@tuned-tensor/cli 0.21.0 → 0.22.1

package/dist/local-runtime.js

@@ -1,11 +1,13 @@
 #!/usr/bin/env node
 import {
   DEFAULT_LOCAL_SPEC_PATH,
+  DatasetNotDownloadedError,
   assertArtifactManifest,
   assertFoundationSpecReady,
   assertLocalRunInputReady,
   assertUsableModelArtifact,
   assessHardware,
+  auditRun,
   baseModelRevisionSchema,
   behaviorSpecExampleSchema,
   behaviorSpecSchema,
@@ -18,14 +20,19 @@ import {
   comparisonReportSchema,
   compileSpecToJsonl,
   createLocalStore,
+  createRunConsoleReporter,
+  datasetColumnsSchema,
   datasetFormatSchema,
+  datasetInputIdentity,
   datasetPrebuiltSchema,
+  datasetSourceSchema,
   decisionHyperparametersSchema,
   decisionLabels,
   decisionPlaceholderIssues,
   decisionQuestionSchema,
   defaultBaseModelRevision,
   defaultLocalHome,
+  ensureDatasetCached,
   evalExampleResultSchema,
   evalReportSchema,
   evalSplitSchema,
@@ -42,10 +49,14 @@ import {
   foundationPlaceholderIssues,
   generatedPlaceholderIssues,
   hardenExistingLocalStore,
+  huggingFaceDataset,
+  huggingFaceDatasetSourceSchema,
+  huggingFaceRepoIdSchema,
   initLocalRunnerConfigFile,
   initLocalSpecFile,
   isDecisionSpecFile,
   isFoundationSpecFile,
+  isRepoRelativePath,
   isTerminalRunState,
   jsonFieldMetricsSchema,
   loadJsonFile,
@@ -66,6 +77,7 @@ import {
   parseLocalRunnerConfig,
   pipelineForRunInput,
   prefetchBaseModel,
+  prefetchHuggingFaceDataset,
   projectRuntimeSchema,
   pythonEnvironmentPath,
   reportInBackground,
@@ -77,8 +89,10 @@ import {
   runGpuProcess,
   runLocalFineTune,
   runLocalPipeline,
+  runProvenanceSchema,
   runReportSchema,
   runRequestFromLocalSpec,
+  safeName,
   sanitizeLogLine,
   specHash,
   specSnapshotSchema,
@@ -94,13 +108,13 @@ import {
   withHuggingFaceCacheEnvironment,
   withOfflineHuggingFaceCacheEnvironment,
   writePrivateJsonAtomic
-} from "./chunk-PVLX7LJI.js";
+} from "./chunk-RPWOCVIL.js";
 
 // src/local-runtime/index.ts
 import { readFileSync as readFileSync3, realpathSync } from "fs";
 import { createHash as createHash2 } from "crypto";
 import { readFile as readFile2, stat } from "fs/promises";
-import { basename, dirname, join as join5, resolve as resolve3 } from "path";
+import { basename, dirname, join as join5, relative, resolve as resolve3 } from "path";
 import { cwd } from "process";
 import { fileURLToPath as fileURLToPath2 } from "url";
 
@@ -932,8 +946,9 @@ Commands:
   validate [tunedtensor.json] [--config local-runner.json]
   run [tunedtensor.json] [--config local-runner.json] [--dry-run] [--verbose] [--quiet]
   serve <model-id|active|base|foundation> [--config local-runner.json] [--host 127.0.0.1] [--port 8000]
-  runs list|get|events|report|compare [args] [--config local-runner.json]
+  runs list|get|events|report|compare|audit [args] [--config local-runner.json]
   models list|get|verify|prefetch|verify-base|active|activate|rollback|serve [args] [--config local-runner.json]
+  datasets prefetch|verify [tunedtensor.json]
 
