Hugging Face's logo

jadechoghari
/
LongVU_Qwen2_7B

Video-Text-to-Text
Transformers
Safetensors
cambrian_qwen
feature-extraction
custom_code
Eval Results
Model card Files Community
LongVU_Qwen2_7B
File size: 4,049 Bytes 
cc3d147
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c076724
cc3d147
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
 
---
datasets:
- shenxq/OneVision
- shenxq/VideoChat2
base_model:
- Vision-CAIR/LongVU_Qwen2_7B_img
pipeline_tag: video-text-to-text
model-index:
- name: llava-onevision-qwen-7b-ov
  results:
  - task:
      type: multimodal
    dataset:
      name: EgoSchema
      type: egoschema
    metrics:
    - type: accuracy
      value: 67.6
      name: accuracy
      verified: true
  - task:
      type: multimodal
    dataset:
      name: MLVU
      type: mlvu
    metrics:
    - type: accuracy
      value: 65.4
      name: accuracy
      verified: true
  - task:
      type: multimodal
    dataset:
      name: MVBench
      type: mvbench
    metrics:
    - type: accuracy
      value: 66.9
      name: accuracy
      verified: true
  - task:
      type: multimodal
    dataset:
      name: VideoMME
      type: videomme
    metrics:
    - type: accuracy
      value: 60.6
      name: accuracy
      verified: true
library_name: transformers
---
# LongVU

This repository contains the model based on Qwen2-7B as presented in [LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding](https://huggingface.co./papers/2410.17434).

Play with the model on the [HF demo](https://huggingface.co./spaces/Vision-CAIR/LongVU).

<div align="left">
    <a href='https://vision-cair.github.io/LongVU'><img src="https://longvu.s3.amazonaws.com/assets/demo.gif" alt="Demo GIF" style="width: 100%; max-width: 650px;"></a>
</div>

# Use

We provide the simple generation process for using our model. For more details, you could refer to [Github](https://github.com/Vision-CAIR/LongVU)

```python
# git clone https://github.com/Vision-CAIR/LongVU
import numpy as np
import torch
from longvu.builder import load_pretrained_model
from longvu.constants import (
    DEFAULT_IMAGE_TOKEN,
    IMAGE_TOKEN_INDEX,
)
from longvu.conversation import conv_templates, SeparatorStyle
from longvu.mm_datautils import (
    KeywordsStoppingCriteria,
    process_images,
    tokenizer_image_token,
)
from decord import cpu, VideoReader

tokenizer, model, image_processor, context_len = load_pretrained_model(
    "./checkpoints/longvu_qwen", None, "cambrian_qwen",
)

model.eval()
video_path = "./examples/video1.mp4"
qs = "Describe this video in detail"

vr = VideoReader(video_path, ctx=cpu(0), num_threads=1)
fps = float(vr.get_avg_fps())
frame_indices = np.array([i for i in range(0, len(vr), round(fps),)])
video = []
for frame_index in frame_indices:
    img = vr[frame_index].asnumpy()
    video.append(img)
video = np.stack(video)
image_sizes = [video[0].shape[:2]]
video = process_images(video, image_processor, model.config)
video = [item.unsqueeze(0) for item in video]

qs = DEFAULT_IMAGE_TOKEN + "\n" + qs
conv = conv_templates["qwen"].copy()
conv.append_message(conv.roles[0], qs)
conv.append_message(conv.roles[1], None)
prompt = conv.get_prompt()

input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors="pt").unsqueeze(0).to(model.device)
stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2
keywords = [stop_str]
stopping_criteria = KeywordsStoppingCriteria(keywords, tokenizer, input_ids)
with torch.inference_mode():
    output_ids = model.generate(
        input_ids,
        images=video,
        image_sizes=image_sizes,
        do_sample=False,
        temperature=0.2,
        max_new_tokens=128,
        use_cache=True,
        stopping_criteria=[stopping_criteria],
    )
pred = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip()
```

# Citation

```
@article{shen2024longvu,
    title={LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding},
    author={Shen, Xiaoqian and Xiong, Yunyang and Zhao, Changsheng and Wu, Lemeng and Chen, Jun and Zhu, Chenchen and Liu, Zechun and Xiao, Fanyi and Varadarajan, Balakrishnan and Bordes, Florian and Liu, Zhuang and Xu, Hu and J. Kim, Hyunwoo and Soran, Bilge and Krishnamoorthi, Raghuraman and Elhoseiny, Mohamed and Chandra, Vikas},
    journal={arXiv:2410.17434},
    year={2024}
  }
```