Jonas

I somewhat fixed this on my machine by removing all backslashes (\) from the end of my lines in _.zsh_history_

+1, Samme issue here using CUDA/12.1.1, Python/3.10.4-GCCcore-11.3.0, vllm==0.2.3. Happened after 5-10 inferences with a lora fine tuned mistral 7b model [vllm_double_free_bug.log](https://github.com/vllm-project/vllm/files/13902192/vllm_double_free_bug.log) EDIT: In our case the fine-tuned model was trained...

Same here