-
Notifications
You must be signed in to change notification settings - Fork 1.2k
Open
Labels
bugSomething isn't workingSomething isn't working
Description
System Info
Running with latest llamastack main branch
Information
- The official example scripts
- My own modified scripts
🐛 Describe the bug
The move to OpenAI chat completions API (7e48cc4) broke the requests as it is sending a wong max_output_tokens: Getting the next BadRequestError:
openai.BadRequestError: Error code: 400 - [{'error': {'code': 400, 'message': '* GenerateContentRequest.generation_config.max_output_tokens: max_output_tokens must be positive.\n', 'status': 'INVALID_ARGUMENT'}}]
Error logs
openai.BadRequestError: Error code: 400 - [{'error': {'code': 400, 'message': '* GenerateContentRequest.generation_config.max_output_tokens: max_output_tokens must be positive.\n', 'status': 'INVALID_ARGUMENT'}}]
Expected behavior
If max token is not specified in the llamastack config it should simply pass None and allow the request to go through as before the referenced commit
Metadata
Metadata
Assignees
Labels
bugSomething isn't workingSomething isn't working