Why Human_Clothes model cann't output normally

[2016110071]

while I infer my videos on "Human_Clothes" , it can not output normally , just output "!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"
how can I solve it?

[zhengdian1]

Hi, the code runs correctly on our machine. Could you please provide the specific execution commands you used, so that we can better identify the issue?

[2016110071]

Hi, the code runs correctly on our machine. Could you please provide the specific execution commands you used, so that we can better identify the issue?

I run the code without changes, but got the following warnings： Loaded LLaVA model:LLaVA-Video-7B-Qwen2 You are using a model of type llava to instantiate a model of type llava_qwen. This is not supported for all configurations of models and can yield errors. Loading vision tower: siglip-so400m-patch14-384 You are using a model of type siglip_vision_model to instantiate a model of type clip_vision_model. This is not supported for all configurations of models and can yield errors. Some weights of CLIPVisionModel were not initialized from the model checkpoint at siglip-so400m-patch14-384 and are newly initialized: ['vision_model.embeddings.class_embedding', 'vision_model.pre_layrnorm.bias', 'vision_model.pre_layrnorm.weight'] You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference. Some weights of CLIPVisionModel were not initialized from the model checkpoint at siglip-so400m-patch14-384 and are newly initialized because the shapes did not match:

• vision_model.embeddings.position_embedding.weight: found shape torch.Size([729, 1152]) in the checkpoint and torch.Size([730, 1152]) in the model instantiated You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference. 2025-07-30 20:56:28,724 - accelerate.utils.modeling - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set max_memory in to a higher value to use more memory (at your own risk).

and then the output is "!!!!!!!!!!!!!"

[zhengdian1]

It seems that the siglip does not load correctly, please check whether the model is downloaded correctly. The correct information is like following:

Loaded LLaVA model: /mnt/petrelfs/xx/.cache/vbench2/lmms-lab/LLaVA-Video-7B-Qwen2 Loading vision tower: google/siglip-so400m-patch14-384 Loading checkpoint shards: 100%|██████████| 4/4 [00:45<00:00, 11.43s/it] Model Class: LlavaQwenForCausalLM