vLLM doesn't factually parse any tools?

#17
by jacksloboda - opened

Used the supplied docker image, and exactly these arguments are included:

  --trust_remote_code \
  --max-model-len 131072 \
  --enable-auto-tool-choice \
  --tool-call-parser apriel \
  --reasoning-parser apriel

But it doesn't seem like tools calls get parsed correctly either way. Currently the model is unusable in Zed for instance:

image

Sign up or log in to comment