a2c77f4932
The auto-fetch of top search results added ~2s latency per search without meaningfully improving results for small models that kept searching the wrong pages. Searches now return snippets only (fast) with a hint telling the model it can fetch any URL explicitly via the url parameter. Direct URL fetch remains available for when the model finds a relevant link and wants the full content. |
||
|---|---|---|
| .. | ||
| __init__.py | ||
| audio_codecs.py | ||
| defaults.py | ||
| inference.py | ||
| llama_cpp.py | ||
| orchestrator.py | ||
| tools.py | ||
| worker.py | ||