Track the completion usage.
(
self,
completion: Union[ChatCompletion, Generator[ChatCompletionChunk, None, None]],
)
| 269 | raise |
| 270 | |
| 271 | def track_completion_usage( |
| 272 | self, |
| 273 | completion: Union[ChatCompletion, Generator[ChatCompletionChunk, None, None]], |
| 274 | ) -> CompletionUsage: |
| 275 | """Track the completion usage.""" |
| 276 | if isinstance(completion, ChatCompletion): |
| 277 | return CompletionUsage( |
| 278 | completion_tokens=completion.usage.completion_tokens, |
| 279 | prompt_tokens=completion.usage.prompt_tokens, |
| 280 | total_tokens=completion.usage.total_tokens, |
| 281 | ) |
| 282 | else: |
| 283 | # For streaming, we can't track usage accurately |
| 284 | return CompletionUsage(completion_tokens=0, prompt_tokens=0, total_tokens=0) |
| 285 | |
| 286 | def parse_embedding_response( |
| 287 | self, response: CreateEmbeddingResponse |
nothing calls this directly
no outgoing calls
no test coverage detected