Does vllm 0.7.3 support this model？
#10
by
traphix
- opened
Does vllm 0.7.3 support this model？
That's an old question, but 0.8.5 does support it, and the latest source 0.10.0rc2 now works correctly with it if you edit the config.json file to remove the "dual_chunk_attention_config" parameter.
Best of luck, and remember to close this issue!
traphix
changed discussion status to
closed