@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.js
CHANGED
|
@@ -1706,14 +1706,17 @@ function processOneEvaluation(task, unit, totalEvaluations, publishEvent, persis
|
|
|
1706
1706
|
triggerId: task.triggerId,
|
|
1707
1707
|
runId: evaluatorRunId,
|
|
1708
1708
|
datasetName: task.dataset.getDisplayLabel(),
|
|
1709
|
+
testCaseId: testCaseItem.id,
|
|
1710
|
+
testCaseName: getTestCaseDisplayLabel(testCaseItem.testCase),
|
|
1709
1711
|
repetitionId,
|
|
1710
1712
|
repetitionIndex,
|
|
1711
1713
|
repetitionCount,
|
|
1712
|
-
runConfigName: task.runConfigName
|
|
1714
|
+
runConfigName: task.runConfigName,
|
|
1715
|
+
...task.experimentName !== void 0 && task.experimentName !== "" ? { experimentName: task.experimentName } : {},
|
|
1716
|
+
testCaseTags: getTestCaseTagList(testCaseItem.testCase),
|
|
1717
|
+
runConfigTags: task.runConfigTags,
|
|
1718
|
+
evaluatorTags: getEvaluatorTagList(evaluator)
|
|
1713
1719
|
},
|
|
1714
|
-
testCaseTags: getTestCaseTagList(testCaseItem.testCase),
|
|
1715
|
-
runConfigTags: task.runConfigTags,
|
|
1716
|
-
evaluatorTags: getEvaluatorTagList(evaluator),
|
|
1717
1720
|
logDiff,
|
|
1718
1721
|
log,
|
|
1719
1722
|
createError
|
|
@@ -2192,7 +2195,8 @@ var EffectRunner = class {
|
|
|
2192
2195
|
globalEvaluationSemaphore: sem,
|
|
2193
2196
|
runConfigName: job.runConfigName,
|
|
2194
2197
|
runConfigTags: job.runConfigTags,
|
|
2195
|
-
repetitions: job.repetitions
|
|
2198
|
+
repetitions: job.repetitions,
|
|
2199
|
+
experimentName: request.experimentName
|
|
2196
2200
|
})
|
|
2197
2201
|
);
|
|
2198
2202
|
}
|
|
@@ -2227,7 +2231,8 @@ var EffectRunner = class {
|
|
|
2227
2231
|
maxConcurrency: request.concurrency ?? this.config.maxConcurrency ?? 1,
|
|
2228
2232
|
repetitions: request.repetitions,
|
|
2229
2233
|
runConfigName,
|
|
2230
|
-
runConfigTags: request.runConfigTags
|
|
2234
|
+
runConfigTags: request.runConfigTags,
|
|
2235
|
+
experimentName: request.experimentName
|
|
2231
2236
|
});
|
|
2232
2237
|
}
|
|
2233
2238
|
async startDatasetRun(params) {
|
|
@@ -2302,7 +2307,8 @@ var EffectRunner = class {
|
|
|
2302
2307
|
globalEvaluationSemaphore: params.globalEvaluationSemaphore,
|
|
2303
2308
|
runConfigName: params.runConfigName,
|
|
2304
2309
|
runConfigTags,
|
|
2305
|
-
repetitions
|
|
2310
|
+
repetitions,
|
|
2311
|
+
experimentName: params.experimentName
|
|
2306
2312
|
})
|
|
2307
2313
|
);
|
|
2308
2314
|
return snapshot;
|