@m4trix/evals 0.16.0 → 0.17.0
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/dist/cli-simple.cjs +8 -0
- package/dist/cli-simple.cjs.map +1 -1
- package/dist/cli-simple.js +8 -0
- package/dist/cli-simple.js.map +1 -1
- package/dist/cli.cjs +8 -0
- package/dist/cli.cjs.map +1 -1
- package/dist/cli.js +8 -0
- package/dist/cli.js.map +1 -1
- package/dist/index.cjs +8 -0
- package/dist/index.cjs.map +1 -1
- package/dist/index.d.ts +19 -1
- package/dist/index.js +8 -0
- package/dist/index.js.map +1 -1
- package/package.json +1 -1
package/dist/cli.js
CHANGED
|
@@ -1261,6 +1261,7 @@ function processOneTestCase(task, testCaseItem, totalEvaluations, publishEvent,
|
|
|
1261
1261
|
const reruns = typeof testCaseItem.testCase.getReruns === "function" ? testCaseItem.testCase.getReruns() : 1;
|
|
1262
1262
|
const rerunPassed = [];
|
|
1263
1263
|
for (let r = 0; r < reruns; r++) {
|
|
1264
|
+
const evaluatorRunId = `run-${randomUUID()}`;
|
|
1264
1265
|
const started = Date.now();
|
|
1265
1266
|
const evaluatorScores = [];
|
|
1266
1267
|
let testCaseError;
|
|
@@ -1287,6 +1288,11 @@ function processOneTestCase(task, testCaseItem, totalEvaluations, publishEvent,
|
|
|
1287
1288
|
input: testCaseItem.testCase.getInput(),
|
|
1288
1289
|
ctx,
|
|
1289
1290
|
output,
|
|
1291
|
+
meta: {
|
|
1292
|
+
triggerId: task.triggerId,
|
|
1293
|
+
runId: evaluatorRunId,
|
|
1294
|
+
datasetId: task.datasetId
|
|
1295
|
+
},
|
|
1290
1296
|
logDiff,
|
|
1291
1297
|
log
|
|
1292
1298
|
})
|
|
@@ -1776,6 +1782,7 @@ var EffectRunner = class {
|
|
|
1776
1782
|
(sum, tc) => sum + (typeof tc.testCase.getReruns === "function" ? tc.testCase.getReruns() : 1),
|
|
1777
1783
|
0
|
|
1778
1784
|
);
|
|
1785
|
+
const triggerId = request.triggerId ?? `trg-${randomUUID()}`;
|
|
1779
1786
|
const runId = `run-${randomUUID()}`;
|
|
1780
1787
|
const artifactPath = createArtifactPath(
|
|
1781
1788
|
this.config.artifactDirectory,
|
|
@@ -1817,6 +1824,7 @@ var EffectRunner = class {
|
|
|
1817
1824
|
await Effect.runPromise(
|
|
1818
1825
|
Queue.offer(this.runQueue, {
|
|
1819
1826
|
runId,
|
|
1827
|
+
triggerId,
|
|
1820
1828
|
datasetId: request.datasetId,
|
|
1821
1829
|
dataset: dataset.dataset,
|
|
1822
1830
|
evaluators: selectedEvaluators,
|