAGE, - `Supported schemas: chatml, dpo, cpt, tts, image.`, - ); - } if (settings.dryRun) { emitResult( { diff --git a/skills/bailian-finetune/reference/dataset.md b/skills/bailian-finetune/reference/dataset.md index 523a421a..22cad6db 100644 --- a/skills/bailian-finetune/reference/dataset.md +++ b/skills/bailian-finetune/reference/dataset.md @@ -107,23 +107,23 @@ bl dataset list --output json ### `bl dataset upload` -| Field | Value | -| --------------- | -------------------------------------------------------------------------------------------------------------------------------- | -| **Name** | `dataset upload` | -| **Description** | Upload a dataset file (.jsonl or .zip) to Bailian | -| **Usage** | `bl dataset upload --file [--purpose ] [--schema ] [--no-validate] [--full-validate]` | +| Field | Value | +| --------------- | --------------------------------------------------------------------------------------------------------------------------------------- | +| **Name** | `dataset upload` | +| **Description** | Upload a dataset file (.jsonl or .zip) to Bailian | +| **Usage** | `bl dataset upload --file [--purpose ] [--schema ] [--no-validate] [--full-validate]` | #### Flags -| Flag | Type | Required | Description | -| ------------------ | ------ | -------- | -------------------------------------------------------------------------------------------------------------------------------------------------------- | -| `--file ` | string | yes | Local dataset file (.jsonl or .zip; ≤300MB text, ≤1GB image) | -| `--purpose ` | string | no | Dataset purpose tag (default: "fine-tune"; e.g. "evaluation") | -| `--schema ` | string | no | Record schema: "chatml" (SFT), "dpo" (chosen/rejected), "cpt" (raw text), "tts" (audio), or "image" (image generation). Default auto-detects per record. | -| `--no-validate` | switch | no | Skip the local JSONL pre-flight check (not recommended) | -| `--full-validate` | switch | no | JSON.parse every line instead of sampling (slower) | -| `--api-key ` | string | no | API key | -| `--base-url ` | string | no | API base URL | +| Flag | Type | Required | Description | +| ------------------ | ------ | -------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | +| `--file ` | string | yes | Local dataset file (.jsonl or .zip; ≤300MB text, ≤1GB image) | +| `--purpose ` | string | no | Dataset purpose tag (default: "fine-tune"; e.g. "evaluation") | +| `--schema ` | string | no | Record schema: "chatml" (SFT), "dpo" (chosen/rejected), "cpt" (raw text), "tts" (audio), "image" (image generation), or "video" (video generation). Default auto-detects per record. | +| `--no-validate` | switch | no | Skip the local JSONL pre-flight check (not recommended) | +| `--full-validate` | switch | no | JSON.parse every line instead of sampling (slower) | +| `--api-key ` | string | no | API key | +| `--base-url ` | string | no | API base URL | #### Notes @@ -170,19 +170,19 @@ bl dataset upload --file train.jsonl --no-validate ### `bl dataset validate` -| Field | Value | -| --------------- | ----------------------------------------------------------------------------------------------- | -| **Name** | `dataset validate` | -| **Description** | Locally validate a dataset file (.jsonl or .zip) without uploading | -| **Usage** | `bl dataset validate --file [--full-validate] [--schema ]` | +| Field | Value | +| --------------- | ------------------------------------------------------------------------------------------------------ | +| **Name** | `dataset validate` | +| **Description** | Locally validate a dataset file (.jsonl or .zip) without uploading | +| **Usage** | `bl dataset validate --file [--full-validate] [--schema ]` | #### Flags -| Flag | Type | Required | Description | -| ----------------- | ------ | -------- | -------------------------------------------------------------------------------------------------------------------------------------------------------- | -| `--file ` | string | yes | Local dataset file (.jsonl or .zip) | -| `--full-validate` | switch | no | JSON.parse every line instead of sampling (slower) | -| `--schema ` | string | no | Record schema: "chatml" (SFT), "dpo" (chosen/rejected), "cpt" (raw text), "tts" (audio), or "image" (image generation). Default auto-detects per record. | +| Flag | Type | Required | Description | +| ----------------- | ------ | -------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | +| `--file ` | string | yes | Local dataset file (.jsonl or .zip) | +| `--full-validate` | switch | no | JSON.parse every line instead of sampling (slower) | +| `--schema ` | string | no | Record schema: "chatml" (SFT), "dpo" (chosen/rejected), "cpt" (raw text), "tts" (audio), "image" (image generation), or "video" (video generation). Default auto-detects per record. | #### Notes @@ -191,13 +191,15 @@ bl dataset upload --file train.jsonl --no-validate - Schemas: chatml = {messages:[...]} (SFT); dpo = {messages:[...], chosen, - rejected}; cpt = {text:"..."} (continual pre-training, raw text); - tts = {wav_fn:"train/xxx.wav", text:"..."} (audio fine-tuning); -- image = {img_path:"..."} (image generation). With no --schema, a record -- carrying wav_fn is validated as TTS, img_path as image, chosen/rejected -- as DPO, text (no messages) as CPT, otherwise ChatML. Pass --schema to -- require a specific shape on every record. ZIP archives (.zip) are -- validated structurally (data.jsonl present, media references resolve) in -- addition to per-record content checks. Use --full-validate to JSON.parse -- every line. +- image = {img_path:"..."} (image generation); +- video = {first_frame_path:"...", video_path:"..."} (video generation, +- i2v first-frame or kf2v first+last-frame with last_frame_path). With no +- --schema, a record carrying wav_fn is validated as TTS, img_path as image, +- first_frame_path/video_path as video, chosen/rejected as DPO, text (no +- messages) as CPT, otherwise ChatML. Pass --schema to require a specific +- shape on every record. ZIP archives (.zip) are validated structurally +- (data.jsonl present, media references resolve) in addition to per-record +- content checks. Use --full-validate to JSON.parse every line. #### Examples @@ -217,6 +219,10 @@ bl dataset validate --file cpt.jsonl --schema cpt bl dataset validate --file audio.zip --schema tts ``` +```bash +bl dataset validate --file wan-i2v-training-dataset.zip --schema video +``` + ```bash bl dataset validate --file eval.jsonl --full-validate ``` From 2965080cb7f33b621832d7b032916b5bcac9fb2a Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Tue, 11 Aug 2026 14:29:10 +0800 Subject: [PATCH 11/14] feat(skills): replace foreign skill dirs containing SKILL.md during fan-out Previously, fan-out skipped any existing real directory not recorded in the lock file, treating it as user content. This left stale skill copies installed by other tools (e.g. npx skills add) permanently out of date. Now: if the directory contains a SKILL.md, it is recognized as a skill artifact and replaced with a symlink to the canonical dir. Directories without SKILL.md are still preserved (user content safety boundary). --- packages/core/src/skills/agents.ts | 20 ++++++++++ packages/core/tests/skills-agents.test.ts | 45 +++++++++++++++++++---- 2 files changed, 58 insertions(+), 7 deletions(-) diff --git a/packages/core/src/skills/agents.ts b/packages/core/src/skills/agents.ts index 1ddd369e..16b2e8d8 100644 --- a/packages/core/src/skills/agents.ts +++ b/packages/core/src/skills/agents.ts @@ -285,6 +285,21 @@ function isRecordedCopy(linkPath: string, recordedLinks: string[]): boolean { } } +/** + * Whether linkPath is a real directory containing a SKILL.md — indicating it is a + * skill artifact installed by another tool (e.g. `npx skills add`) or an older + * version predating bl's symlink management. These are safe to replace: they are + * not arbitrary user content but the same kind of artifact we manage. + */ +function isForeignSkillDir(linkPath: string): boolean { + try { + if (!lstatSync(linkPath).isDirectory()) return false; + return existsSync(join(linkPath, "SKILL.md")); + } catch { + return false; + } +} + export interface LinkResult { agent: string; path: string; @@ -326,6 +341,11 @@ export function linkSkillToAgents( // Copy-fallback artifact from a previous install → replace so updates // reach agents that have no symlink permission rmSync(linkPath, { recursive: true, force: true }); + } else if (isForeignSkillDir(linkPath)) { + // A real directory containing SKILL.md — a skill installed by another + // tool (e.g. `npx skills add`) or predating bl's symlink management. + // Replace with our symlink so future updates propagate automatically. + rmSync(linkPath, { recursive: true, force: true }); } else { results.push({ agent: agent.id, diff --git a/packages/core/tests/skills-agents.test.ts b/packages/core/tests/skills-agents.test.ts index dc80ca31..7fa8125b 100644 --- a/packages/core/tests/skills-agents.test.ts +++ b/packages/core/tests/skills-agents.test.ts @@ -293,20 +293,51 @@ test("agents: Amp-style XDG config dir lights up the universal-xdg shared target }); }); -test("agents: recorded copy-fallback artifact is replaced; unrecorded dir stays skipped", async () => { +test("agents: foreign skill dir (contains SKILL.md) is replaced even without lock record", async () => { await inFakeHome(async (home) => { mkdirSync(join(home, ".claude"), { recursive: true }); const canonical = seedCanonicalSkill("demo"); const copyPath = join(home, ".claude", "skills", "demo"); - // Simulate a previous install that fell back to copy (no symlink permission, e.g. Windows) + // Simulate a skill installed by another tool (e.g. `npx skills add`) — a real + // directory containing SKILL.md, not recorded in our lock. mkdirSync(copyPath, { recursive: true }); - writeFileSync(join(copyPath, "SKILL.md"), "stale copy"); + writeFileSync(join(copyPath, "SKILL.md"), "stale copy from another tool"); + + // New behavior: contains SKILL.md → recognized as a skill artifact → replaced + const result = linkSkillToAgents("demo"); + expect(result[0]).toMatchObject({ agent: "claude-code", mode: "symlink" }); + expect(lstatSync(copyPath).isSymbolicLink()).toBe(true); + expect(readlinkSync(copyPath)).toBe(canonical); + }); +}); + +test("agents: foreign non-skill dir (no SKILL.md) stays skipped", async () => { + await inFakeHome(async (home) => { + mkdirSync(join(home, ".claude"), { recursive: true }); + seedCanonicalSkill("demo"); + const foreignPath = join(home, ".claude", "skills", "demo"); - // Without a lock record the dir is foreign → skipped, content untouched - const unrecorded = linkSkillToAgents("demo"); - expect(unrecorded[0].mode).toBe("skipped"); - expect(readFileSync(join(copyPath, "SKILL.md"), "utf-8")).toBe("stale