YanweiLi commited on
Commit
3ce9a9e
1 Parent(s): 64b201f

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +46 -0
README.md ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - vision-language model
4
+ - llama
5
+ - video understanding
6
+ ---
7
+
8
+
9
+ # LLaMA-VID Model Card
10
+ <a href='https://llama-vid.github.io/'><img src='https://img.shields.io/badge/Project-Page-Green'></a>
11
+ <a href='https://arxiv.org/abs/2311.17043'><img src='https://img.shields.io/badge/Paper-Arxiv-red'></a>
12
+
13
+ ## Model details
14
+ LLaMA-VID empowers existing frameworks to support hour-long videos and pushes their upper limit with an extra context token.
15
+
16
+
17
+ **Model type:**
18
+ LLaMA-VID is an open-source chatbot trained by fine-tuning LLaMA/Vicuna on GPT-generated multimodal instruction-following data.
19
+ LLaMA-VID empowers existing frameworks to support hour-long videos and pushes their upper limit with an extra context token. We build this repo based on LLaVA.
20
+
21
+
22
+ **Model date:**
23
+ llama-vid-13b-full-224-video-fps-1 was trained on 11/2023.
24
+
25
+ ## License
26
+ Llama 2 is licensed under the LLAMA 2 Community License,
27
+ Copyright (c) Meta Platforms, Inc. All Rights Reserved.
28
+
29
+ **Where to send questions or comments about the model:**
30
+ https://github.com/dvlab-research/LLaMA-VID/issues
31
+
32
+ ## Intended use
33
+ **Primary intended uses:**
34
+ The primary use of LLaMA-VID is research on large multimodal models and chatbots.
35
+
36
+ **Primary intended users:**
37
+ The primary intended users of the model are researchers and hobbyists in computer vision, natural language processing, machine learning, and artificial intelligence.
38
+
39
+ ## Training data
40
+ This model is trained based on image data from LLaVA-1.5 dataset, and video data from WebVid and ActivityNet datasets, including
41
+ - 558K filtered image-text pairs from LAION/CC/SBU, captioned by BLIP.
42
+ - 158K GPT-generated multimodal instruction-following data.
43
+ - 450K academic-task-oriented VQA data mixture.
44
+ - 40K ShareGPT data.
45
+ - 232K video-caption pairs sampled from the WebVid 2.5M dataset.
46
+ - 98K videos from ActivityNet with QA pairs from Video-ChatGPT.