-
Notifications
You must be signed in to change notification settings - Fork 1.1k
/
metafile.yml
71 lines (70 loc) · 2.38 KB
/
metafile.yml
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
Collections:
- Name: BLIP-2
Metadata:
Training Data:
- COCO
- VG
- CC3M
- CC12M
- SBU
- LAION-400M
Training Resources: 8x A100 GPUs
Architecture:
- Transformer
- Q-Former
Paper:
Title: 'BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image
Encoders and Large Language Models'
URL: https://arxiv.org/abs/2301.12597
README: configs/blip2/README.md
Models:
- Name: blip2_3rdparty_retrieval
Metadata:
FLOPs: null
Parameters: 1173191358
In Collection: BLIP-2
Results:
- Task: Image-To-Text Retrieval
Dataset: COCO
Metrics:
Recall@1: 85.4
- Task: Text-To-Image Retrieval
Dataset: COCO
Metrics:
Recall@1: 68.3
Weights: https://download.openmmlab.com/mmclassification/v1/blip2/blip2_3rdparty_pretrain_20230505-f7ef4390.pth
Config: configs/blip2/blip2_8xb32_retrieval.py
Converted From:
Weights: https://storage.googleapis.com/sfr-vision-language-research/LAVIS/models/BLIP2/blip2_pretrained_opt2.7b.pth
Code: https://github.com/salesforce/LAVIS
- Name: blip2-opt2.7b_3rdparty-zeroshot_vqa
Metadata:
FLOPs: null
Parameters: 3770465152
In Collection: BLIP-2
Results:
- Task: Visual Question Answering
Dataset: VQAv2
Metrics:
Accuracy: 53.5
Weights: https://download.openmmlab.com/mmclassification/v1/blip2/blip2-opt2.7b_3rdparty_pretrain_20230505-b51db4e1.pth
Config: configs/blip2/blip2-opt2.7b_8xb16_vqa.py
Converted From:
Weights: https://storage.googleapis.com/sfr-vision-language-research/LAVIS/models/BLIP2/blip2_pretrained_opt2.7b.pth
Code: https://github.com/salesforce/LAVIS
- Name: blip2-opt2.7b_3rdparty-zeroshot_caption
Metadata:
FLOPs: null
Parameters: 3770465152
In Collection: BLIP-2
Results:
- Task: Image Caption
Dataset: COCO
Metrics:
BLEU-4: 32.90
CIDER: 111.10
Weights: https://download.openmmlab.com/mmclassification/v1/blip2/blip2-opt2.7b_3rdparty_pretrain_20230505-b51db4e1.pth
Config: configs/blip2/blip2-opt2.7b_8xb32_caption.py
Converted From:
Weights: https://storage.googleapis.com/sfr-vision-language-research/LAVIS/models/BLIP2/blip2_pretrained_opt2.7b.pth
Code: https://github.com/salesforce/LAVIS