copy"); + // A user's own directory that happens to share the skill name but has no SKILL.md + mkdirSync(foreignPath, { recursive: true }); + writeFileSync(join(foreignPath, "my-notes.txt"), "user content"); + + const result = linkSkillToAgents("demo"); + expect(result[0].mode).toBe("skipped"); + // User content untouched + expect(readFileSync(join(foreignPath, "my-notes.txt"), "utf-8")).toBe("user content"); + }); +}); + +test("agents: recorded copy-fallback artifact is replaced with symlink", async () => { + await inFakeHome(async (home) => { + mkdirSync(join(home, ".claude"), { recursive: true }); + const canonical = seedCanonicalSkill("demo"); + const copyPath = join(home, ".claude", "skills", "demo"); + + // Simulate a previous install that fell back to copy (no symlink permission) + mkdirSync(copyPath, { recursive: true }); + writeFileSync(join(copyPath, "SKILL.md"), "stale copy"); // With the recorded link the artifact is rebuilt and points at canonical again const recorded = linkSkillToAgents("demo", detectInstalledAgents(), [copyPath]); From 9749a11d76943d1f35d8cb17f8e12494676b5c6f Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Tue, 11 Aug 2026 14:50:58 +0800 Subject: [PATCH 12/14] fix(finetune): detect video-kf2v sub-variant from local dataset MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit finetune video create passed a fixed modality "video" to the profile validator without probing the data for last_frame_path. This caused a false KF2V_DATA_MISMATCH error when training kf2v models (wan2.2-kf2v-*) with datasets that correctly contain last_frame_path. Add sub-variant detection symmetric to the existing image-i2i upgrade: when a local file is provided, detectModality() inspects the first record and upgrades "video" → "video-kf2v" if last_frame_path is present. --- packages/commands/src/commands/finetune/create.ts | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/packages/commands/src/commands/finetune/create.ts b/packages/commands/src/commands/finetune/create.ts index 0f83f980..834168c0 100644 --- a/packages/commands/src/commands/finetune/create.ts +++ b/packages/commands/src/commands/finetune/create.ts @@ -468,6 +468,10 @@ async function runCreate( if (detected === "image-i2i") modality = "image-i2i"; } } + if (commandModality === "video" && firstLocalPath && !settings.dryRun) { + const detected = await detectModality(firstLocalPath); + if (detected === "video-kf2v") modality = "video-kf2v"; + } const training = await analyzeDatasetTokens( settings, From e7422bd2e5e3357968e0517d77ba568cbb5dcb2f Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Wed, 12 Aug 2026 07:48:11 +0800 Subject: [PATCH 13/14] fix(dataset): align validation rules with platform data format spec MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - Support content array format [{text/image/video}] alongside legacy string - Add tool role support with tool_calls structure and tool_call_id validation - Add thinking tag placement check (only in last assistant message) - Add OpenAI migration guards: warn on unsupported name/weight fields - Add loss_weight range validation (0.0–1.0) - Fix size limits: SFT/DPO 200MB, CPT 300MB, media ZIP 2GB - Enforce data.jsonl at ZIP root (reject nested wrapping folders) - Add ZIP filename constraints: charset [a-zA-Z0-9_-], length ≤120, uniqueness - Add .tif to accepted image extensions - Add DPO_LAST_MSG_NOT_USER warning when messages don't end with user role - CPT profile now uses dedicated 300MB cap instead of shared default - Expand unit tests from 19 to 45 covering all new validation paths --- .../commands/src/commands/dataset/upload.ts | 22 +- packages/core/src/dataset/index.ts | 1 + packages/core/src/dataset/validate/common.ts | 22 +- packages/core/src/dataset/validate/index.ts | 7 +- .../src/dataset/validate/schemas/chatml.ts | 398 +++++++++++++++++- .../core/src/dataset/validate/schemas/dpo.ts | 16 + .../src/dataset/validate/schemas/image.ts | 10 +- packages/core/src/dataset/validate/zip.ts | 184 ++++++-- packages/core/src/finetune/profiles/cpt.ts | 36 +- packages/core/tests/dataset-validate.test.ts | 280 +++++++++++- skills/bailian-finetune/reference/dataset.md | 11 +- 11 files changed, 918 insertions(+), 69 deletions(-) diff --git a/packages/commands/src/commands/dataset/upload.ts b/packages/commands/src/commands/dataset/upload.ts index 6888b115..822c5a9f 100644 --- a/packages/commands/src/commands/dataset/upload.ts +++ b/packages/commands/src/commands/dataset/upload.ts @@ -5,6 +5,7 @@ import { parseDatasetSchemaFlag, formatIssue, MAX_DATASET_BYTES, + MAX_CPT_BYTES, MAX_MEDIA_ZIP_BYTES, BailianError, ExitCode, @@ -16,7 +17,7 @@ const UPLOAD_FLAGS = { file: { type: "string", valueHint: "", - description: "Local dataset file (.jsonl or .zip; ≤300MB text, ≤1GB image)", + description: "Local dataset file (.jsonl or .zip; ≤200MB SFT/DPO, ≤300MB CPT, ≤2GB media zip)", required: true, }, purpose: { @@ -57,13 +58,14 @@ export default defineCommand({ ], notes: [ "Supports .jsonl (text) and .zip (audio/image archives with a data.jsonl", - "manifest). Five record schemas are recognized: chatml = {messages:[...]}", + "manifest). Six record schemas are recognized: chatml = {messages:[...]}", '(SFT); dpo = {messages:[...], chosen, rejected}; cpt = {text:"..."}', '(continual pre-training, raw text); tts = {wav_fn:"train/xxx.wav",', 'text:"..."} (audio fine-tuning); image = {img_path:"..."} (image', - "generation). With no --schema, a record carrying wav_fn is validated as", - "TTS, img_path as image, chosen/rejected as DPO, text (no messages) as CPT,", - "otherwise ChatML. Upload cap: 300MB text, 1GB image. Upload uses the", + "generation); video = {first_frame_path:...} (video generation). With no", + "--schema, a record carrying wav_fn is validated as TTS, img_path as image,", + "chosen/rejected as DPO, text (no messages) as CPT, otherwise ChatML.", + "Upload cap: 200MB SFT/DPO text, 300MB CPT, 2GB media zip. Upload uses the", "OpenAI-compatible /compatible-mode/v1/files endpoint so the purpose tag is", "persisted (the DashScope-native /api/v1/files drops it).", ], @@ -72,11 +74,15 @@ export default defineCommand({ const filePath = flags.file; const purpose = flags.purpose || "fine-tune"; const schema = parseDatasetSchemaFlag(flags.schema); - // Image and video schemas allow larger ZIPs (1 GB vs 300 MB for text). + // Size caps differ per training type: SFT/DPO 200MB, CPT 300MB, media ZIP 2GB. const isMediaSchema = schema === "image" || schema === "video"; + const maxBytes = isMediaSchema + ? MAX_MEDIA_ZIP_BYTES + : schema === "cpt" + ? MAX_CPT_BYTES + : MAX_DATASET_BYTES; if (!flags.noValidate) { - const maxBytes = isMediaSchema ? MAX_MEDIA_ZIP_BYTES : MAX_DATASET_BYTES; const result = await validateDataset(filePath, { fullValidate: flags.fullValidate, schema, @@ -116,7 +122,7 @@ export default defineCommand({ action: "dataset.upload", file: filePath, purpose, - max_bytes: isMediaSchema ? MAX_MEDIA_ZIP_BYTES : MAX_DATASET_BYTES, + max_bytes: maxBytes, validate: !flags.noValidate, schema: schema ?? "auto", }, diff --git a/packages/core/src/dataset/index.ts b/packages/core/src/dataset/index.ts index cc605c44..b6e581c2 100644 --- a/packages/core/src/dataset/index.ts +++ b/packages/core/src/dataset/index.ts @@ -7,6 +7,7 @@ export { registerValidator, listSupportedFormats, MAX_DATASET_BYTES, + MAX_CPT_BYTES, MAX_MEDIA_ZIP_BYTES, parseDatasetSchemaFlag, formatIssue, diff --git a/packages/core/src/dataset/validate/common.ts b/packages/core/src/dataset/validate/common.ts index 17bdbad5..212fe7ba 100644 --- a/packages/core/src/dataset/validate/common.ts +++ b/packages/core/src/dataset/validate/common.ts @@ -12,18 +12,24 @@ import { ExitCode } from "../../errors/codes.ts"; import type { DatasetSchema, ValidationIssue, ValidationStats } from "./types.ts"; /** - * The platform caps dataset uploads at 300MB per file. `bl dataset upload` - * enforces this client-side so users learn early. Update if the platform - * raises the cap or differentiates per-purpose limits. + * The platform caps SFT/DPO text dataset uploads at 200MB per file. + * `bl dataset upload` enforces this client-side so users learn early. + * CPT uses 300MB (see MAX_CPT_BYTES); API general upload is also 300MB. */ -export const MAX_DATASET_BYTES = 300 * 1024 * 1024; +export const MAX_DATASET_BYTES = 200 * 1024 * 1024; /** - * Image / video ZIP size cap — 1 GB per the platform docs (vs 300 MB for - * text / audio). Used by `bl dataset upload` for media schemas and by the - * `sft-lora` training profile for image / video validation. + * CPT text dataset size cap — 300 MB per the platform docs. + * CPT requires at least 50M tokens; larger files are expected. */ -export const MAX_MEDIA_ZIP_BYTES = 1024 * 1024 * 1024; +export const MAX_CPT_BYTES = 300 * 1024 * 1024; + +/** + * Image / video ZIP size cap — 2 GB per the platform docs. Used by + * `bl dataset upload` for media schemas and by the `sft-lora` training + * profile for image / video validation. + */ +export const MAX_MEDIA_ZIP_BYTES = 2 * 1024 * 1024 * 1024; export interface PreflightResult { bytes: number; diff --git a/packages/core/src/dataset/validate/index.ts b/packages/core/src/dataset/validate/index.ts index 42df4421..169a32c7 100644 --- a/packages/core/src/dataset/validate/index.ts +++ b/packages/core/src/dataset/validate/index.ts @@ -4,7 +4,12 @@ export { registerValidator, listSupportedFormats, } from "./registry.ts"; -export { MAX_DATASET_BYTES, MAX_MEDIA_ZIP_BYTES, parseDatasetSchemaFlag } from "./common.ts"; +export { + MAX_DATASET_BYTES, + MAX_CPT_BYTES, + MAX_MEDIA_ZIP_BYTES, + parseDatasetSchemaFlag, +} from "./common.ts"; export { formatIssue } from "./format.ts"; export type { ValidatorSpec, diff --git a/packages/core/src/dataset/validate/schemas/chatml.ts b/packages/core/src/dataset/validate/schemas/chatml.ts index d4c032e6..959149e8 100644 --- a/packages/core/src/dataset/validate/schemas/chatml.ts +++ b/packages/core/src/dataset/validate/schemas/chatml.ts @@ -5,12 +5,215 @@ * and no more specific schema matches, ChatML is selected. `inspectMessageObject` * lives here because it is the canonical per-message check; the DPO schema * imports it to validate `chosen` / `rejected` preference messages. + * + * Content format: supports both legacy plain-string content (`"content": "…"`) + * and the current platform array format (`"content": [{"text": "…"}, …]`). + * The array format may also carry `image` / `video` items for VL multimodal + * understanding data. + * + * Tool calling: supports `role: "tool"` messages with `tool_call_id`, and + * `assistant.tool_calls` arrays. Validates id correspondence. */ import { makeIssue } from "../common.ts"; import type { ValidationIssue } from "../types.ts"; import type { RecordSchemaSpec } from "./types.ts"; -const VALID_ROLES = new Set(["system", "user", "assistant"]); +const VALID_ROLES = new Set(["system", "user", "assistant", "tool"]); + +/** + * Validate a content field that may be: + * - A plain string (legacy format) + * - An array of content items: `[{text: "…"}, {image: "…"}, {video: "…"|"…"}, …]` + * + * Returns issues found. `path` scopes the location for error reporting. + */ +export function inspectContentField( + content: unknown, + lineNo: number, + path: string, +): ValidationIssue[] { + const out: ValidationIssue[] = []; + if (typeof content === "string") { + // Legacy string format — always valid. + return out; + } + if (!Array.isArray(content)) { + out.push( + makeIssue( + "error", + "INVALID_CONTENT", + `"content" must be a string or an array of content items (got ${typeof content}).