@m4trix/evals 0.27.0 → 0.29.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +5 -2
- package/dist/cli-simple.cjs +38 -15
- package/dist/cli-simple.cjs.map +1 -1
- package/dist/cli-simple.js +38 -15
- package/dist/cli-simple.js.map +1 -1
- package/dist/cli.cjs +13 -7
- package/dist/cli.cjs.map +1 -1
- package/dist/cli.js +13 -7
- package/dist/cli.js.map +1 -1
- package/dist/index.cjs +17 -10
- package/dist/index.cjs.map +1 -1
- package/dist/index.d.ts +39 -14
- package/dist/index.js +17 -10
- package/dist/index.js.map +1 -1
- package/package.json +1 -1
package/dist/cli.cjs
CHANGED
|
@@ -1733,14 +1733,17 @@ function processOneEvaluation(task, unit, totalEvaluations, publishEvent, persis
|
|
|
1733
1733
|
triggerId: task.triggerId,
|
|
1734
1734
|
runId: evaluatorRunId,
|
|
1735
1735
|
datasetName: task.dataset.getDisplayLabel(),
|
|
1736
|
+
testCaseId: testCaseItem.id,
|
|
1737
|
+
testCaseName: getTestCaseDisplayLabel(testCaseItem.testCase),
|
|
1736
1738
|
repetitionId,
|
|
1737
1739
|
repetitionIndex,
|
|
1738
1740
|
repetitionCount,
|
|
1739
|
-
runConfigName: task.runConfigName
|
|
1741
|
+
runConfigName: task.runConfigName,
|
|
1742
|
+
...task.experimentName !== void 0 && task.experimentName !== "" ? { experimentName: task.experimentName } : {},
|
|
1743
|
+
testCaseTags: getTestCaseTagList(testCaseItem.testCase),
|
|
1744
|
+
runConfigTags: task.runConfigTags,
|
|
1745
|
+
evaluatorTags: getEvaluatorTagList(evaluator)
|
|
1740
1746
|
},
|
|
1741
|
-
testCaseTags: getTestCaseTagList(testCaseItem.testCase),
|
|
1742
|
-
runConfigTags: task.runConfigTags,
|
|
1743
|
-
evaluatorTags: getEvaluatorTagList(evaluator),
|
|
1744
1747
|
logDiff,
|
|
1745
1748
|
log,
|
|
1746
1749
|
createError
|
|
@@ -2219,7 +2222,8 @@ var EffectRunner = class {
|
|
|
2219
2222
|
globalEvaluationSemaphore: sem,
|
|
2220
2223
|
runConfigName: job.runConfigName,
|
|
2221
2224
|
runConfigTags: job.runConfigTags,
|
|
2222
|
-
repetitions: job.repetitions
|
|
2225
|
+
repetitions: job.repetitions,
|
|
2226
|
+
experimentName: request.experimentName
|
|
2223
2227
|
})
|
|
2224
2228
|
);
|
|
2225
2229
|
}
|
|
@@ -2254,7 +2258,8 @@ var EffectRunner = class {
|
|
|
2254
2258
|
maxConcurrency: request.concurrency ?? this.config.maxConcurrency ?? 1,
|
|
2255
2259
|
repetitions: request.repetitions,
|
|
2256
2260
|
runConfigName,
|
|
2257
|
-
runConfigTags: request.runConfigTags
|
|
2261
|
+
runConfigTags: request.runConfigTags,
|
|
2262
|
+
experimentName: request.experimentName
|
|
2258
2263
|
});
|
|
2259
2264
|
}
|
|
2260
2265
|
async startDatasetRun(params) {
|
|
@@ -2329,7 +2334,8 @@ var EffectRunner = class {
|
|
|
2329
2334
|
globalEvaluationSemaphore: params.globalEvaluationSemaphore,
|
|
2330
2335
|
runConfigName: params.runConfigName,
|
|
2331
2336
|
runConfigTags,
|
|
2332
|
-
repetitions
|
|
2337
|
+
repetitions,
|
|
2338
|
+
experimentName: params.experimentName
|
|
2333
2339
|
})
|
|
2334
2340
|
);
|
|
2335
2341
|
return snapshot;
|