 Global options:
   -h, --help                       Show help
@@ -1037,6 +1052,7 @@ var COMMAND_GROUPS = {
       get: { usage: "tt runs get <run-id> [--config path]", description: "Get a local run.", options: [CONFIG_OPTION], minPositionals: 1, maxPositionals: 1, missingPositionalsMessage: "runs get requires <run-id>" },
       events: { usage: "tt runs events <run-id> [--config path]", description: "List run events.", options: [CONFIG_OPTION], minPositionals: 1, maxPositionals: 1, missingPositionalsMessage: "runs events requires <run-id>" },
       report: { usage: "tt runs report <run-id> [--config path]", description: "Show the baseline-vs-tuned report, including deltas and regressions.", options: [CONFIG_OPTION], minPositionals: 1, maxPositionals: 1, missingPositionalsMessage: "runs report requires <run-id>" },
+      audit: { usage: "tt runs audit <run-id> [--config path]", description: "Verify the run's event chain and report digest, and show its timeline and provenance.", options: [CONFIG_OPTION], minPositionals: 1, maxPositionals: 1, missingPositionalsMessage: "runs audit requires <run-id>" },
       compare: { usage: "tt runs compare <run-id-a> <run-id-b> [--config path]", description: "Compare two run reports.", options: [CONFIG_OPTION], minPositionals: 2, maxPositionals: 2, missingPositionalsMessage: "runs compare requires <run-id-a> <run-id-b>" }
     }
   },
@@ -1095,6 +1111,24 @@ var COMMAND_GROUPS = {
         missingPositionalsMessage: "models serve requires <model-id>"
       }
     }
+  },
+  datasets: {
+    description: "Download or verify the Hugging Face dataset a behavior spec trains on.",
+    defaultSubcommand: "prefetch",
+    subcommands: {
+      prefetch: {
+        usage: "tt datasets prefetch [tunedtensor.json] [options]",
+        description: "Download dataset_prebuilt.huggingface files into the local Hugging Face cache. Resumes after interruption.",
+        options: [CONFIG_OPTION, VERBOSE_OPTION, QUIET_OPTION],
+        maxPositionals: 1
+      },
+      verify: {
+        usage: "tt datasets verify [tunedtensor.json] [options]",
+        description: "Verify the cached dataset files and checksums without network access.",
+        options: [CONFIG_OPTION, QUIET_OPTION],
+        maxPositionals: 1
+      }
+    }
   }
 };
 function hasHelpFlag(argv) {
@@ -1255,38 +1289,9 @@ async function loadCliBehaviorSpec(inputPath, runId) {
 function printJson(value) {
   console.log(JSON.stringify(value, null, 2));
 }
-function shortValue(value) {
-  if (value === void 0 || value === null) return null;
-  if (typeof value === "string" || typeof value === "number" || typeof value === "boolean") {
-    return String(value);
-  }
-  if (Array.isArray(value)) return `[${value.length} items]`;
-  return null;
-}
-function formatEvent(event) {
-  const detailText = Object.entries(event.details ?? {}).filter(([key]) => key !== "metrics").map(([key, value]) => {
-    const formatted = key === "command" && Array.isArray(value) ? value.join(" ") : shortValue(value);
-    return formatted ? `${key}=${formatted}` : null;
-  }).filter((value) => Boolean(value)).slice(0, 5).join(" ");
-  return sanitizeLogLine(`[tt] ${event.stage}: ${event.message}${detailText ? ` (${detailText})` : ""}`);
-}
 function createConsoleReporter(options) {
   if (options.quiet) return void 0;
-  let lastLogLine = "";
-  return {
-    verbose: options.verbose,
-    onEvent(event) {
-      process.stderr.write(`${formatEvent(event)}
-`);
-    },
-    onLog(log) {
-      const line = sanitizeLogLine(`[tt] ${log.stage}${log.stream ? ` ${log.stream}` : ""}: ${log.message}`);
-      if (line === lastLogLine) return;
-      lastLogLine = line;
-      process.stderr.write(`${line}
-`);
-    }
-  };
+  return createRunConsoleReporter({ verbose: options.verbose });
 }
 async function verifyStoredModel(model, config) {
   const manifestPath = join5(model.artifact_dir, "artifact-manifest.json");
@@ -1805,10 +1810,32 @@ ${validation.errors.join("\n")}`);
     const input = loaded;
     assertLocalRunInputReady(input.request);
     const configSelection = await configSelectionFromArgv(argv, inputPath);
-    const validated = await validateLocalFineTuneInput({
-      request: input.request,
-      config: configSelection.config
-    });
+    const remoteDataset = huggingFaceDataset(input.request);
+    let validated;
+    try {
+      validated = await validateLocalFineTuneInput({
+        request: input.request,
+        config: configSelection.config
+      });
+    } catch (error) {
+      if (!(error instanceof DatasetNotDownloadedError) || !remoteDataset) throw error;
+      const config2 = localRunnerConfigSchema.parse(configSelection.config);
+      printJson({
+        ok: true,
+        input_path: input.path,
+        config_path: configSelection.path ?? null,
+        behavior_spec_id: input.request.behavior_spec_id,
+        base_model: input.request.spec_snapshot.base_model,
+        dataset_format: remoteDataset.format,
+        dataset_status: "not_downloaded",
+        dataset_huggingface: remoteDataset.huggingface,