`, + { line: lineNo, path }, + ), + ); + return out; + } + if (content.length === 0) { + out.push( + makeIssue("error", "EMPTY_CONTENT_ARRAY", `"content" array must not be empty.`, { + line: lineNo, + path, + }), + ); + return out; + } + for (let idx = 0; idx < content.length; idx++) { + const item = content[idx]; + const itemPath = `${path}[${idx}]`; + if (item === null || typeof item !== "object" || Array.isArray(item)) { + out.push( + makeIssue("error", "INVALID_CONTENT_ITEM", `Content item must be an object.`, { + line: lineNo, + path: itemPath, + }), + ); + continue; + } + const obj = item as Record; + const hasText = "text" in obj; + const hasImage = "image" in obj; + const hasVideo = "video" in obj; + if (!hasText && !hasImage && !hasVideo) { + out.push( + makeIssue( + "error", + "CONTENT_ITEM_NO_KNOWN_FIELD", + `Content item must contain at least one of: "text", "image", "video".`, + { line: lineNo, path: itemPath }, + ), + ); + continue; + } + if (hasText && typeof obj.text !== "string") { + out.push( + makeIssue("error", "INVALID_CONTENT_TEXT", `"text" in content item must be a string.`, { + line: lineNo, + path: `${itemPath}.text`, + }), + ); + } + if (hasImage && typeof obj.image !== "string") { + out.push( + makeIssue("error", "INVALID_CONTENT_IMAGE", `"image" in content item must be a string.`, { + line: lineNo, + path: `${itemPath}.image`, + }), + ); + } + if (hasVideo) { + // video can be a string (file path) or an array of strings (frame list) + const video = obj.video; + if (typeof video !== "string" && !Array.isArray(video)) { + out.push( + makeIssue( + "error", + "INVALID_CONTENT_VIDEO", + `"video" in content item must be a string (file path) or an array of strings (frame list).`, + { line: lineNo, path: `${itemPath}.video` }, + ), + ); + } else if (Array.isArray(video)) { + for (let frameIdx = 0; frameIdx < video.length; frameIdx++) { + if (typeof video[frameIdx] !== "string") { + out.push( + makeIssue( + "error", + "INVALID_VIDEO_FRAME", + `Video frame list item at index ${frameIdx} must be a string.`, + { line: lineNo, path: `${itemPath}.video[${frameIdx}]` }, + ), + ); + } + } + } + } + } + return out; +} + +/** + * Validate `tool_calls` array on an assistant message. + * Each entry: `{id: string, type: "function", function: {name: string, arguments: string}}`. + */ +function inspectToolCalls(toolCalls: unknown, lineNo: number, path: string): ValidationIssue[] { + const out: ValidationIssue[] = []; + if (!Array.isArray(toolCalls)) { + out.push( + makeIssue("error", "INVALID_TOOL_CALLS", `"tool_calls" must be an array.`, { + line: lineNo, + path, + }), + ); + return out; + } + for (let idx = 0; idx < toolCalls.length; idx++) { + const call = toolCalls[idx]; + const callPath = `${path}[${idx}]`; + if (call === null || typeof call !== "object" || Array.isArray(call)) { + out.push( + makeIssue("error", "INVALID_TOOL_CALL", `tool_calls item must be an object.`, { + line: lineNo, + path: callPath, + }), + ); + continue; + } + const obj = call as Record; + if (typeof obj.id !== "string" || obj.id.length === 0) { + out.push( + makeIssue("error", "TOOL_CALL_MISSING_ID", `tool_calls item must have a non-empty "id".`, { + line: lineNo, + path: `${callPath}.id`, + }), + ); + } + if (obj.type !== "function") { + out.push( + makeIssue( + "warning", + "TOOL_CALL_TYPE_NOT_FUNCTION", + `tool_calls item "type" should be "function" (got "${String(obj.type)}").`, + { line: lineNo, path: `${callPath}.type` }, + ), + ); + } + const fn = obj.function; + if (fn === null || typeof fn !== "object" || Array.isArray(fn)) { + out.push( + makeIssue( + "error", + "TOOL_CALL_MISSING_FUNCTION", + `tool_calls item must have a "function" object.`, + { + line: lineNo, + path: `${callPath}.function`, + }, + ), + ); + } else { + const fnObj = fn as Record; + if (typeof fnObj.name !== "string" || fnObj.name.length === 0) { + out.push( + makeIssue("error", "TOOL_CALL_FN_NO_NAME", `tool_calls function must have a "name".`, { + line: lineNo, + path: `${callPath}.function.name`, + }), + ); + } + if (typeof fnObj.arguments !== "string") { + out.push( + makeIssue( + "error", + "TOOL_CALL_FN_ARGS_NOT_STRING", + `tool_calls function "arguments" must be a JSON string.`, + { line: lineNo, path: `${callPath}.function.arguments` }, + ), + ); + } + } + } + return out; +} /** * Structural checks for a single message object `{role, content}`. Shared by @@ -35,25 +238,73 @@ export function inspectMessageObject( } const record = msg as Record; const role = record.role; - const content = record.content; if (typeof role !== "string" || !VALID_ROLES.has(role)) { out.push( makeIssue( "error", "INVALID_ROLE", - `Invalid role "${String(role)}". Expected one of: system, user, assistant.`, + `Invalid role "${String(role)}". Expected one of: system, user, assistant, tool.`, { line: lineNo, path: `${path}.role` }, ), ); } - if (typeof content !== "string") { + + // tool role: must have tool_call_id + if (role === "tool") { + if (typeof record.tool_call_id !== "string" || record.tool_call_id.length === 0) { + out.push( + makeIssue( + "error", + "TOOL_MISSING_CALL_ID", + `A "tool" role message must have a non-empty "tool_call_id".`, + { line: lineNo, path: `${path}.tool_call_id` }, + ), + ); + } + } + + // content validation: string or array format + if (!("content" in record)) { + // assistant messages with tool_calls may omit content + if (role !== "assistant" || !("tool_calls" in record)) { + out.push( + makeIssue("error", "MISSING_CONTENT", `"content" field is missing.`, { + line: lineNo, + path: `${path}.content`, + }), + ); + } + } else { + out.push(...inspectContentField(record.content, lineNo, `${path}.content`)); + } + + // tool_calls on assistant + if ("tool_calls" in record) { + out.push(...inspectToolCalls(record.tool_calls, lineNo, `${path}.tool_calls`)); + } + + // OpenAI migration guard: name / weight are not supported by Bailian + if ("name" in record) { out.push( - makeIssue("error", "INVALID_CONTENT", `"content" must be a string (got ${typeof content}).`, { - line: lineNo, - path: `${path}.content`, - }), + makeIssue( + "warning", + "UNSUPPORTED_FIELD_NAME", + `Field "name" is not supported by Bailian. Remove it when migrating from OpenAI/Azure.`, + { line: lineNo, path: `${path}.name` }, + ), + ); + } + if ("weight" in record) { + out.push( + makeIssue( + "warning", + "UNSUPPORTED_FIELD_WEIGHT", + `Field "weight" is not supported by Bailian. Remove it when migrating from OpenAI/Azure.`, + { line: lineNo, path: `${path}.weight` }, + ), ); } + return out; } @@ -91,19 +342,24 @@ export function inspectChatMLRecord( let sawSystem = false; let lastRole: string | undefined; - for (let i = 0; i < messages.length; i++) { - const msg = messages[i]; - const path = `messages[${i}]`; + let lastAssistantIdx = -1; + const toolCallIds = new Set(); + const toolResponseIds = new Set(); + + for (let idx = 0; idx < messages.length; idx++) { + const msg = messages[idx]; + const path = `messages[${idx}]`; out.push(...inspectMessageObject(msg, lineNo, path)); - const role = (msg as Record | null)?.role; + const msgObj = msg as Record | null; + const role = msgObj?.role; if (role === "system") { - if (i !== 0) { + if (idx !== 0) { out.push( makeIssue( "warning", "SYSTEM_NOT_FIRST", - `"system" message should appear at index 0; found at index ${i}.`, + `"system" message should appear at index 0; found at index ${idx}.`, { line: lineNo, path: `${path}.role` }, ), ); @@ -111,6 +367,27 @@ export function inspectChatMLRecord( sawSystem = true; } + if (role === "assistant") { + lastAssistantIdx = idx; + // Collect tool_calls ids + if (msgObj && Array.isArray(msgObj.tool_calls)) { + for (const call of msgObj.tool_calls) { + const callObj = call as Record | null; + if (callObj && typeof callObj.id === "string") { + toolCallIds.add(callObj.id); + } + } + } + } + + if (role === "tool") { + const callId = msgObj?.tool_call_id; + if (typeof callId === "string" && callId.length > 0) { + toolResponseIds.add(callId); + } + } + + // Consecutive same-role warning (skip tool — multiple tool responses are normal) if (lastRole === role && (role === "user" || role === "assistant")) { out.push( makeIssue( @@ -123,6 +400,7 @@ export function inspectChatMLRecord( } if (typeof role === "string") lastRole = role; } + // Soft check: messages without any user role almost certainly indicate a bug. if (!messages.some((m) => (m as Record).role === "user")) { out.push( @@ -140,9 +418,101 @@ export function inspectChatMLRecord( }), ); } + + // tool_call_id correspondence: every tool response should reference a known call id + for (const responseId of toolResponseIds) { + if (!toolCallIds.has(responseId)) { + out.push( + makeIssue( + "warning", + "TOOL_CALL_ID_UNMATCHED", + `tool message references tool_call_id "${responseId}" which does not match any assistant tool_calls[].id.`, + { line: lineNo, path: "messages" }, + ), + ); + } + } + + // thinking tag check: …` should only appear in the last assistant message + if (lastAssistantIdx >= 0) { + for (let idx = 0; idx < messages.length; idx++) { + if (idx === lastAssistantIdx) continue; + const msg = messages[idx] as Record | null; + if (msg?.role !== "assistant") continue; + const content = msg?.content; + if (contentHasThinkTag(content)) { + out.push( + makeIssue( + "warning", + "THINK_TAG_NOT_LAST", + `Thinking tags (…) should only appear in the last assistant message, found at messages[${idx}].`, + { line: lineNo, path: `messages[${idx}].content` }, + ), + ); + } + } + } + + // loss_weight validation (record-level, invite-only parameter) + if ("loss_weight" in record) { + const lossWeight = record.loss_weight; + if (typeof lossWeight !== "number" || lossWeight < 0 || lossWeight > 1) { + out.push( + makeIssue( + "error", + "INVALID_LOSS_WEIGHT", + `"loss_weight" must be a number between 0.0 and 1.0 (got ${JSON.stringify(lossWeight)}).