wire generation params into cfg and backend requests
Add top_p, top_k, min_p, top_a, repetition_penalty, max_completion_tokens, response_format, reasoning_effort, include_reasoning, stop, and verbosity to GenerationParams, config, cfg file, and backend request construction.
This commit is contained in:
parent
1e1e36bf36
commit
6fd52f6fe0
Loading…
Reference in New Issue