Video-LLaMA icon indicating copy to clipboard operation
Video-LLaMA copied to clipboard

demo运行

Open luyao-cv opened this issue 1 year ago • 9 comments

9fb057dd904c82588aa15dbf6996a441 跑了两个demo.py,用的给的例子,都会出现这样的错误

luyao-cv avatar Aug 06 '23 05:08 luyao-cv

哈喽,是不是没升级到最新的代码?刚我这边测了一下最新代码没问题,可否把详细地配置信息和报错信息报一下,在启动demo的终端里面应该有

hangzhang-nlp avatar Aug 07 '23 02:08 hangzhang-nlp

python demo_audiovideo.py --cfg-path eval_configs/video_llama_eval_withaudio.yaml --model_type llama_v2 --gpu-id 0 运行demo的时候失败了,配置是下面这样的,哪里出问题了吗? llama_model: "/group/30155/jasoncjxcai/Video-LLaMA/Video-LLaMA-2-7B-Finetuned/llama-2-7b-chat-hf" imagebind_ckpt_path: "/group/30155/jasoncjxcai/Video-LLaMA/Video-LLaMA-2-7B-Finetuned/imagebind_huge.pth" ckpt: '/group/30155/jasoncjxcai/Video-LLaMA/Video-LLaMA-2-7B-Finetuned/VL_LLaMA_2_7B_Finetuned.pth' # you can use our pretrained ckpt from https://huggingface.co/DAMO-NLP-SG/Video-LLaMA-2-13B-Pretrained/ ckpt_2: '/group/30155/jasoncjxcai/Video-LLaMA/Video-LLaMA-2-7B-Finetuned/AL_LLaMA_2_7B_Finetuned.pth' image

CaicaiJason avatar Aug 10 '23 03:08 CaicaiJason

python demo_audiovideo.py --cfg-path eval_configs/video_llama_eval_withaudio.yaml --model_type llama_v2 --gpu-id 0 运行demo的时候失败了,配置是下面这样的,哪里出问题了吗? llama_model: "/group/30155/jasoncjxcai/Video-LLaMA/Video-LLaMA-2-7B-Finetuned/llama-2-7b-chat-hf" imagebind_ckpt_path: "/group/30155/jasoncjxcai/Video-LLaMA/Video-LLaMA-2-7B-Finetuned/imagebind_huge.pth" ckpt: '/group/30155/jasoncjxcai/Video-LLaMA/Video-LLaMA-2-7B-Finetuned/VL_LLaMA_2_7B_Finetuned.pth' # you can use our pretrained ckpt from https://huggingface.co/DAMO-NLP-SG/Video-LLaMA-2-13B-Pretrained/ ckpt_2: '/group/30155/jasoncjxcai/Video-LLaMA/Video-LLaMA-2-7B-Finetuned/AL_LLaMA_2_7B_Finetuned.pth' image

看起来是没有自动下载bert-base-uncased,应该是huggingface transformer的bug。

hangzhang-nlp avatar Aug 11 '23 08:08 hangzhang-nlp

image

Dylandtt avatar Sep 13 '23 08:09 Dylandtt

视频一直上传不上去

Dylandtt avatar Sep 13 '23 08:09 Dylandtt

你好,llama13B版本在推理时只能使用单张A100(或其他大显存机器),请问如果没有A100,如何使用多张例如4张v100代替?

flying2023 avatar Sep 27 '23 15:09 flying2023

视频传不上去能如何解决吗?

jqtangust avatar Oct 16 '23 15:10 jqtangust

你好,llama13B版本在推理时只能使用单张A100(或其他大显存机器),请问如果没有A100,如何使用多张例如4张v100代替?

加载模型的device改成auto就可以两张v100共同部署测试了device_map = 'auto',不过这个模型真的太重了,两个qformer,加vit加vicuna

xmy0916 avatar Oct 17 '23 06:10 xmy0916

视频一直上传不上去

老哥解决了吗,我也卡在这了

pzqez avatar Jan 18 '24 07:01 pzqez