DAMO-NLP-SG/VideoLLaMA3

View on GitHub

Frontier Multimodal Foundation Models for Image and Video Understanding

Stars: 1.2k
Forks: 89
Open beginner issues: 0
Indexed issues: 69

Dominant language: Jupyter Notebook
License: Apache-2.0
Last GitHub push: Aug 14, 2025
Latest indexed: Sep 15, 2026
Contributing guide: No contributing guide
Code of conduct: No code of conduct
Beginner labels: good first issue
PR merge metrics: No merged PRs in 30d

Open All

69 open issues indexed

when would you write the instuctions about how to do sft on custom data? Open

DAMO-NLP-SG/VideoLLaMA3#1 · 6 comments · 1 reaction · 0 assignees · Indexed Sep 15, 2026
About the relative compression ratio Open

DAMO-NLP-SG/VideoLLaMA3#2 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
What is the max_frames for inference on long videos in the paper? Open

DAMO-NLP-SG/VideoLLaMA3#4 · 2 comments · 1 reaction · 0 assignees · Indexed Sep 15, 2026
Audio-visual video understanding Open

DAMO-NLP-SG/VideoLLaMA3#5 · 2 comments · 3 reactions · 0 assignees · Indexed Sep 15, 2026
The model does not follow the instructions Open

DAMO-NLP-SG/VideoLLaMA3#6 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Steaming video understanding Open

DAMO-NLP-SG/VideoLLaMA3#8 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
the vocab.json has no Chinese word Open

DAMO-NLP-SG/VideoLLaMA3#9 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Question about memory needed to run the various models Open

DAMO-NLP-SG/VideoLLaMA3#13 · 0 comments · 3 reactions · 0 assignees · Indexed Sep 15, 2026
Trainable parameters in each stage Open

good first issue

DAMO-NLP-SG/VideoLLaMA3#15 · 2 comments · 1 reaction · 0 assignees · Indexed Sep 15, 2026
Requesting training docs Open

DAMO-NLP-SG/VideoLLaMA3#18 · 5 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Compute Resources and Training Time Open

DAMO-NLP-SG/VideoLLaMA3#19 · 1 comment · 3 reactions · 0 assignees · Indexed Sep 15, 2026
does it support chinese prompt as input? Open

DAMO-NLP-SG/VideoLLaMA3#21 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Does it support real time video analysis? Open

DAMO-NLP-SG/VideoLLaMA3#22 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Multi-turn Dialogue Open

DAMO-NLP-SG/VideoLLaMA3#23 · 1 comment · 1 reaction · 0 assignees · Indexed Sep 15, 2026
video processing code failed Open

DAMO-NLP-SG/VideoLLaMA3#25 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
training error Open

DAMO-NLP-SG/VideoLLaMA3#26 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Question about Differential Frame Pruner Open

DAMO-NLP-SG/VideoLLaMA3#28 · 3 comments · 1 reaction · 0 assignees · Indexed Sep 15, 2026
Is it possible to perform inference with 2gpu Open

DAMO-NLP-SG/VideoLLaMA3#29 · 4 comments · 0 reactions · 1 assignee · Indexed Sep 15, 2026
Why does the VideoLLaMA3 demo webpage test exhibit output errors in the time granularity of video appearance? Open

DAMO-NLP-SG/VideoLLaMA3#30 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
微调代码 Open

DAMO-NLP-SG/VideoLLaMA3#31 · 6 comments · 2 reactions · 0 assignees · Indexed Sep 15, 2026
How to Perform Inference with Fine-Tuned VideoLLaMA3? Open

DAMO-NLP-SG/VideoLLaMA3#32 · 5 comments · 2 reactions · 0 assignees · Indexed Sep 15, 2026
feature_extraction Open

good first issue

DAMO-NLP-SG/VideoLLaMA3#34 · 1 comment · 4 reactions · 0 assignees · Indexed Sep 15, 2026
Hosting the model Open

DAMO-NLP-SG/VideoLLaMA3#35 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Is "Convert HF checkpoint" part necessary for fintuning on custom dataset? Open

DAMO-NLP-SG/VideoLLaMA3#36 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
cutlass call "implicit_gemm_op.can_implement(arguments)" got error "Error Invalid Problem" at /job_1901480/source/acompute/dnn/cuda/acdnn_cnn_train/kernel_conv_igemm_first_layer_wgrad.cu:245 Open

DAMO-NLP-SG/VideoLLaMA3#37 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
CUDA out of memory Open

DAMO-NLP-SG/VideoLLaMA3#38 · 8 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Video Data Open

DAMO-NLP-SG/VideoLLaMA3#40 · 0 comments · 8 reactions · 0 assignees · Indexed Sep 15, 2026
multi images sft Open

DAMO-NLP-SG/VideoLLaMA3#41 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Get Video Embedding for Video Search Open

DAMO-NLP-SG/VideoLLaMA3#44 · 5 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
can i finetune it with unsloth? Open

DAMO-NLP-SG/VideoLLaMA3#45 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Streaming and video data Open

DAMO-NLP-SG/VideoLLaMA3#47 · 0 comments · 2 reactions · 0 assignees · Indexed Sep 15, 2026
grounding任务的返回坐标格式不固定？ Open

DAMO-NLP-SG/VideoLLaMA3#50 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
How to use video reasoning for video frames？ Open

DAMO-NLP-SG/VideoLLaMA3#51 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
what dose this code do ? Open

DAMO-NLP-SG/VideoLLaMA3#52 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
could you tell me which part of code is implemented for "Any-resolutionVisionTokenization(AVT)" Open

DAMO-NLP-SG/VideoLLaMA3#53 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
How to prepare fine-tuning data for the temporal grounding task? Open

DAMO-NLP-SG/VideoLLaMA3#54 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
The A100-80G graphics card is running the videollama3-7b model, and the error message is that the video memory is overflowing！！！ Open

DAMO-NLP-SG/VideoLLaMA3#56 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
nextqa eval Open

DAMO-NLP-SG/VideoLLaMA3#57 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
output_ids are all zeros Open

DAMO-NLP-SG/VideoLLaMA3#58 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Plan to release ckpts for each training stage Open

DAMO-NLP-SG/VideoLLaMA3#61 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
目前脚本可否支持 Lora 微调 Open

DAMO-NLP-SG/VideoLLaMA3#63 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Which file to run for multi image captioning Open

DAMO-NLP-SG/VideoLLaMA3#64 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Request for Top 5 Logits at Each Output Position Open

DAMO-NLP-SG/VideoLLaMA3#65 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
evaluate the 7b model on charades-sta benchmark Open

DAMO-NLP-SG/VideoLLaMA3#67 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
evaluate the 2b and 7b model on MLVU or longVideoBench Open

DAMO-NLP-SG/VideoLLaMA3#68 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
how could i acclerate the inference on millions of videos? Open

DAMO-NLP-SG/VideoLLaMA3#69 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Training error on stage 1 when following 'bash scripts/train/stage1_2b.sh': AttributeError: 'Linear' object has no attribute 'cal_proj_size' Open

DAMO-NLP-SG/VideoLLaMA3#71 · 5 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
VideoLLaMA3-7B 视频分析质量远低于VideoLLaMA3-2B是什么原因 Open

DAMO-NLP-SG/VideoLLaMA3#73 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
does learning rate affect much for further finetune? Open

DAMO-NLP-SG/VideoLLaMA3#74 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
TypeError: argument of type 'NoneType' is not iterable Open

DAMO-NLP-SG/VideoLLaMA3#75 · 5 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026

DAMO-NLP-SG/VideoLLaMA3

Get new issues in your inbox