Ecosyste.ms: Timeline
Browse the timeline of events for every public repo on GitHub. Data updated hourly from GH Archive.
perp opened an issue on congvvc/HyperSeg
Demo for inference a single image with prompt?
could you provide a demo for inference a single image with prompt?
perp created a comment on an issue on NexaAI/nexa-sdk
 ok,tokens num and time eplased is as expected
perp created a comment on an issue on NexaAI/nexa-sdk
> Thank you for your feedback, it seems that context buffer allocation is not correctly compatible with the shape of omnivision-968M vision embedding, we will fix it as soon as possible (in 1 or 2 ...
perp created a comment on an issue on NexaAI/nexa-sdk
> I'm not quite sure what your question is. Could you describe the problem more clearly? like tokenpacker, image tokens are compressed ，for example original image tokens num is about 27*27 = 7...
perp opened an issue on NexaAI/nexa-sdk
[QUESTION] <title> Android inference too long
### Question or Issue using llama cpp to inference OmniVision-968M on an android device. it shows that image and text tokens are too many not same as mentioned in hugging face. can someone help?...