dvlab-research LLaMA-VID issues

dvlab-research / LLaMA-VID

Official Implementation for LLaMA-VID: An Image is Worth 2 Tokens in Large Language Models

Apache License 2.0

618 stars 39 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

unable to get results when evaluating on msvd-qa benchmark

#96 irisgong1020 opened 18 hours ago
0
AssertionError: Size mismatch! image_features: 1, prompts: 8

#95 szbcasia opened 1 month ago
0
I was reasoning on the GPU L20(48GB) machine and still burst the video memory

#94 try2020-code opened 1 month ago
0
OOM in stage2 finetuning

#93 Nastu-Ho opened 1 month ago
0
_StoreAction.__init__() got an unexpected keyword argument 'defalut'

#92 try2020-code opened 1 month ago
1
2 tokens in inference

#91 XinyuJiang closed 1 month ago
1
About mm_projector loading issue

#90 rubylan opened 1 month ago
1
[h264 @ 0x871b380] mmco: unref short failure during stage-2 training

#89 Nastu-Ho opened 1 month ago
0
training loss in stage-1

#88 Nastu-Ho opened 2 months ago
0
code details

#87 Nastu-Ho closed 2 months ago
0
Extract context relevancy

#86 IgnacioSan22 opened 2 months ago
0
KeyError: 'LlavaConfig'

#85 skyol99 opened 2 months ago
0
How to resume the checkpoint to continue pretraining？

#84 Einstone-rose opened 2 months ago
0
About the WebVid dataset

#83 szbcasia opened 2 months ago
0
Are all video-based checkpoints trained with 2 tokens?

#82 haodi19 opened 2 months ago
0
HF model format : vlm weights not in llama-vid-7b-full-336

#81 nileshkokane01 opened 2 months ago
0
Questions about Text Decoder and Text Query

#80 SeuXiao opened 3 months ago
0
About the json in stage2 and stage3

#79 liziming5353 opened 3 months ago
1
about the context length for long video

#78 zhuqiangLu opened 3 months ago
0
Confusion in pre-process images for long video

#77 zhuqiangLu closed 2 months ago
0
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cuda:1! (when checking argument for argument weight in method wrapper_CUDA__native_layer_norm)

#76 daocodedao opened 3 months ago
2
About ZERO3

#75 xxtars closed 3 months ago
7
An error occurs during the stage 2 fine-tuning

#74 ShuoZhang2003 opened 3 months ago
1
AttributeError: 'NoneType' object has no attribute 'is_loaded'

#73 sykuann opened 3 months ago
1
why not use LoRA for tunning Vicuna?

#72 dragen1860 closed 2 months ago
1
Multi-image inference

#71 g-h-chen opened 3 months ago
1
Computation costs for each stage?

#70 Becomebright closed 3 months ago
1
Requirements needed for inferring llama-vid llama-vid-13b-full-224-video-fps-1

#69 sykuann opened 3 months ago
1
abnormal outputs for llama-vid-7b-full-224-video-fps-1 ckpt

#68 YulongBonjour opened 3 months ago
1
How to change default path for model_zoo

#67 sykuann opened 3 months ago
2
Questions about the subtitles.

#66 Yxxxb opened 4 months ago
1
flash-attn

#65 ismailukman closed 4 months ago
2
error: llava key

#64 menahem-borges-rodrigues closed 4 months ago
1
About evaluation on vqav2 dataset

#63 liziming5353 opened 4 months ago
1
Long video dataset (only available 167 movies)

#62 KerolosAtef closed 3 months ago
2
Long Video dataset

#61 eslambakr opened 4 months ago
1
Zero-3 offload support

#60 XenonLamb opened 4 months ago
5
Sharing training loss

#59 Deaddawn opened 5 months ago
2
MSVD ACC decrease after stage3

#58 Deaddawn closed 4 months ago
3
The GPU's graphics card usage is also constantly increasing,

#57 kunkunsheng closed 5 months ago
3
is eva_vit_g.pth trained by yourself?

#56 Deaddawn closed 5 months ago
1
why stage 1 and 2 use differenct ` --version plain_guided ` ` --version imgsp_v1 ` parameters?

#55 dragen1860 closed 5 months ago
1
自定义长视频完全跑不了

#54 TotoroDHL closed 4 months ago
1
Enquiry on Download Permission

#53 HenryHZY closed 5 months ago
2
Incomplete evaluation on MSVD-QA dataset.

#52 XenonLamb opened 5 months ago
5
About text encoder

#51 liziming5353 closed 4 months ago
3
Logic error in code: img_in_text and img_token not in sentence["value"]

#50 dragen1860 closed 5 months ago
3
is the LLM weight trainable during stage1-2-3?

#49 dragen1860 closed 5 months ago
1
Long Video CLI wrong

#48 QiSu77 closed 5 months ago
2
why delay_load in build_vision_tower(config, delay_load=True)?

#47 dragen1860 closed 5 months ago
1