Please fill this field
Please fill this field
openai/clip-vit-large-patch14-336 laion/CLIP-convnext_large_d_320.laion2B-s29B-b131K-ft-soup
how can we run the llama 3 70b ? I used your own pre-converted repo and it cannot find .pth file
it works ! thanks, fyi, im getting 39 tks/s on 4xH100.
please do, I know you do not owe the open source community anything but you have claimed to os 'soon' the code of training for llava-next since january. Llava 1.6...
uncomment concurrency_limit and upgrade gradio and it should work just fine. however im running into NameError: name 'PaddleOCR' is not defined, while made sure to install paddleocr as documented: `python3...
 This is the result of the finetune at the end of the training run, only 0 to 4 model.safetensors are saved whilst the original model_base which is your instruction...
[model.safetensors.index.json](https://github.com/dvlab-research/MiniGemini/files/14920819/model.safetensors.index.json) Please have a look, the mixtral model should normally have 20 safetensors.
@yanwei-li please help, thank you.