File size: 13,763 Bytes
1188cc9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d948b64
46d4e53
02277ae
 
c41c00f
 
 
 
1188cc9
fca3735
d4d0f79
fca3735
 
1188cc9
 
a605c65
1188cc9
ce6d29f
1188cc9
c269769
1188cc9
2766807
9faf4db
2766807
9faf4db
 
 
38d492c
 
 
 
 
1188cc9
a605c65
ecb789e
 
a605c65
1188cc9
a1eb664
a9f189f
36e4d64
1188cc9
 
 
c2f1a37
1188cc9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c2f1a37
a1eb664
a9f189f
 
a1eb664
1188cc9
 
 
ea5cacd
1188cc9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ea5cacd
a9f189f
 
 
 
1188cc9
 
 
 
a9f189f
 
 
6ce6a2d
1188cc9
a9f189f
1188cc9
 
6ce6a2d
1188cc9
bc875ce
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a605c65
ac6a587
ce6d29f
 
8220107
1188cc9
a605c65
605e2f3
1188cc9
 
 
605e2f3
83d6be0
605e2f3
 
 
 
b0c5e0b
 
 
605e2f3
 
1188cc9
a605c65
ecb789e
 
b0c5e0b
9e8eeca
b0c5e0b
1188cc9
b0c5e0b
1188cc9
b0c5e0b
1188cc9
b0c5e0b
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
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
---
license: apache-2.0
language:
- vi
- en
- zh
base_model:
- Qwen/Qwen2-VL-7B-Instruct
library_name: transformers
tags:
- erax
- multimodal
- erax-vl-7b
- insurance
- ocr
- vietnamese
- bcg
pipeline_tag: visual-question-answering

widget:
- src: images/photo-1-16505057982762025719470.webp
  example_title: Test 1
- src: images/vt-don-thuoc-f0-7417.jpeg
  example_title: Test 2
---
<p align="center">
  <img src="https://cdn-uploads.huggingface.co/production/uploads/66e93d483745423cbb14c5ff/fNxjr3en_onzbOv0sghpE.jpeg" alt="Logo">
</p>
<!-- ![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/66e93d483745423cbb14c5ff/fNxjr3en_onzbOv0sghpE.jpeg) -->

# EraX-VL-7B-V1
## Introduction🎉

We are excited to introduce **EraX-VL-7B-v1**, a robust multimodal model for **OCR (optical character recognition)** and **VQA (visual question-answering)** that excels in various languages🌍, with a particular focus on Vietnamese🇻🇳. The `EraX-VL-7B` model stands out for its precise recognition capabilities across a range of documents📝, including medical forms 🩺, invoices 🧾, bills of sale 💳, quotes 📄, and medical records 💊. This functionality is expected to be highly beneficial for hospitals 🏥, clinics 💉, insurance companies 🛡️, and other similar applications 📋. Built on the solid foundation of the [Qwen/Qwen2-VL-7B-Instruct](https://huggingface.co/Qwen/Qwen2-VL-7B-Instruct)[1], which we found to be of high quality and fluent in Vietnamese, `EraX-VL-7B` has been fine-tuned to enhance its performance. We plan to continue improving and releasing new versions for free, along with sharing performance benchmarks in the near future.

**EraX-VL-7B-V1** is a young member of our **EraX's LànhGPT** collection of LLM models.

- **Developed by:**
  - Nguyễn Anh Nguyên (nguyen@erax.ai)
  - Nguyễn Hồ Nam (BCG)
  - Hoàng Tiến Dũng (dung.ht@erax.ai)
  - Phạm Huỳnh Nhật (nhat.ph@erax.ai)
  - Phạm Đình Thục (thuc.pd@erax.ai)
- **Funded by:** [Bamboo Capital Group](https://bamboocap.com.vn) and EraX
- **Model type:** Multimodal Transformer with over 7B parameters
- **Languages (NLP):** Primarily Vietnamese with multilingual capabilities
- **License:** Apache 2.0
- **Fine-tuned from:** [Qwen/Qwen2-VL-7B-Instruct](https://huggingface.co/Qwen/Qwen2-VL-7B-Instruct)

