Hugging Face's logo

yuanzhoulvpi
/
llava_qwen15-4b-chat_openai-clip-vit-large-patch14-336

Image-Text-to-Text
Transformers
Safetensors
Chinese
English
llava
conversational
Inference Endpoints
Model card Files Community
1
llava_qwen15-4b-chat_openai-clip-vit-large-patch14-336
File size: 2,718 Bytes 
baf6b33
 
 
 
 
 
 
4514bcd
 
 
 
 
 
 
 
 
 
 
 
baf6b33
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
 
---
license: apache-2.0
language:
- zh
- en
pipeline_tag: image-text-to-text
---
## cite this model

```bash
@misc {yuanz_2024,
	author       = { {yuanz} },
	title        = { llava_qwen15-4b-chat_openai-clip-vit-large-patch14-336 (Revision 5070a27) },
	year         = 2024,
	url          = { https://huggingface.co/yuanzhoulvpi/llava_qwen15-4b-chat_openai-clip-vit-large-patch14-336 },
	doi          = { 10.57967/hf/3146 },
	publisher    = { Hugging Face }
}
```

# 从0到1训练一个定制版的llava模型
1. 基于openai/clip-vit-large-patch14-336 和Qwen1.5-4B-Chat模型，构建一个llava模型
2. 使用数据liuhaotian/LLaVA-CC3M-Pretrain-595K
3. 训练方式是deepspeed-zero2、lora进行微调。



# 关联的github
1. [https://github.com/yuanzhoulvpi2017/zero_nlp/tree/main/train_llava](https://github.com/yuanzhoulvpi2017/zero_nlp/tree/main/train_llava)



# 关联的b站学习视频

1. 待填充



# 推理代码


```python

from transformers import LlavaForConditionalGeneration, AutoProcessor
import torch
from PIL import Image
```

```python

raw_model_name_or_path = "yuanzhoulvpi/llava_qwen15-4b-chat_openai-clip-vit-large-patch14-336"
model = LlavaForConditionalGeneration.from_pretrained(raw_model_name_or_path,device_map="cuda:0", torch_dtype=torch.bfloat16)
processor = AutoProcessor.from_pretrained(raw_model_name_or_path)
model.eval()
print('ok')
```

```python
testdata = (
'<image>\nRelay a brief, clear account of the picture shown.', # 提问
 'large kitchen island with an overhang and dining space next to it', # 真实答案
 'data/liuhaotian/LLaVA-CC3M-Pretrain-595K/images_dl/GCC_train_001899387.jpg' # 图片路径
) 

```


```python
def build_model_input(model, processor, testdata:tuple):
    messages = [
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": testdata[0]},
    ]
    prompt = processor.tokenizer.apply_chat_template(
        messages, tokenize=False, add_generation_prompt=True
    )
    # print(prompt)
    # print("*"*20)
    image = Image.open(testdata[2])
    inputs = processor(text=prompt, images=image, return_tensors="pt")
    
    for tk in inputs.keys():
        inputs[tk] = inputs[tk].to(model.device)
    generate_ids = model.generate(**inputs, max_new_tokens=20)
    
    generate_ids = [
        oid[len(iids):] for oid, iids in zip(generate_ids, inputs.input_ids)
    ]

    

    gen_text = processor.batch_decode(generate_ids, skip_special_tokens=False, clean_up_tokenization_spaces=False)[0]
    return gen_text



```

```python
build_model_input(model, processor, testdata)

# 'the kitchen is a bright yellow with a glass top island and a large window that looks out to the'
```