agentjido / req_llm / 45c8cd05e603f622a2f4a3f81b2d53d4e5abfd76

49%

Build Type

push

github

Committed by web-flow

Commit Message

Add Cerebras provider implementation (#78)

* feat(provider/cerebras): add Cerebras provider implementation and associated fixtures

- Introduced a new provider module for Cerebras, implementing OpenAI-compatible Chat Completions API with specific adjustments.
- Updated usage calculations to include cached tokens and reasoning tokens in the provider's response structure.
- Added comprehensive test coverage for the Cerebras provider, including various fixture files for different model scenarios.
- Enhanced the supported models list to include new Cerebras models with their respective statuses and last checked timestamps.
- Refactored existing usage parsing methods to accommodate the new structure and ensure consistency across providers.

* quality and tests

* Deps

* feat(provider/cerebras): update Cerebras provider implementation and fixtures

- Removed the `decode_response` function from the Cerebras provider, streamlining response handling.
- Updated model configurations in `priv/models_dev/cerebras.json` to enable reasoning capabilities for the GPT OSS 120B model.
- Adjusted various fixture files to reflect changes in token limits and response structures, including updates to `max_tokens` for multiple scenarios.
- Added new reasoning-related fixtures to enhance testing coverage for reasoning capabilities in the Cerebras provider.
- Updated timestamps in supported models and fixture files for consistency.

Run Details

4 of 29 new or added lines in 3 files covered. (13.79%)

6 existing lines in 1 file now uncovered.

2671 of 5302 relevant lines covered (50.38%)

184.6 hits per line