Add model name to completions telemetry #3689
Conversation
There was a problem hiding this comment.
Pull request overview
This PR adds the active model identifier to the “provideInlineEdit” (NES/completions) telemetry payload so completions telemetry can be correlated with the model used.
Changes:
- Extend
LlmNESTelemetryBuilderto optionally include amodelNameproperty in its built telemetry object. - Populate
modelNamefor completions by wiringengineInfo.modelIdinto the telemetry builder during ghost text computation.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
| src/extension/inlineEdits/node/nextEditProviderTelemetry.ts | Adds modelName support to LlmNESTelemetryBuilder so it can be emitted in the telemetry payload. |
| src/extension/completions-core/vscode-node/lib/src/ghostText/ghostText.ts | Sets telemetry builder model name from the selected completion engine model id. |
| const engineInfo = this.instantiationService.invokeFunction(getEngineRequestInfo, preIssuedTelemetryDataWithExp); | ||
| telemetryBuilder.setModelName(engineInfo.modelId); | ||
| const ghostTextOptions = { ...defaultOptions, ...options, tokenizer: engineInfo.tokenizer }; |
There was a problem hiding this comment.
New telemetry behavior (populating modelName via telemetryBuilder.setModelName(engineInfo.modelId)) isn’t covered by a unit test. Since this value ultimately affects the provideInlineEdit / copilot-nes/provideInlineEdit events, please add/extend a test (e.g., in ghostText.test.ts or a provider-level test using the in-memory telemetry spy) to assert modelName is present in the emitted telemetry for completions.
1 participant
No description provided.