+        next_step: `tt datasets prefetch ${relative(cwd(), input.path) || input.path}`,
+        artifact_root: config2.artifactRoot,
+        store_root: config2.storeRoot,
+        dry_run: config2.dryRun
+      });
+      return;
+    }
     const request = validated.request;
     const config = validated.config;
     printJson({
@@ -1818,6 +1845,7 @@ ${validation.errors.join("\n")}`);
       behavior_spec_id: request.behavior_spec_id,
       base_model: request.spec_snapshot.base_model,
       dataset_format: request.dataset_prebuilt?.format ?? null,
+      ...request.dataset_source ? { dataset_status: "cached", dataset_source: request.dataset_source } : {},
       artifact_root: config.artifactRoot,
       store_root: config.storeRoot,
       dry_run: config.dryRun
@@ -1834,15 +1862,16 @@ ${validation.errors.join("\n")}`);
       dryRun: hasFlag(argv, "--dry-run") ? true : configInput.dryRun
     });
     assertLocalRunInputReady(input.request);
-    const validated = await validateLocalFineTuneInput({
-      request: input.request,
-      config
-    });
-    let request = validated.request;
     const reporter = createConsoleReporter({
       verbose: hasFlag(argv, "--verbose"),
       quiet: hasFlag(argv, "--quiet")
     });
+    const ensured = await ensureDatasetCached({ request: input.request, config, reporter });
+    const validated = await validateLocalFineTuneInput({
+      request: ensured.request,
+      config
+    });
+    let request = validated.request;
     if (!config.dryRun && !config.paths.baseModel && !request.hyperparameters.base_model_revision) {
       const prefetch = await prefetchBaseModel({
         request,
@@ -1860,12 +1889,13 @@ ${validation.errors.join("\n")}`);
         }
       });
     }
+    const specFile = /^\..+\.tt-local-/.test(basename(input.path)) ? void 0 : { path: input.path, sha256: specHash(await readFile2(input.path, "utf8")) };
     if (input.spec.pipeline) {
-      const result2 = await runLocalPipeline({ request, config, reporter, pipeline: pipelineForRunInput(input), projectSpec: input.spec });
+      const result2 = await runLocalPipeline({ request, config, reporter, pipeline: pipelineForRunInput(input), projectSpec: input.spec, specFile });
       printJson(result2);
       return;
     }
-    const result = await runLocalFineTune({ request, config, reporter });
+    const result = await runLocalFineTune({ request, config, reporter, specFile });
     printJson({
       status: result.report.status,
       run_id: result.report.run_id,
@@ -1913,6 +1943,14 @@ ${validation.errors.join("\n")}`);
       if (!id) throw new Error("runs report requires <run-id>");
       return printJson(await store.getRunReport(id));
     }
+    if (subcommand === "audit") {
+      const id = cli.positionals[0];
+      if (!id) throw new Error("runs audit requires <run-id>");
+      const audit = await auditRun(store, id);
+      printJson(audit);
+      if (audit.verdict === "tampered") process.exitCode = 2;
+      return;
+    }
     if (subcommand === "compare") {
       const idA = cli.positionals[0];
       const idB = cli.positionals[1];
@@ -1925,6 +1963,23 @@ ${validation.errors.join("\n")}`);
     }
     throw new Error(`Unknown runs command: ${subcommand}`);
   }
+  if (command === "datasets") {
+    const subcommand = cli.subcommand;
+    const inputPath = resolve3(cli.positionals[0] ?? DEFAULT_LOCAL_SPEC_PATH);
+    const config = await configFromArgv(argv, inputPath);
+    const input = await loadCliBehaviorSpec(inputPath);
+    if (subcommand !== "prefetch" && subcommand !== "verify") throw new Error(`Unknown datasets command: ${subcommand}`);
+    const report = await prefetchHuggingFaceDataset({
+      request: input.request,
+      config,
+      localOnly: subcommand === "verify",
+      reporter: createConsoleReporter({
+        verbose: hasFlag(argv, "--verbose"),
+        quiet: hasFlag(argv, "--quiet")
+      })
+    });
+    return printJson({ ...report, input_path: input.path });
+  }
   if (command === "models") {
     const subcommand = cli.subcommand;
     const modelInputPath = subcommand === "prefetch" || subcommand === "verify-base" ? resolve3(cli.positionals[0] ?? DEFAULT_LOCAL_SPEC_PATH) : void 0;
@@ -2040,8 +2095,11 @@ export {
   comparisonReportSchema,
   compileSpecToJsonl,
   createLocalStore,
+  datasetColumnsSchema,
   datasetFormatSchema,
+  datasetInputIdentity,
   datasetPrebuiltSchema,
+  datasetSourceSchema,
   decisionHyperparametersSchema,
   decisionLabels,
   decisionPlaceholderIssues,
@@ -2065,10 +2123,13 @@ export {
   getActiveModel,
   getLocalRunnerInfo,
   hardenExistingLocalStore,
+  huggingFaceDatasetSourceSchema,
+  huggingFaceRepoIdSchema,
   initLocalRunnerConfigFile,
   initLocalSpecFile,
   isDecisionSpecFile,
   isFoundationSpecFile,
+  isRepoRelativePath,
   isTerminalRunState,
   jsonFieldMetricsSchema,
   loadJsonFile,
@@ -2092,8 +2153,10 @@ export {
   rollbackActiveModel,
   runLocalFineTune,
   runLocalPipeline,
+  runProvenanceSchema,
   runReportSchema,
   runRequestFromLocalSpec,
+  safeName,
   sanitizeLogLine,
   serveLocalModel,
   specHash,