@wix/evalforge-evaluator 0.207.0 → 0.209.0
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/build/index.js +19 -10
- package/build/index.js.map +2 -2
- package/build/index.mjs +19 -10
- package/build/index.mjs.map +2 -2
- package/build/types/run-scenario/agents/simple-agent/cost-calculation.d.ts +7 -2
- package/build/types/run-scenario/agents/simple-agent/execute.d.ts +1 -5
- package/package.json +2 -2
package/build/index.js
CHANGED
|
@@ -10966,7 +10966,14 @@ function calculateFromPricing(modelId, tokenUsage) {
|
|
|
10966
10966
|
return tokenUsage.prompt / 1e6 * pricing.input + tokenUsage.completion / 1e6 * pricing.output;
|
|
10967
10967
|
}
|
|
10968
10968
|
function calculateStepCost(step, modelId, provider, tokenUsage) {
|
|
10969
|
-
|
|
10969
|
+
const gatewayCost = extractGatewayCost(step, provider);
|
|
10970
|
+
if (gatewayCost !== void 0) {
|
|
10971
|
+
return gatewayCost;
|
|
10972
|
+
}
|
|
10973
|
+
console.warn(
|
|
10974
|
+
`[cost-calculation] gateway cost missing; using list-price fallback (provider=${provider}, model=${modelId}, promptTokens=${tokenUsage.prompt}, completionTokens=${tokenUsage.completion})`
|
|
10975
|
+
);
|
|
10976
|
+
return calculateFromPricing(modelId, tokenUsage);
|
|
10970
10977
|
}
|
|
10971
10978
|
|
|
10972
10979
|
// src/run-scenario/agents/simple-agent/build-conversation.ts
|
|
@@ -11246,13 +11253,13 @@ async function executeWithAiSdk(context) {
|
|
|
11246
11253
|
[text, steps, rawUsage] = await Promise.all([
|
|
11247
11254
|
stream.text,
|
|
11248
11255
|
stream.steps,
|
|
11249
|
-
stream.
|
|
11256
|
+
stream.totalUsage
|
|
11250
11257
|
]);
|
|
11251
11258
|
} else {
|
|
11252
11259
|
const result = await (0, import_ai.generateText)(sdkParams);
|
|
11253
11260
|
text = result.text;
|
|
11254
11261
|
steps = result.steps;
|
|
11255
|
-
rawUsage = result.
|
|
11262
|
+
rawUsage = result.totalUsage;
|
|
11256
11263
|
}
|
|
11257
11264
|
const durationMs = Date.now() - startTime;
|
|
11258
11265
|
const usage = {
|
|
@@ -11263,7 +11270,6 @@ async function executeWithAiSdk(context) {
|
|
|
11263
11270
|
const llmTrace = buildLLMTrace2(
|
|
11264
11271
|
steps,
|
|
11265
11272
|
durationMs,
|
|
11266
|
-
usage,
|
|
11267
11273
|
modelId,
|
|
11268
11274
|
provider,
|
|
11269
11275
|
startTime,
|
|
@@ -11412,7 +11418,7 @@ function findToolResultError(step) {
|
|
|
11412
11418
|
}
|
|
11413
11419
|
return null;
|
|
11414
11420
|
}
|
|
11415
|
-
function buildLLMTrace2(steps, totalDurationMs,
|
|
11421
|
+
function buildLLMTrace2(steps, totalDurationMs, modelId, provider, executionStartMs, stepTimestamps) {
|
|
11416
11422
|
const traceSteps = steps.map((step, i) => {
|
|
11417
11423
|
const stepFinishedAt = stepTimestamps[i] ?? executionStartMs;
|
|
11418
11424
|
const stepStartedAt = i === 0 ? executionStartMs : stepTimestamps[i - 1] ?? executionStartMs;
|
|
@@ -11444,11 +11450,14 @@ function buildLLMTrace2(steps, totalDurationMs, totalUsage, modelId, provider, e
|
|
|
11444
11450
|
};
|
|
11445
11451
|
});
|
|
11446
11452
|
const totalCostUsd = traceSteps.reduce((sum, s) => sum + s.costUsd, 0);
|
|
11447
|
-
const finalTokens =
|
|
11448
|
-
|
|
11449
|
-
|
|
11450
|
-
|
|
11451
|
-
|
|
11453
|
+
const finalTokens = traceSteps.reduce(
|
|
11454
|
+
(acc, s) => ({
|
|
11455
|
+
prompt: acc.prompt + s.tokenUsage.prompt,
|
|
11456
|
+
completion: acc.completion + s.tokenUsage.completion,
|
|
11457
|
+
total: acc.total + s.tokenUsage.total
|
|
11458
|
+
}),
|
|
11459
|
+
{ prompt: 0, completion: 0, total: 0 }
|
|
11460
|
+
);
|
|
11452
11461
|
return {
|
|
11453
11462
|
id: (0, import_crypto4.randomUUID)(),
|
|
11454
11463
|
steps: traceSteps,
|