vik
Can you share more details about your setup? I assume you have to install a node (?) after setting up ComfyUI?
I did not -- did you email it to [email protected]?
What prompt are you using? Which version of the model?
Just created a pull request to add support to vLLM: https://github.com/vllm-project/vllm/pull/4228
Definitely in the roadmap!
You're right on that being the fix - would love a PR!
I'm getting this too, investigating.
Will have a fix for the `ValueError: not enough values to unpack (expected 4, got 3)` bug pushed shortly.
Are you using ollama? The image needs to be passed in at the start of the prompt, I've seen this happen when the image is passed in at the end.
I've experimented with various projection architectures (perceiver-resampler, MAP, C-Abstractor). C-Abstractor worked best, but even it comes with the trade-off of 2-5% drop on the benchmarks we're measuring. It is a...