bokesyo commited on
Commit
3081d81
1 Parent(s): 0297db8

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +16 -5
README.md CHANGED
@@ -18,23 +18,34 @@ With MiniCPM-Visual-Embedding, it is possible to directly build knowledge base w
18
 
19
  # News
20
 
21
- - 2024-06-27: We released our first visual embedding model minicpm-visual-embedding-v0.1 on [huggingface](https://huggingface.co/RhapsodyAI/minicpm-visual-embedding-v0.1).
22
 
23
  - 2024-05-08: We [committed](https://github.com/bokesyo/minicpm-visual-embedding) our training code (full-parameter tuning with GradCache and DeepSpeed, supports large batch size across multiple GPUs with zero-stage1) and eval code.
24
 
25
  # Get started
26
 
 
 
 
 
 
 
 
 
 
 
 
27
  First you are suggested to git clone this huggingface repo or download repo with `huggingface_cli`.
28
 
29
  ```bash
30
  git lfs install
31
- git clone https://huggingface.co/RhapsodyAI/minicpm-visual-embedding-v0.1
32
  ```
33
 
34
  or
35
 
36
  ```bash
37
- huggingface-cli download RhapsodyAI/minicpm-visual-embedding-v0.1
38
  ```
39
 
40
  ```python
@@ -56,8 +67,8 @@ def last_token_pool(last_hidden_states: Tensor,
56
  return last_hidden_states[torch.arange(batch_size, device=last_hidden_states.device), sequence_lengths]
57
 
58
 
59
- tokenizer = AutoTokenizer.from_pretrained('/local/path/to/minicpm-visual-embedding-v0.1')
60
- model = AutoModel.from_pretrained('/local/path/to/minicpm-visual-embedding-v0.1')
61
 
62
  image_1 = Image.open('/local/path/to/document1.png').convert('RGB')
63
  image_2 = Image.open('/local/path/to/document2.png').convert('RGB')
 
18
 
19
  # News
20
 
21
+ - 2024-06-27: We released our first visual embedding model minicpm-visual-embedding-v0.1 on [huggingface](https://huggingface.co/RhapsodyAI/minicpm-visual-embedding-v0).
22
 
23
  - 2024-05-08: We [committed](https://github.com/bokesyo/minicpm-visual-embedding) our training code (full-parameter tuning with GradCache and DeepSpeed, supports large batch size across multiple GPUs with zero-stage1) and eval code.
24
 
25
  # Get started
26
 
27
+ Pip install all dependencies:
28
+
29
+ ```
30
+ Pillow==10.1.0
31
+ timm==0.9.10
32
+ torch==2.1.2
33
+ torchvision==0.16.2
34
+ transformers==4.36.0
35
+ sentencepiece==0.1.99
36
+ ```
37
+
38
  First you are suggested to git clone this huggingface repo or download repo with `huggingface_cli`.
39
 
40
  ```bash
41
  git lfs install
42
+ git clone https://huggingface.co/RhapsodyAI/minicpm-visual-embedding-v0
43
  ```
44
 
45
  or
46
 
47
  ```bash
48
+ huggingface-cli download RhapsodyAI/minicpm-visual-embedding-v0
49
  ```
50
 
51
  ```python
 
67
  return last_hidden_states[torch.arange(batch_size, device=last_hidden_states.device), sequence_lengths]
68
 
69
 
70
+ tokenizer = AutoTokenizer.from_pretrained('/local/path/to/minicpm-visual-embedding-v0', trust_remote_code=True)
71
+ model = AutoModel.from_pretrained('/local/path/to/minicpm-visual-embedding-v0', trust_remote_code=True)
72
 
73
  image_1 = Image.open('/local/path/to/document1.png').convert('RGB')
74
  image_2 = Image.open('/local/path/to/document2.png').convert('RGB')