`, + { line: lineNo, path: "loss_weight" }, + ), + ); + } + } + + // OpenAI migration guard at record level + if ("name" in record && !("messages" in record)) { + // Only warn at record level if it's not inside messages (messages handled above) + out.push( + makeIssue( + "warning", + "UNSUPPORTED_FIELD_NAME", + `Record-level field "name" is not supported by Bailian.`, + { line: lineNo, path: "name" }, + ), + ); + } + if ("weight" in record) { + out.push( + makeIssue( + "warning", + "UNSUPPORTED_FIELD_WEIGHT", + `Record-level field "weight" is not supported by Bailian. Remove it when migrating from OpenAI/Azure.`, + { line: lineNo, path: "weight" }, + ), + ); + } + return out; } +/** Check whether content (string or array) contains a tag. */ +function contentHasThinkTag(content: unknown): boolean { + if (typeof content === "string") { + return content.includes(""); + } + if (Array.isArray(content)) { + return content.some((item) => { + if (item && typeof item === "object" && "text" in item) { + return ( + typeof (item as Record).text === "string" && + ((item as Record).text as string).includes("") + ); + } + return false; + }); + } + return false; +} + /** * ChatML / SFT schema. The auto-detect predicate is `true` so it acts as the * registry fallback — any record that isn't picked up by a more specific diff --git a/packages/core/src/dataset/validate/schemas/dpo.ts b/packages/core/src/dataset/validate/schemas/dpo.ts index 5dc69370..26c3bcc3 100644 --- a/packages/core/src/dataset/validate/schemas/dpo.ts +++ b/packages/core/src/dataset/validate/schemas/dpo.ts @@ -18,6 +18,22 @@ function inspectDPORecord(record: Record, lineNo: number): Vali const messages = record.messages; if (!Array.isArray(messages) || messages.length === 0) return out; + // DPO convention: messages should end with a user message (the prompt that + // chosen/rejected respond to). If the last message is assistant, it's likely + // a structural mistake. + const lastMsg = messages[messages.length - 1] as Record | null; + if (lastMsg && lastMsg.role !== "user") { + out.push( + makeIssue( + "warning", + "DPO_LAST_MSG_NOT_USER", + `DPO "messages" should end with a "user" message (the prompt for chosen/rejected). ` + + `Got "${String(lastMsg.role)}" as the last message.`, + { line: lineNo, path: `messages[${messages.length - 1}].role` }, + ), + ); + } + const hasChosen = "chosen" in record; const hasRejected = "rejected" in record; diff --git a/packages/core/src/dataset/validate/schemas/image.ts b/packages/core/src/dataset/validate/schemas/image.ts index daef3804..248619c7 100644 --- a/packages/core/src/dataset/validate/schemas/image.ts +++ b/packages/core/src/dataset/validate/schemas/image.ts @@ -16,7 +16,15 @@ import type { ValidationIssue } from "../types.ts"; import type { RecordSchemaSpec } from "./types.ts"; /** Accepted image file extensions (lower-case, with dot). */ -export const IMAGE_EXTENSIONS = new Set([".png", ".jpg", ".jpeg", ".bmp", ".webp", ".tiff"]); +export const IMAGE_EXTENSIONS = new Set([ + ".png", + ".jpg", + ".jpeg", + ".bmp", + ".tif", + ".tiff", + ".webp", +]); /** * Check that a path string ends with an accepted image extension. diff --git a/packages/core/src/dataset/validate/zip.ts b/packages/core/src/dataset/validate/zip.ts index 43091337..da1c3622 100644 --- a/packages/core/src/dataset/validate/zip.ts +++ b/packages/core/src/dataset/validate/zip.ts @@ -3,14 +3,15 @@ * * A training data ZIP must have: * - `data.jsonl` at the root — the manifest mapping media files to labels. - * - A `train/` subfolder (or media files at the root) referenced by the - * manifest entries. + * The platform requires data.jsonl to be directly visible when opening the + * ZIP (no wrapping folder). + * - Media files referenced by the manifest entries. * * This validator owns the **ZIP-level structural checks** (entries present, - * references resolve). The **per-record JSONL content validation** is delegated - * to the existing `jsonlValidator` — we extract `data.jsonl` to a temp file, - * run the full pipeline (quickScan + deepCheck + schema dispatch), and stitch - * the results together. + * references resolve, filename constraints). The **per-record JSONL content + * validation** is delegated to the existing `jsonlValidator` — we extract + * `data.jsonl` to a temp file, run the full pipeline (quickScan + deepCheck + + * schema dispatch), and stitch the results together. * * The schema for `data.jsonl` records is passed via `opts.schema` (typically * `"tts"` for audio). The profile layer decides which schema to use based on @@ -92,6 +93,121 @@ function collectZipEntries(zipPath: string): Promise { }); } +/** + * Platform filename constraints: + * - Allowed charset: ASCII letters (a-z, A-Z), digits (0-9), underscore (_), hyphen (-) + * - Filename (without extension) ≤ 120 characters + * - Filenames must be globally unique (ignoring extension) + */ +const FILENAME_CHARSET_RE = /^[a-zA-Z0-9_-]+$/; +const MAX_FILENAME_BASE_LENGTH = 120; + +/** Extract the base name (no extension) from a path segment. */ +function basenameNoExt(segment: string): string { + const dot = segment.lastIndexOf("."); + return dot > 0 ? segment.slice(0, dot) : segment; +} + +/** + * Validate ZIP entry filenames against platform constraints. + * Returns issues for charset violations, over-length names, and duplicates. + */ +function validateZipFilenames(entries: string[]): ValidationIssue[] { + const out: ValidationIssue[] = []; + const seenBasenames = new Map(); // basename (no ext) → first full path + const MAX_REPORTED = 10; + + for (const entry of entries) { + // Skip directory entries + if (entry.endsWith("/")) continue; + + // Check each path segment (folder names + file name) + const segments = entry.split("/").filter((s) => s.length > 0); + for (const segment of segments) { + // Strip extension for the charset check on the base part + const base = basenameNoExt(segment); + const ext = segment.slice(base.length); // includes dot, e.g. ".jpg" + + // Charset check on base name (extension checked separately) + if (base.length > 0 && !FILENAME_CHARSET_RE.test(base)) { + if (out.length < MAX_REPORTED) { + out.push( + makeIssue( + "error", + "INVALID_FILENAME_CHARSET", + `File/folder name "${segment}" contains invalid characters. ` + + `Only a-z, A-Z, 0-9, underscore (_), and hyphen (-) are allowed.`, + { path: entry }, + ), + ); + } + } + + // Extension charset (allow dot + alphanumeric) + if (ext.length > 0 && !/^\.[a-zA-Z0-9]+$/.test(ext)) { + if (out.length < MAX_REPORTED) { + out.push( + makeIssue( + "error", + "INVALID_FILENAME_CHARSET", + `File extension "${ext}" in "${segment}" contains invalid characters.`, + { path: entry }, + ), + ); + } + } + } + + // Filename length check (base name without extension) + const fileName = segments[segments.length - 1] ?? ""; + const baseName = basenameNoExt(fileName); + if (baseName.length > MAX_FILENAME_BASE_LENGTH) { + if (out.length < MAX_REPORTED) { + out.push( + makeIssue( + "error", + "FILENAME_TOO_LONG", + `Filename "${fileName}" (without extension) exceeds ${MAX_FILENAME_BASE_LENGTH} characters ` + + `(got ${baseName.length}). Shorten the name and re-upload.`, + { path: entry }, + ), + ); + } + } + + // Global uniqueness check (ignoring extension, case-sensitive) + if (baseName.length > 0) { + const existing = seenBasenames.get(baseName); + if (existing !== undefined) { + if (out.length < MAX_REPORTED) { + out.push( + makeIssue( + "error", + "DUPLICATE_FILENAME", + `Filename "${fileName}" conflicts with "${existing}" — names must be globally unique ` + + `(ignoring extension) even across different folders.`, + { path: entry }, + ), + ); + } + } else { + seenBasenames.set(baseName, entry); + } + } + } + + if (out.length >= MAX_REPORTED) { + out.push( + makeIssue( + "warning", + "FILENAME_ISSUES_TRUNCATED", + `More filename issues exist but reporting is capped at ${MAX_REPORTED}.`, + ), + ); + } + return out; +} + /** * Extract a single entry from a ZIP archive to a destination path. */ @@ -201,19 +317,37 @@ export const zipValidator: ValidatorSpec = { }; } - // --- 2. Check for data.jsonl --- - const hasDataJsonl = entries.some( - (entry) => entry === "data.jsonl" || entry.endsWith("/data.jsonl"), - ); - if (!hasDataJsonl) { - errors.push( - makeIssue( - "error", - "MISSING_DATA_JSONL", - `ZIP archive must contain "data.jsonl" at the root. ` + - `This file maps media files (e.g. .wav) to their labels.`, - ), - ); + // --- 2. Check for data.jsonl (must be at ZIP root) --- + const hasRootDataJsonl = entries.some((entry) => entry === "data.jsonl"); + const nestedDataJsonl = + !hasRootDataJsonl && entries.find((entry) => entry.endsWith("/data.jsonl")); + if (!hasRootDataJsonl) { + if (nestedDataJsonl) { + errors.push( + makeIssue( + "error", + "DATA_JSONL_NOT_AT_ROOT", + `"data.jsonl" must be at the ZIP root (found "${nestedDataJsonl}"). ` + + `Re-package so that opening the ZIP shows data.jsonl directly, without a wrapping folder.`, + ), + ); + } else { + errors.push( + makeIssue( + "error", + "MISSING_DATA_JSONL", + `ZIP archive must contain "data.jsonl" at the root. ` + + `This file maps media files (e.g. .wav, .jpg) to their labels.