Evaluation UI: Model Config #8
Description
Is your feature request related to a problem?
The current evaluation config only supports basic options: model name, vector store ID, and system prompt. However, different models
have different configuration parameters—newer models support reasoning and verbosity, while older models use temperature and top_p.
This limits our ability to properly evaluate models with their optimal settings.
Describe the solution you'd like
Add support for model-specific configuration options in the evaluation config:
- Newer models: reasoning, verbosity
- Older models: temperature, top_p, max_tokens
This would:
- Allow evaluations with model-appropriate parameters
- Enable testing different config combinations (e.g., high vs low temperature)