dvlab-research / LLaMA-VID

LLaMA-VID: An Image is Worth 2 Tokens in Large Language Models (ECCV 2024)
Apache License 2.0
693 stars 43 forks source link

AttributeError: 'NoneType' object has no attribute 'is_loaded' #73

Open sykuann opened 6 months ago

sykuann commented 6 months ago

File "/mnt/llmdownloads/LLaMA-VID/llamavid/model/builder.py", line 92, in load_pretrained_model if not vision_tower.is_loaded: AttributeError: 'NoneType' object has no attribute 'is_loaded'

python -m llamavid.serve.cli --model-path /mnt/llmdownloads/LLaMA-VID/llamavid/work_dirs/llama-vid-7b-full-224 --image-file /mnt/llmdownloads/LLaMA-VID/AI_crypto_trade_scam.mp4 --temperature 0.5

Error arises when run code.

yanwei-li commented 6 months ago

Hi, it seems this issue could casued by this line, please check this and make sure the CLIPVisionTower in llamavid/model/multimodal_encoder/clip_encoder.py is normally loaded.