`, + ), + ); + } + } + + // --- 2b. Filename constraints (charset, length, uniqueness) --- + const filenameIssues = validateZipFilenames(entries); + for (const issue of filenameIssues) { + if (issue.severity === "error") errors.push(issue); + else warnings.push(issue); } // --- 3. Check for train/ directory (modality-aware) --- @@ -265,8 +399,8 @@ export const zipValidator: ValidatorSpec = { } } - // If data.jsonl is missing, we can't do JSONL content validation. - if (!hasDataJsonl) { + // If data.jsonl is missing entirely, we can't do JSONL content validation. + if (!hasRootDataJsonl && !nestedDataJsonl) { return { valid: false, format: "zip", @@ -278,9 +412,11 @@ export const zipValidator: ValidatorSpec = { } // --- 4. Extract data.jsonl to a temp file and run jsonlValidator --- - const dataJsonlEntry = entries.find( - (entry) => entry === "data.jsonl" || entry.endsWith("/data.jsonl"), - )!; + // Prefer root data.jsonl; fall back to nested for content validation even + // though we already reported the root-placement error above. + const dataJsonlEntry = hasRootDataJsonl + ? "data.jsonl" + : entries.find((entry) => entry.endsWith("/data.jsonl"))!; const tmpDir = join(tmpdir(), `bl-zip-${randomBytes(6).toString("hex")}`); mkdirSync(tmpDir, { recursive: true }); const tmpJsonl = join(tmpDir, "data.jsonl"); diff --git a/packages/core/src/finetune/profiles/cpt.ts b/packages/core/src/finetune/profiles/cpt.ts index ab5aba7f..f9629445 100644 --- a/packages/core/src/finetune/profiles/cpt.ts +++ b/packages/core/src/finetune/profiles/cpt.ts @@ -1,7 +1,39 @@ /** * `cpt` profile — Continual Pre-Training (full-parameter). * Maps to the server's `cpt` training type. CPT record schema. + * CPT allows larger files (300 MB) compared to SFT/DPO (200 MB). */ -import { textProfile } from "./common.ts"; +import type { TrainingProfile, DataModality } from "./types.ts"; +import type { ValidateOpts, ValidationResult } from "../../dataset/validate/types.ts"; +import { validateDataset } from "../../dataset/validate/registry.ts"; +import { resolveTextHyperParameters } from "./common.ts"; +import { MAX_CPT_BYTES } from "../../dataset/validate/common.ts"; -export const cptProfile = textProfile("cpt", "cpt", "cpt"); +export const cptProfile: TrainingProfile = { + clientTrainingType: "cpt", + serverTrainingType: "cpt", + acceptedExtensions: [".jsonl"], + + async validate( + filePath: string, + _modality: DataModality, + opts: ValidateOpts, + ): Promise { + return validateDataset(filePath, { ...opts, schema: "cpt", maxBytes: MAX_CPT_BYTES }); + }, + + resolveHyperParameters( + _modality: DataModality, + flags: Record, + ): Record { + return resolveTextHyperParameters(flags); + }, + + shouldSkipGate(_gate: string, _modality: DataModality): boolean { + return false; + }, + + shouldSkipCapabilityCheck(_modality: DataModality): boolean { + return false; + }, +}; diff --git a/packages/core/tests/dataset-validate.test.ts b/packages/core/tests/dataset-validate.test.ts index b4b2b52b..c1e67c5f 100644 --- a/packages/core/tests/dataset-validate.test.ts +++ b/packages/core/tests/dataset-validate.test.ts @@ -56,8 +56,6 @@ describe("validateDataset — DPO schema", () => { }); test('schema "dpo" requires both chosen and rejected on every record', async () => { - // A record with neither chosen nor rejected is SFT-shaped; under --schema dpo - // it must be flagged as missing both preferences. const p = file("sft_under_dpo.jsonl", [SFT_OK]); const r = await validateDataset(p, { fullValidate: true, schema: "dpo" }); expect(r.valid).toBe(false); @@ -107,6 +105,15 @@ describe("validateDataset — DPO schema", () => { const r = await validateDataset(p, { fullValidate: true, schema: "dpo" }); expect(r.valid).toBe(true); }); + + test("DPO messages ending with assistant → DPO_LAST_MSG_NOT_USER warning", async () => { + const p = file("dpo_last_asst.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"yo"}],"chosen":{"role":"assistant","content":"good"},"rejected":{"role":"assistant","content":"bad"}}', + ]); + const r = await validateDataset(p, { fullValidate: true, schema: "dpo" }); + expect(r.valid).toBe(true); + expect(codes(r).warnings).toContain("DPO_LAST_MSG_NOT_USER"); + }); }); describe("validateDataset — CPT schema", () => { @@ -142,8 +149,6 @@ describe("validateDataset — CPT schema", () => { }); test("auto-detect routes a {text} record to CPT, not ChatML", async () => { - // A CPT record has no `messages`; under auto-detect it must NOT produce a - // ChatML MISSING_MESSAGES error — it should be validated as CPT and pass. const p = file("cpt_auto.jsonl", [CPT_OK]); const r = await validateDataset(p, { fullValidate: true }); expect(r.valid).toBe(true); @@ -151,8 +156,6 @@ describe("validateDataset — CPT schema", () => { }); test("SFT record with a stray text field still routes to ChatML", async () => { - // {messages, text} is ambiguous; CPT detect requires text AND no messages, - // so this falls through to ChatML and validates as SFT (text ignored). const p = file("mixed.jsonl", [ '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"yo"}],"text":"noise"}', ]); @@ -162,6 +165,271 @@ describe("validateDataset — CPT schema", () => { }); }); +describe("validateDataset — content array format", () => { + test("content as [{text}] array passes validation", async () => { + const p = file("content_arr.jsonl", [ + '{"messages":[{"role":"system","content":[{"text":"sys"}]},{"role":"user","content":[{"text":"hi"}]},{"role":"assistant","content":[{"text":"hello"}]}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).errors).toEqual([]); + }); + + test("content as plain string still passes (legacy format)", async () => { + const p = file("content_str.jsonl", [SFT_OK]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + }); + + test("content array with image item passes (VL multimodal)", async () => { + const p = file("content_img.jsonl", [ + '{"messages":[{"role":"user","content":[{"text":"describe"},{"image":"img1.jpg"}]},{"role":"assistant","content":[{"text":"a cat"}]}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + }); + + test("content array with video string item passes", async () => { + const p = file("content_vid.jsonl", [ + '{"messages":[{"role":"user","content":[{"text":"describe"},{"video":"vid1.mp4"}]},{"role":"assistant","content":[{"text":"a car"}]}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + }); + + test("content array with video frame list passes", async () => { + const p = file("content_frames.jsonl", [ + '{"messages":[{"role":"user","content":[{"text":"describe"},{"video":["0.jpg","1.jpg","2.jpg"]}]},{"role":"assistant","content":[{"text":"frames"}]}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + }); + + test("content array with invalid item (no text/image/video) → error", async () => { + const p = file("content_bad_item.jsonl", [ + '{"messages":[{"role":"user","content":[{"foo":"bar"}]},{"role":"assistant","content":"ok"}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("CONTENT_ITEM_NO_KNOWN_FIELD"); + }); + + test("content as number → INVALID_CONTENT error", async () => { + const p = file("content_num.jsonl", [ + '{"messages":[{"role":"user","content":42},{"role":"assistant","content":"ok"}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("INVALID_CONTENT"); + }); + + test("empty content array → EMPTY_CONTENT_ARRAY error", async () => { + const p = file("content_empty_arr.jsonl", [ + '{"messages":[{"role":"user","content":[]},{"role":"assistant","content":"ok"}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("EMPTY_CONTENT_ARRAY"); + }); + + test("DPO with content array format passes", async () => { + const p = file("dpo_arr.jsonl", [ + '{"messages":[{"role":"user","content":[{"text":"hi"}]}],"chosen":{"role":"assistant","content":[{"text":"good"}]},"rejected":{"role":"assistant","content":[{"text":"bad"}]}}', + ]); + const r = await validateDataset(p, { fullValidate: true, schema: "dpo" }); + expect(r.valid).toBe(true); + }); +}); + +describe("validateDataset — tool calling (function calling)", () => { + const TOOL_OK = JSON.stringify({ + tools: [ + { + type: "function", + function: { + name: "get_weather", + description: "get weather", + parameters: { + type: "object", + properties: { city: { type: "string" } }, + required: ["city"], + }, + }, + }, + ], + messages: [ + { role: "user", content: [{ text: "weather in Beijing" }] }, + { + role: "assistant", + content: [{ text: "let me check" }], + tool_calls: [ + { + id: "call_1", + type: "function", + function: { name: "get_weather", arguments: '{"city":"Beijing"}' }, + }, + ], + }, + { role: "tool", tool_call_id: "call_1", content: [{ text: '{"weather":"sunny"}' }] }, + { role: "assistant", content: [{ text: "It is sunny." }] }, + ], + }); + + test("valid tool calling record passes", async () => { + const p = file("tool_ok.jsonl", [TOOL_OK]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).errors).toEqual([]); + }); + + test("tool role is accepted (no INVALID_ROLE)", async () => { + const p = file("tool_role.jsonl", [TOOL_OK]); + const r = await validateDataset(p, { fullValidate: true }); + expect(codes(r).errors).not.toContain("INVALID_ROLE"); + }); + + test("tool message without tool_call_id → TOOL_MISSING_CALL_ID", async () => { + const p = file("tool_no_id.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"","tool_calls":[{"id":"c1","type":"function","function":{"name":"f","arguments":"{}"}}]},{"role":"tool","content":"result"},{"role":"assistant","content":"done"}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("TOOL_MISSING_CALL_ID"); + }); + + test("tool_call_id unmatched → TOOL_CALL_ID_UNMATCHED warning", async () => { + const p = file("tool_unmatched.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"","tool_calls":[{"id":"c1","type":"function","function":{"name":"f","arguments":"{}"}}]},{"role":"tool","tool_call_id":"WRONG_ID","content":"result"},{"role":"assistant","content":"done"}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).warnings).toContain("TOOL_CALL_ID_UNMATCHED"); + }); + + test("tool_calls with missing function name → TOOL_CALL_FN_NO_NAME", async () => { + const p = file("tool_no_fn_name.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"","tool_calls":[{"id":"c1","type":"function","function":{"arguments":"{}"}}]},{"role":"tool","tool_call_id":"c1","content":"r"},{"role":"assistant","content":"ok"}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("TOOL_CALL_FN_NO_NAME"); + }); + + test("assistant with tool_calls but no content is valid", async () => { + const p = file("tool_no_content.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","tool_calls":[{"id":"c1","type":"function","function":{"name":"f","arguments":"{}"}}]},{"role":"tool","tool_call_id":"c1","content":"r"},{"role":"assistant","content":"ok"}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).errors).not.toContain("MISSING_CONTENT"); + }); +}); + +describe("validateDataset — thinking tags", () => { + test("think tag in last assistant is valid", async () => { + const p = file("think_ok.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"\\nreasoning\\n\\n\\nanswer"}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).warnings).not.toContain("THINK_TAG_NOT_LAST"); + }); + + test("think tag in non-last assistant → THINK_TAG_NOT_LAST warning", async () => { + const p = file("think_mid.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"\\nearly\\n\\n\\nmid"},{"role":"user","content":"more"},{"role":"assistant","content":"final"}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).warnings).toContain("THINK_TAG_NOT_LAST"); + }); + + test("think tag in content array format detected", async () => { + const p = file("think_arr.jsonl", [ + '{"messages":[{"role":"user","content":[{"text":"hi"}]},{"role":"assistant","content":[{"text":"\\nreason\\n\\n\\nans"}]},{"role":"user","content":[{"text":"more"}]},{"role":"assistant","content":[{"text":"final"}]}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).warnings).toContain("THINK_TAG_NOT_LAST"); + }); +}); + +describe("validateDataset — OpenAI migration guards", () => { + test("message-level name field → UNSUPPORTED_FIELD_NAME warning", async () => { + const p = file("openai_name.jsonl", [ + '{"messages":[{"role":"user","content":"hi","name":"alice"},{"role":"assistant","content":"hello"}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).warnings).toContain("UNSUPPORTED_FIELD_NAME"); + }); + + test("message-level weight field → UNSUPPORTED_FIELD_WEIGHT warning", async () => { + const p = file("openai_weight.