## Benchmarks📊
Coming Soon!!!

## Examples🧩

<details>
  <summary>Example 01:</summary>

![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/66e93d483745423cbb14c5ff/Q5GkK8vuZ9zDVPkhwu4yH.jpeg)

* Kết quả:
```
{
    "document": {
        "header": {
            "title": "GIẤY HẸN KHÁM LẠI",
            "organization": "SỞ Y TẾ NGHỆ AN\nBỆNH VIỆN UNG BƯỚU NGHỆ AN",
            "address": "Võ Thị Sáu, Thủy Tùng - TP Vinh - Nghệ An"
        },
        "patient_info": {
            "name": "NGUYỄN THỊ LUÂN",
            "date_of_birth": "03/07/1976",
            "gender": "40",
            "address": "Xã Nghĩa Khánh-Huyện Nghĩa Đàn-Nghệ An",
            "medical_card_number": "CN 3 40 40 168 60413",
            "registration_date": "16/12/2016",
            "admission_date": "Từ 01/03/2016",
            "diagnosis": "C20-Bướu ac trực tràng",
            "revisit_date": "17/01/2017"
        },
        "administrative_details": {
            "department": "Trung tâm điều trị ung bướu",
            "revisit_instruction": "vào ngày 17/01/2017, hoặc đến hết kỳ thời gian nếu nước ngoài hẹn khám lại nếu có dấu hiệu (triệu chứng)",
            "note": "nếu KCB ban đầu: Trạm y tế xã Nghĩa Khánh",
            "signature": "Trưởng khoa",
            "doctor_signature": "Lâm Nguyễn Khang",
            "revisiting_date_confirmation": "Ngày 16 tháng 12 năm 2016",
            "confirmation_signature": "Bác sĩ điều trị",
            "physician_signature": "Nguyễn Văn Việt"
        }
    }
}
```
</details>
<details>
  <summary>Example 02:</summary>

![image/png](https://cdn-uploads.huggingface.co/production/uploads/66e93d483745423cbb14c5ff/IrX-QP67TZTcTl3vlp1uZ.png)

* Kết quả:
```
{
  "header": {
    "title": "PHIẾU KHÁM BỆNH",
    "date": "Hà Nội, ngày 23 tháng 3 năm 2020",
    "patient_info": {
      "id": "HN011000002",
      "name": "Vương Hồng Thắng - Năm sinh: 1978",
      "address": "Số 10 tầng 2, TTTM V+, Số 505 Phố Minh Khai, Quận Hai Bà Trưng, Hà Nội",
      "phone": "+0942116117",
      "email": "bamu.vn@gmail.com"
    },
    "contact_info": {
      "address": "Nhà Khoa Bamufit\nĐịa chỉ: 505, Phố Minh Khai, Hai Bà Trưng, Hà Nội, Việt Nam",
      "phone": "0942484784",
      "email": "info@bamufit.vn",
      "website": "https://bamufit.vn"
    }
  },
  "treatment_details": [
    {
      "visit_date": "13-09-2019",
      "treatment_type": "Chẩn đoán: Abscess chẽ",
      "procedure": "Cắt lợi bằng Laser r23",
      "doctor": "THỊ HIEN",
      "price": "500,000",
      "quantity": "1",
      "discounted_price": "0",
      "total_cost": "500,000"
    },
    {
      "visit_date": "13-09-2019",
      "treatment_type": "Chẩn đoán: Abscess quanh chóp",
      "procedure": "Bám gai xuống ở răng r23",
      "doctor": "THỊ HIEN",
      "price": "100,000",
      "quantity": "1",
      "discounted_price": "0",
      "total_cost": "100,000"
    }
  ],
  "financial_details": {
    "total_cost": "600,000",
    "discounted_total": "0",
    "paid_amount": "1,114,000",
    "remaining_balance": "1,714,000"
  },
  "notes": "- Kiêng ăn uống đồ để gây nhiễm mủ như chè, cà phê, thuốc lá, rượu vang đỏ .. và hạn chế dùng đồ quá nóng, quá lạnh sau khi tẩy trắng răng ít nhất 2 tuần.",
  "footer": {
    "customer_signature": "(Ký và ghi rõ họ tên)",
    "accountant_signature": "(Ký và ghi rõ họ tên)",
    "dentist_signature": "(Ký và ghi rõ họ tên)"
  }
}
```
</details>
<details>
  <summary>Example 03:</summary>

