excom-ai-demo / app.py

Commit History

Add tool calling support with Hermes parser for Qwen 2.5
a19cb21

Peter Larnholt commited on

Remove --enable-auto-tool-choice flag (Qwen has native tool calling)
80b0386

Peter Larnholt commited on

Enable auto tool choice for better tool calling support
55381a2

Peter Larnholt commited on

Upgrade vLLM to 0.6.4.post1 and remove explicit outlines dependencies
65659d1

Peter Larnholt commited on

Add outlines package and fix Gradio message format compatibility
fa4aba4

Peter Larnholt commited on

Enable verbose logging to diagnose chat completion errors
0e074c5

Peter Larnholt commited on

Fix invalid --disable-guided-decoding flag and add airportsdata dependency
2e9c870

Peter Larnholt commited on

Disable guided decoding to resolve chat completion errors
3356350

Peter Larnholt commited on

Remove guided decoding backend override and add airportsdata dependency
4142581

Peter Larnholt commited on

Switch guided decoding backend to lm-format-enforcer to avoid outlines issues
82c6e6d

Peter Larnholt commited on

Upgrade to vLLM 0.6.3.post1 and remove pyairports workarounds
e48919a

Peter Larnholt commited on

Add sitecustomize.py to patch pyairports module globally
5130472

Peter Larnholt commited on

Use py-airports package instead of pyairports and add debugging
951c827

Peter Larnholt commited on

Pass VLLM_USE_OUTLINES=0 to subprocess and remove outlines dependencies
aaa0398

Peter Larnholt commited on

Update app.py
eaa5923
verified

plarnholt commited on

Update app.py
002f0f2
verified

plarnholt commited on

Initial version
0d3d2c6

Peter Larnholt commited on