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"hello","weight":0.5}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).warnings).toContain("UNSUPPORTED_FIELD_WEIGHT"); + }); + + test("record-level weight field → UNSUPPORTED_FIELD_WEIGHT warning", async () => { + const p = file("record_weight.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"hello"}],"weight":1}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).warnings).toContain("UNSUPPORTED_FIELD_WEIGHT"); + }); +}); + +describe("validateDataset — loss_weight", () => { + test("valid loss_weight (0.5) passes", async () => { + const p = file("lw_ok.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"hello"}],"loss_weight":0.5}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).errors).not.toContain("INVALID_LOSS_WEIGHT"); + }); + + test("loss_weight out of range (1.5) → INVALID_LOSS_WEIGHT", async () => { + const p = file("lw_bad.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"hello"}],"loss_weight":1.5}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("INVALID_LOSS_WEIGHT"); + }); + + test("loss_weight negative → INVALID_LOSS_WEIGHT", async () => { + const p = file("lw_neg.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"hello"}],"loss_weight":-0.1}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("INVALID_LOSS_WEIGHT"); + }); + + test("loss_weight non-number → INVALID_LOSS_WEIGHT", async () => { + const p = file("lw_str.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"hello"}],"loss_weight":"high"}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("INVALID_LOSS_WEIGHT"); + }); + + test("loss_weight boundary values 0 and 1 pass", async () => { + const p = file("lw_boundary.jsonl", [ + '{"messages":[{"role":"user","content":"a"},{"role":"assistant","content":"b"}],"loss_weight":0}', + '{"messages":[{"role":"user","content":"c"},{"role":"assistant","content":"d"}],"loss_weight":1}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + }); +}); + describe("parseDatasetSchemaFlag", () => { test("undefined / empty → undefined (auto)", () => { expect(parseDatasetSchemaFlag(undefined)).toBeUndefined(); diff --git a/skills/bailian-finetune/reference/dataset.md b/skills/bailian-finetune/reference/dataset.md index 22cad6db..2a3b9810 100644 --- a/skills/bailian-finetune/reference/dataset.md +++ b/skills/bailian-finetune/reference/dataset.md @@ -117,7 +117,7 @@ bl dataset list --output json | Flag | Type | Required | Description | | ------------------ | ------ | -------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | -| `--file ` | string | yes | Local dataset file (.jsonl or .zip; ≤300MB text, ≤1GB image) | +| `--file ` | string | yes | Local dataset file (.jsonl or .zip; ≤200MB SFT/DPO, ≤300MB CPT, ≤2GB media zip) | | `--purpose ` | string | no | Dataset purpose tag (default: "fine-tune"; e.g. "evaluation") | | `--schema ` | string | no | Record schema: "chatml" (SFT), "dpo" (chosen/rejected), "cpt" (raw text), "tts" (audio), "image" (image generation), or "video" (video generation). Default auto-detects per record. | | `--no-validate` | switch | no | Skip the local JSONL pre-flight check (not recommended) | @@ -128,13 +128,14 @@ bl dataset list --output json #### Notes - Supports .jsonl (text) and .zip (audio/image archives with a data.jsonl -- manifest). Five record schemas are recognized: chatml = {messages:[...]} +- manifest). Six record schemas are recognized: chatml = {messages:[...]} - (SFT); dpo = {messages:[...], chosen, rejected}; cpt = {text:"..."} - (continual pre-training, raw text); tts = {wav_fn:"train/xxx.wav", - text:"..."} (audio fine-tuning); image = {img_path:"..."} (image -- generation). With no --schema, a record carrying wav_fn is validated as -- TTS, img_path as image, chosen/rejected as DPO, text (no messages) as CPT, -- otherwise ChatML. Upload cap: 300MB text, 1GB image. Upload uses the +- generation); video = {first_frame_path:...} (video generation). With no +- --schema, a record carrying wav_fn is validated as TTS, img_path as image, +- chosen/rejected as DPO, text (no messages) as CPT, otherwise ChatML. +- Upload cap: 200MB SFT/DPO text, 300MB CPT, 2GB media zip. Upload uses the - OpenAI-compatible /compatible-mode/v1/files endpoint so the purpose tag is - persisted (the DashScope-native /api/v1/files drops it). From d74d4efcd015c28d7f35f7a1668cfd1e2bda55bc Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Wed, 12 Aug 2026 10:09:10 +0800 Subject: [PATCH 14/14] fix(dataset): align validation with the platform data-format rules doc MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Reviewed against the official text-tuning data rules; fixes two confirmed mismatches and fills enforcement gaps: - thinking: exempt assistant messages carrying tool_calls from the THINK_TAG_NOT_LAST check — the spec's tool+thinking combo example puts on a non-last assistant and was previously false-flagged - DPO support matrix: reject image/video content items, tools, tool_calls and role:tool (DPO_UNSUPPORTED_ELEMENT); also scan chosen/rejected - DPO: messages not ending with user upgraded warning -> error - OpenAI migration: name/weight upgraded warning -> error (spec: must not carry); drop dead record-level name branch - tool_call_id: unmatched tool response upgraded warning -> error (one-to-one per spec); new TOOL_CALL_NO_RESPONSE warning for orphan calls - loss_weight: validate range at message level too; warn when placed on anything but the last assistant message (LOSS_WEIGHT_PLACEMENT) - video params: fps/sample_fps must be within [0.1, 10] (INVALID_VIDEO_FPS); mode-mismatched params warned (VIDEO_PARAM_MODE_MISMATCH); video_start/video_end type-checked - zip: skip macOS packaging metadata (__MACOSX/, .DS_Store, ._*) in filename constraints and image counting to stop false failures on Finder-created archives Tests 45 -> 59 covering every new/changed rule, including a replica of the spec's official tool+thinking example. --- .../src/dataset/validate/schemas/chatml.ts | 178 +++++++++++++---- .../core/src/dataset/validate/schemas/dpo.ts | 68 ++++++- packages/core/src/dataset/validate/zip.ts | 23 ++- packages/core/tests/dataset-validate.test.ts | 187 ++++++++++++++++-- 4 files changed, 394 insertions(+), 62 deletions(-) diff --git a/packages/core/src/dataset/validate/schemas/chatml.ts b/packages/core/src/dataset/validate/schemas/chatml.ts index 959149e8..00b4ca20 100644 --- a/packages/core/src/dataset/validate/schemas/chatml.ts +++ b/packages/core/src/dataset/validate/schemas/chatml.ts @@ -20,6 +20,69 @@ import type { RecordSchemaSpec } from "./types.ts"; const VALID_ROLES = new Set(["system", "user", "assistant", "tool"]); +/** Platform bounds for video sampling rate params (`fps` / `sample_fps`). */ +const VIDEO_FPS_MIN = 0.1; +const VIDEO_FPS_MAX = 10; + +/** + * Validate the sampling/clipping params carried by a video content item. + * Mode rules (platform spec): + * - path mode (video: string): `fps`, `video_start`, `video_end` allowed; `sample_fps` is not + * - frame-list mode (video: string[]): `sample_fps` allowed; `fps` / `video_start` / `video_end` are not + * `fps` / `sample_fps` must be numbers within [0.1, 10] when present. + */ +function inspectVideoParams( + item: Record, + isFrameList: boolean, + lineNo: number, + itemPath: string, +): ValidationIssue[] { + const out: ValidationIssue[] = []; + const checkFpsRange = (field: "fps" | "sample_fps"): void => { + if (!(field in item)) return; + const value = item[field]; + if (typeof value !== "number" || value < VIDEO_FPS_MIN || value > VIDEO_FPS_MAX) { + out.push( + makeIssue( + "error", + "INVALID_VIDEO_FPS", + `"${field}" must be a number between ${VIDEO_FPS_MIN} and ${VIDEO_FPS_MAX} (got ${JSON.stringify(value)}).`, + { line: lineNo, path: `${itemPath}.${field}` }, + ), + ); + } + }; + checkFpsRange("fps"); + checkFpsRange("sample_fps"); + + const wrongModeFields = isFrameList ? ["fps", "video_start", "video_end"] : ["sample_fps"]; + const modeName = isFrameList ? "frame-list" : "file-path"; + for (const field of wrongModeFields) { + if (field in item) { + out.push( + makeIssue( + "warning", + "VIDEO_PARAM_MODE_MISMATCH", + `"${field}" does not apply to ${modeName} video mode and will be ignored by the platform.`, + { line: lineNo, path: `${itemPath}.${field}` }, + ), + ); + } + } + + for (const field of ["video_start", "video_end"] as const) { + if (field in item && !isFrameList && typeof item[field] !== "number") { + out.push( + makeIssue("error", "INVALID_VIDEO_CLIP_TIME", `"${field}" must be a number (seconds).`, { + line: lineNo, + path: `${itemPath}.${field}`, + }), + ); + } + } + return out; +} + /** * Validate a content field that may be: * - A plain string (legacy format) @@ -112,19 +175,22 @@ export function inspectContentField( { line: lineNo, path: `${itemPath}.video` }, ), ); - } else if (Array.isArray(video)) { - for (let frameIdx = 0; frameIdx < video.length; frameIdx++) { - if (typeof video[frameIdx] !== "string") { - out.push( - makeIssue( - "error", - "INVALID_VIDEO_FRAME", - `Video frame list item at index ${frameIdx} must be a string.`, - { line: lineNo, path: `${itemPath}.video[${frameIdx}]` }, - ), - ); + } else { + if (Array.isArray(video)) { + for (let frameIdx = 0; frameIdx < video.length; frameIdx++) { + if (typeof video[frameIdx] !== "string") { + out.push( + makeIssue( + "error", + "INVALID_VIDEO_FRAME", + `Video frame list item at index ${frameIdx} must be a string.`, + { line: lineNo, path: `${itemPath}.video[${frameIdx}]` }, + ), + ); + } } } + out.push(...inspectVideoParams(obj, Array.isArray(video), lineNo, itemPath)); } } } @@ -283,13 +349,13 @@ export function inspectMessageObject( out.push(...inspectToolCalls(record.tool_calls, lineNo, `${path}.tool_calls`)); } - // OpenAI migration guard: name / weight are not supported by Bailian + // OpenAI migration guard: the platform rejects data carrying name / weight if ("name" in record) { out.push( makeIssue( - "warning", + "error", "UNSUPPORTED_FIELD_NAME", - `Field "name" is not supported by Bailian. Remove it when migrating from OpenAI/Azure.`, + `Field "name" is not supported by Bailian and must be removed when migrating from OpenAI/Azure.`, { line: lineNo, path: `${path}.name` }, ), ); @@ -297,9 +363,10 @@ export function inspectMessageObject( if ("weight" in record) { out.push( makeIssue( - "warning", + "error", "UNSUPPORTED_FIELD_WEIGHT", - `Field "weight" is not supported by Bailian. Remove it when migrating from OpenAI/Azure.