test it yourself
short token output, but maybe not that good at bug fixing
Hi @rosspanda0 ,
Thank you for reaching out to us . Could you please share a bit more details, such as what you’d like tested, the issue you’re encountering, any relevant code or examples, and the expected vs actual output? We'd be happy to help!
last time in claude code, it fell into a loop trying to fix a simple python coding bug, after cancel the task and clear the session, it solve it. the very charming aspect of this model is that it produce much shorter response than others which make it very suitable for agent using, make the whole interactions more responsive even running on 6 years old gpus.
Same issue. Deployed by vllm. It is very likely to produce dead loop response in claude code.