`, + `Field "weight" is not supported by Bailian and must be removed when migrating from OpenAI/Azure. ` + + `All assistant outputs are trained; per-line importance uses "loss_weight" (invite-only).`, { line: lineNo, path: `${path}.weight` }, ), ); @@ -419,12 +486,15 @@ export function inspectChatMLRecord( ); } - // tool_call_id correspondence: every tool response should reference a known call id + // tool_call_id correspondence must be one-to-one (platform spec): + // every tool response must reference a known call id (hard error), and every + // tool_call should receive a response (advisory — trailing calls are dubious + // in training data but we cannot rule out platform-side tolerance). for (const responseId of toolResponseIds) { if (!toolCallIds.has(responseId)) { out.push( makeIssue( - "warning", + "error", "TOOL_CALL_ID_UNMATCHED", `tool message references tool_call_id "${responseId}" which does not match any assistant tool_calls[].id.`, { line: lineNo, path: "messages" }, @@ -432,20 +502,37 @@ export function inspectChatMLRecord( ); } } + for (const callId of toolCallIds) { + if (!toolResponseIds.has(callId)) { + out.push( + makeIssue( + "warning", + "TOOL_CALL_NO_RESPONSE", + `assistant tool_calls[].id "${callId}" has no matching tool response message.`, + { line: lineNo, path: "messages" }, + ), + ); + } + } - // thinking tag check: …` should only appear in the last assistant message + // thinking tag check: … should only appear in the last + // assistant message. Exemption (platform spec, tool+thinking combo): an + // assistant that carries tool_calls may legitimately hold a block + // even when it is not the last assistant message. if (lastAssistantIdx >= 0) { for (let idx = 0; idx < messages.length; idx++) { if (idx === lastAssistantIdx) continue; const msg = messages[idx] as Record | null; if (msg?.role !== "assistant") continue; + if (msg && Array.isArray(msg.tool_calls)) continue; const content = msg?.content; if (contentHasThinkTag(content)) { out.push( makeIssue( "warning", "THINK_TAG_NOT_LAST", - `Thinking tags (…) should only appear in the last assistant message, found at messages[${idx}].`, + `Thinking tags (…) should only appear in the last assistant message ` + + `(or an assistant message carrying tool_calls), found at messages[${idx}].`, { line: lineNo, path: `messages[${idx}].content` }, ), ); @@ -453,39 +540,50 @@ export function inspectChatMLRecord( } } - // loss_weight validation (record-level, invite-only parameter) - if ("loss_weight" in record) { - const lossWeight = record.loss_weight; - if (typeof lossWeight !== "number" || lossWeight < 0 || lossWeight > 1) { + // loss_weight validation (invite-only parameter). + // Range is enforced wherever the field appears (record level and message + // level); placement follows the spec: only the LAST assistant message line + // supports loss_weight — misplaced occurrences are advisory (invite-only + // semantics are account-specific, so we do not hard-fail). + const checkLossWeightRange = (value: unknown, path: string): void => { + if (typeof value !== "number" || value < 0 || value > 1) { out.push( makeIssue( "error", "INVALID_LOSS_WEIGHT", - `"loss_weight" must be a number between 0.0 and 1.0 (got ${JSON.stringify(lossWeight)}).`, - { line: lineNo, path: "loss_weight" }, + `"loss_weight" must be a number between 0.0 and 1.0 (got ${JSON.stringify(value)}).`, + { line: lineNo, path }, ), ); } + }; + if ("loss_weight" in record) { + checkLossWeightRange(record.loss_weight, "loss_weight"); } - - // OpenAI migration guard at record level - if ("name" in record && !("messages" in record)) { - // Only warn at record level if it's not inside messages (messages handled above) - out.push( - makeIssue( - "warning", - "UNSUPPORTED_FIELD_NAME", - `Record-level field "name" is not supported by Bailian.`, - { line: lineNo, path: "name" }, - ), - ); + for (let idx = 0; idx < messages.length; idx++) { + const msg = messages[idx] as Record | null; + if (!msg || !("loss_weight" in msg)) continue; + checkLossWeightRange(msg.loss_weight, `messages[${idx}].loss_weight`); + if (!(msg.role === "assistant" && idx === lastAssistantIdx)) { + out.push( + makeIssue( + "warning", + "LOSS_WEIGHT_PLACEMENT", + `"loss_weight" is only supported on the last assistant message; found at messages[${idx}] (role "${String(msg.role)}").`, + { line: lineNo, path: `messages[${idx}].loss_weight` }, + ), + ); + } } + + // OpenAI migration guard at record level (message-level occurrences are + // handled by inspectMessageObject above) if ("weight" in record) { out.push( makeIssue( - "warning", + "error", "UNSUPPORTED_FIELD_WEIGHT", - `Record-level field "weight" is not supported by Bailian. Remove it when migrating from OpenAI/Azure.`, + `Record-level field "weight" is not supported by Bailian and must be removed when migrating from OpenAI/Azure.`, { line: lineNo, path: "weight" }, ), ); diff --git a/packages/core/src/dataset/validate/schemas/dpo.ts b/packages/core/src/dataset/validate/schemas/dpo.ts index 26c3bcc3..86de3aa9 100644 --- a/packages/core/src/dataset/validate/schemas/dpo.ts +++ b/packages/core/src/dataset/validate/schemas/dpo.ts @@ -18,16 +18,70 @@ function inspectDPORecord(record: Record, lineNo: number): Vali const messages = record.messages; if (!Array.isArray(messages) || messages.length === 0) return out; - // DPO convention: messages should end with a user message (the prompt that - // chosen/rejected respond to). If the last message is assistant, it's likely - // a structural mistake. + /** image / video content items are outside the DPO support matrix */ + const mediaIssues = (content: unknown, basePath: string): ValidationIssue[] => { + if (!Array.isArray(content)) return []; + const found: ValidationIssue[] = []; + for (let itemIdx = 0; itemIdx < content.length; itemIdx++) { + const item = content[itemIdx] as Record | null; + if (!item || typeof item !== "object") continue; + for (const mediaField of ["image", "video"] as const) { + if (mediaField in item) { + found.push( + makeIssue( + "error", + "DPO_UNSUPPORTED_ELEMENT", + `DPO training data does not support ${mediaField} inputs; found at ${basePath}.content[${itemIdx}].`, + { line: lineNo, path: `${basePath}.content[${itemIdx}].${mediaField}` }, + ), + ); + } + } + } + return found; + }; + + // Support matrix (platform spec): DPO is text + thinking ONLY — no image / + // video inputs and no tool calling. Reject multimodal items and tool fields + // that the SFT-oriented ChatML inspector would otherwise accept. + if ("tools" in record) { + out.push( + makeIssue( + "error", + "DPO_UNSUPPORTED_ELEMENT", + `DPO training data does not support tool calling; remove the "tools" definition.`, + { line: lineNo, path: "tools" }, + ), + ); + } + for (let idx = 0; idx < messages.length; idx++) { + const msg = messages[idx] as Record | null; + if (!msg) continue; + const msgPath = `messages[${idx}]`; + if (msg.role === "tool" || "tool_calls" in msg) { + out.push( + makeIssue( + "error", + "DPO_UNSUPPORTED_ELEMENT", + `DPO training data does not support tool calling; found ${ + msg.role === "tool" ? `role "tool"` : `"tool_calls"` + } at ${msgPath}.`, + { line: lineNo, path: msgPath }, + ), + ); + } + out.push(...mediaIssues(msg.content, msgPath)); + } + + // DPO trains the preference for the LAST user input — messages ending with + // any other role make the chosen/rejected pair semantically meaningless. const lastMsg = messages[messages.length - 1] as Record | null; if (lastMsg && lastMsg.role !== "user") { out.push( makeIssue( - "warning", + "error", "DPO_LAST_MSG_NOT_USER", - `DPO "messages" should end with a "user" message (the prompt for chosen/rejected). ` + + `DPO "messages" must end with a "user" message (the prompt for chosen/rejected). ` + `Got "${String(lastMsg.role)}" as the last message.`, { line: lineNo, path: `messages[${messages.length - 1}].role` }, ), @@ -55,6 +109,7 @@ function inspectDPORecord(record: Record, lineNo: number): Vali } if (hasChosen) { out.push(...inspectMessageObject(record.chosen, lineNo, "chosen")); + out.push(...mediaIssues((record.chosen as Record | null)?.content, "chosen")); const role = (record.chosen as Record | null)?.role; if (typeof role === "string" && role !== "assistant") { out.push( @@ -69,6 +124,9 @@ function inspectDPORecord(record: Record, lineNo: number): Vali } if (hasRejected) { out.push(...inspectMessageObject(record.rejected, lineNo, "rejected")); + out.push( + ...mediaIssues((record.rejected as Record | null)?.content, "rejected"), + ); const role = (record.rejected as Record | null)?.role; if (typeof role === "string" && role !== "assistant") { out.push( diff --git a/packages/core/src/dataset/validate/zip.ts b/packages/core/src/dataset/validate/zip.ts index da1c3622..232a5f95 100644 --- a/packages/core/src/dataset/validate/zip.ts +++ b/packages/core/src/dataset/validate/zip.ts @@ -108,18 +108,36 @@ function basenameNoExt(segment: string): string { return dot > 0 ? segment.slice(0, dot) : segment; } +/** + * macOS Finder/zip metadata entries (`__MACOSX/` resource forks, `.DS_Store`, + * AppleDouble `._*` files). They are packaging noise, not training data: + * exclude them from filename constraints and media counting so Mac-created + * archives don't fail on artifacts the user never sees. + */ +function isZipMetadataEntry(entry: string): boolean { + if (entry === "__MACOSX" || entry.startsWith("__MACOSX/")) return true; + const lastSegment = + entry + .split("/") + .filter((segment) => segment.length > 0) + .pop() ?? ""; + return lastSegment === ".DS_Store" || lastSegment.startsWith("._"); +} + /** * Validate ZIP entry filenames against platform constraints. * Returns issues for charset violations, over-length names, and duplicates. + * Exported for direct unit testing (not re-exported by the barrel). */ -function validateZipFilenames(entries: string[]): ValidationIssue[] { +export function validateZipFilenames(entries: string[]): ValidationIssue[] { const out: ValidationIssue[] = []; const seenBasenames = new Map(); // basename (no ext) → first full path const MAX_REPORTED = 10; for (const entry of entries) { - // Skip directory entries + // Skip directory entries and macOS packaging metadata if (entry.endsWith("/")) continue; + if (isZipMetadataEntry(entry)) continue; // Check each path segment (folder names + file name) const segments = entry.split("/").filter((s) => s.length > 0); @@ -383,6 +401,7 @@ export const zipValidator: ValidatorSpec = { const imageFiles = entries.filter((entry) => { if (entry === "data.jsonl" || entry.endsWith("/data.jsonl")) return false; if (entry.endsWith("/")) return false; // directory entries + if (isZipMetadataEntry(entry)) return false; // __MACOSX/._x.jpg is not an image const dot = entry.lastIndexOf("."); const ext = dot >= 0 ? entry.slice(dot).toLowerCase() : ""; return IMAGE_EXTENSIONS.has(ext); diff --git a/packages/core/tests/dataset-validate.test.ts b/packages/core/tests/dataset-validate.test.ts index c1e67c5f..b478cca6 100644 --- a/packages/core/tests/dataset-validate.test.ts +++ b/packages/core/tests/dataset-validate.test.ts @@ -106,13 +106,41 @@ describe("validateDataset — DPO schema", () => { expect(r.valid).toBe(true); }); - test("DPO messages ending with assistant → DPO_LAST_MSG_NOT_USER warning", async () => { + test("DPO messages ending with assistant → DPO_LAST_MSG_NOT_USER error", async () => { const p = file("dpo_last_asst.jsonl", [ '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"yo"}],"chosen":{"role":"assistant","content":"good"},"rejected":{"role":"assistant","content":"bad"}}', ]); const r = await validateDataset(p, { fullValidate: true, schema: "dpo" }); - expect(r.valid).toBe(true); - expect(codes(r).warnings).toContain("DPO_LAST_MSG_NOT_USER"); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("DPO_LAST_MSG_NOT_USER"); + }); + + test("DPO with image content item → DPO_UNSUPPORTED_ELEMENT error", async () => { + const p = file("dpo_image.jsonl", [ + '{"messages":[{"role":"user","content":[{"text":"look"},{"image":"a.jpg"}]}],"chosen":{"role":"assistant","content":[{"text":"good"}]},"rejected":{"role":"assistant","content":[{"text":"bad"}]}}', + ]); + const r = await validateDataset(p, { fullValidate: true, schema: "dpo" }); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("DPO_UNSUPPORTED_ELEMENT"); + }); + + test("DPO with tools / tool_calls → DPO_UNSUPPORTED_ELEMENT error", async () => { + const p = file("dpo_tools.jsonl", [ + '{"tools":[{"type":"function","function":{"name":"f","parameters":{}}}],"messages":[{"role":"user","content":"hi"}],"chosen":{"role":"assistant","content":"good"},"rejected":{"role":"assistant","content":"bad"}}', + ]); + const r = await validateDataset(p, { fullValidate: true, schema: "dpo" }); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("DPO_UNSUPPORTED_ELEMENT"); + }); + + test("DPO chosen carrying an image item → DPO_UNSUPPORTED_ELEMENT error", async () => { + const p = file("dpo_chosen_image.jsonl", [ + '{"messages":[{"role":"user","content":"hi"}],"chosen":{"role":"assistant","content":[{"text":"good"},{"image":"x.png"}]},"rejected":{"role":"assistant","content":"bad"}}', + ]); + const r = await validateDataset(p, { fullValidate: true, schema: "dpo" }); + expect(r.valid).toBe(false); + const err = r.errors.find((e) => e.code === "DPO_UNSUPPORTED_ELEMENT"); + expect(err?.path).toContain("chosen"); }); }); @@ -297,13 +325,23 @@ describe("validateDataset — tool calling (function calling)", () => { expect(codes(r).errors).toContain("TOOL_MISSING_CALL_ID"); }); - test("tool_call_id unmatched → TOOL_CALL_ID_UNMATCHED warning", async () => { + test("tool_call_id unmatched → TOOL_CALL_ID_UNMATCHED error", async () => { const p = file("tool_unmatched.jsonl", [ '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"","tool_calls":[{"id":"c1","type":"function","function":{"name":"f","arguments":"{}"}}]},{"role":"tool","tool_call_id":"WRONG_ID","content":"result"},{"role":"assistant","content":"done"}]}', ]); const r = await validateDataset(p, { fullValidate: true }); - expect(r.valid).toBe(true); - expect(codes(r).warnings).toContain("TOOL_CALL_ID_UNMATCHED"); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("TOOL_CALL_ID_UNMATCHED"); + // The orphaned call side is advisory + expect(codes(r).warnings).toContain("TOOL_CALL_NO_RESPONSE"); + }); + + test("tool_calls without a tool response → TOOL_CALL_NO_RESPONSE warning", async () => { + const p = file("tool_no_resp.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"","tool_calls":[{"id":"c1","type":"function","function":{"name":"f","arguments":"{}"}}]},{"role":"assistant","content":"done"}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(codes(r).warnings).toContain("TOOL_CALL_NO_RESPONSE"); }); test("tool_calls with missing function name → TOOL_CALL_FN_NO_NAME", async () => { @@ -352,34 +390,68 @@ describe("validateDataset — thinking tags", () => { expect(r.valid).toBe(true); expect(codes(r).warnings).toContain("THINK_TAG_NOT_LAST"); }); + + test("official tool+thinking combo: think in non-last assistant WITH tool_calls is exempt", async () => { + // Mirrors the platform spec's 工具与思考组合 example: the assistant that + // issues tool_calls carries the block, the final assistant answers. + const p = file("think_tool_combo.jsonl", [ + JSON.stringify({ + tools: [ + { + type: "function", + function: { name: "get_weather", description: "d", parameters: { type: "object" } }, + }, + ], + messages: [ + { role: "user", content: [{ text: "weather in Beijing?" }] }, + { + role: "assistant", + content: [{ text: "\nneed the weather tool\n\n" }], + tool_calls: [ + { + id: "call_1", + type: "function", + function: { name: "get_weather", arguments: '{"city":"Beijing"}' }, + }, + ], + }, + { role: "tool", tool_call_id: "call_1", content: [{ text: '{"weather":"sunny"}' }] }, + { role: "assistant", content: [{ text: "It is sunny in Beijing." }] }, + ], + }), + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).warnings).not.toContain("THINK_TAG_NOT_LAST"); + }); }); describe("validateDataset — OpenAI migration guards", () => { - test("message-level name field → UNSUPPORTED_FIELD_NAME warning", async () => { + test("message-level name field → UNSUPPORTED_FIELD_NAME error", async () => { const p = file("openai_name.jsonl", [ '{"messages":[{"role":"user","content":"hi","name":"alice"},{"role":"assistant","content":"hello"}]}', ]); const r = await validateDataset(p, { fullValidate: true }); - expect(r.valid).toBe(true); - expect(codes(r).warnings).toContain("UNSUPPORTED_FIELD_NAME"); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("UNSUPPORTED_FIELD_NAME"); }); - test("message-level weight field → UNSUPPORTED_FIELD_WEIGHT warning", async () => { + test("message-level weight field → UNSUPPORTED_FIELD_WEIGHT error", async () => { const p = file("openai_weight.jsonl", [ '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"hello","weight":0.5}]}', ]); const r = await validateDataset(p, { fullValidate: true }); - expect(r.valid).toBe(true); - expect(codes(r).warnings).toContain("UNSUPPORTED_FIELD_WEIGHT"); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("UNSUPPORTED_FIELD_WEIGHT"); }); - test("record-level weight field → UNSUPPORTED_FIELD_WEIGHT warning", async () => { + test("record-level weight field → UNSUPPORTED_FIELD_WEIGHT error", async () => { const p = file("record_weight.jsonl", [ '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"hello"}],"weight":1}', ]); const r = await validateDataset(p, { fullValidate: true }); - expect(r.valid).toBe(true); - expect(codes(r).warnings).toContain("UNSUPPORTED_FIELD_WEIGHT"); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("UNSUPPORTED_FIELD_WEIGHT"); }); }); @@ -428,6 +500,91 @@ describe("validateDataset — loss_weight", () => { const r = await validateDataset(p, { fullValidate: true }); expect(r.valid).toBe(true); }); + + test("message-level loss_weight on the LAST assistant passes without warning", async () => { + const p = file("lw_msg_last.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"hello","loss_weight":0.8}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).warnings).not.toContain("LOSS_WEIGHT_PLACEMENT"); + }); + + test("message-level loss_weight on a non-last assistant → LOSS_WEIGHT_PLACEMENT warning", async () => { + const p = file("lw_msg_mid.jsonl", [ + '{"messages":[{"role":"user","content":"a"},{"role":"assistant","content":"b","loss_weight":0.8},{"role":"user","content":"c"},{"role":"assistant","content":"d"}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(codes(r).warnings).toContain("LOSS_WEIGHT_PLACEMENT"); + }); + + test("message-level loss_weight out of range → INVALID_LOSS_WEIGHT", async () => { + const p = file("lw_msg_range.jsonl", [ + '{"messages":[{"role":"user","content":"hi"},{"role":"assistant","content":"hello","loss_weight":2}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("INVALID_LOSS_WEIGHT"); + }); +}); + +describe("validateDataset — video content params", () => { + test("path-mode video with in-range fps and clip times passes", async () => { + const p = file("video_path_ok.jsonl", [ + '{"messages":[{"role":"user","content":[{"text":"desc"},{"video":"v.mp4","fps":3.0,"video_start":0.0,"video_end":3.0}]},{"role":"assistant","content":[{"text":"ok"}]}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).warnings).not.toContain("VIDEO_PARAM_MODE_MISMATCH"); + }); + + test("fps out of [0.1, 10] → INVALID_VIDEO_FPS error", async () => { + const p = file("video_fps_bad.jsonl", [ + '{"messages":[{"role":"user","content":[{"text":"desc"},{"video":"v.mp4","fps":30}]},{"role":"assistant","content":[{"text":"ok"}]}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(false); + expect(codes(r).errors).toContain("INVALID_VIDEO_FPS"); + }); + + test("sample_fps on path-mode video → VIDEO_PARAM_MODE_MISMATCH warning", async () => { + const p = file("video_mode_mix.jsonl", [ + '{"messages":[{"role":"user","content":[{"text":"desc"},{"video":"v.mp4","sample_fps":2.0}]},{"role":"assistant","content":[{"text":"ok"}]}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(codes(r).warnings).toContain("VIDEO_PARAM_MODE_MISMATCH"); + }); + + test("frame-list video with sample_fps passes; fps there is flagged", async () => { + const p = file("video_frames.jsonl", [ + '{"messages":[{"role":"user","content":[{"text":"desc"},{"video":["0.jpg","1.jpg"],"sample_fps":5.0}]},{"role":"assistant","content":[{"text":"ok"}]}]}', + '{"messages":[{"role":"user","content":[{"text":"desc"},{"video":["0.jpg"],"fps":2.0}]},{"role":"assistant","content":[{"text":"ok"}]}]}', + ]); + const r = await validateDataset(p, { fullValidate: true }); + expect(r.valid).toBe(true); + expect(codes(r).warnings).toContain("VIDEO_PARAM_MODE_MISMATCH"); + }); +}); + +describe("validateZipFilenames — macOS metadata entries", () => { + test("__MACOSX / .DS_Store / ._resource-fork entries are ignored", async () => { + const { validateZipFilenames } = await import("../src/dataset/validate/zip.ts"); + const issues = validateZipFilenames([ + "data.jsonl", + "image_1.jpg", + "__MACOSX/._image_1.jpg", + "__MACOSX/", + ".DS_Store", + "train/._clip.wav", + ]); + expect(issues).toEqual([]); + }); + + test("real charset violations are still reported", async () => { + const { validateZipFilenames } = await import("../src/dataset/validate/zip.ts"); + const issues = validateZipFilenames(["data.jsonl", "图片1.jpg"]); + expect(issues.map((issue) => issue.code)).toContain("INVALID_FILENAME_CHARSET"); + }); }); describe("parseDatasetSchemaFlag", () => {