![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/66e93d483745423cbb14c5ff/pSSqEOXQCsvz9H76CGQXa.jpeg)

* Kết quả:
Hình ảnh là một biểu đồ thể hiện mối quan hệ giữa chỉ số BMI (Body Mass Index) và tuổi, được chia thành các mức độ khác nhau dựa trên phần trăm percentile. Trục hoành của biểu đồ đại diện cho tuổi từ 2 đến 20 năm, trong khi trục tung đại diện cho chỉ số BMI từ 10 đến 32. Biểu đồ này có ba khu vực chính: vùng màu đỏ ở phía dưới cùng đại diện cho mức béo phì với chỉ số BMI cao hơn 30; vùng màu vàng nằm giữa đại diện cho nguy cơ béo phì với chỉ số BMI từ khoảng 25 đến 30; và vùng màu xanh lá cây ở phía trên đại diện cho mức cân nặng khỏe mạnh hoặc thiếu cân với chỉ số BMI thấp hơn 25. Trên biểu đồ còn có đường cong màu xám chạy qua các mức độ BMI theo tuổi, đánh dấu các mức 5th, 50th, và 95th percentile. Văn bản trong hình gồm các cụm từ 'Béo phì', 'Nguy cơ béo phì', 'Sức khỏe dinh dưỡng tốt', và 'Thiếu cân' để mô tả từng khu vực tương ứng với chỉ số BMI.
</details>
<details>
  <summary>Example 04:</summary>
### Example 04:
![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/66e93d483745423cbb14c5ff/kXrR6s990p0EgphZEtEnJ.jpeg)
</details>

* Kết quả:
Hình ảnh chụp một khung cảnh đồng quê Việt Nam với những nét đặc trưng riêng biệt. Bối cảnh là một cánh đồng lúa xanh mướt, trải dài bất tận, được bao phủ bởi một lớp sương mù nhẹ nhàng. Phía xa là những dãy núi xanh ngắt, tạo nên một bức tranh thiên nhiên thơ mộng và trữ tình. Ở trung tâm bức ảnh, hai cậu bé đang ngồi trên lưng con trâu nước đen tuyền. Cậu bé phía trước mặc quần đỏ, đang cười tươi rói, trong khi cậu bé phía sau mặc quần đen, vẻ mặt trầm tư hơn. Con trâu nước đang đứng yên, đầu hướng về phía bên phải, nhìn về phía xa xăm. Phía bên phải hình ảnh, một người đàn ông đang cầm một cái chậu màu đen, đang đổ nước lên đầu mình. Nước từ chậu chảy xuống tạo thành những giọt nước li ti, tung tóe khắp không gian. Trên bầu trời, có những đám mây trắng bồng bềnh, tạo nên một khung cảnh thanh bình và lãng mạn. Ánh nắng vàng óng chiếu rọi xuống cánh đồng, tạo nên một bức tranh tuyệt đẹp.

## Quickstart🎮
[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1CnSxtWDLG48-NQh7wk9_z8WI7J4OY_Ci?usp=sharing)

Install the necessary packages:
```curl
python -m pip install git+https://github.com/huggingface/transformers accelerate
python -m pip install qwen-vl-utils
pip install flash-attn --no-build-isolation
```

Then you can use `EraX-VL-7B-V1` like this:
```python
import os
import base64
import json

import cv2
import numpy as np
import matplotlib.pyplot as  plt

import torch
from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor
from qwen_vl_utils import process_vision_info

model_path = "erax/EraX-VL-7B-V1"

model = Qwen2VLForConditionalGeneration.from_pretrained(
    model_path,
    torch_dtype=torch.bfloat16,
    attn_implementation="eager", # replace with "flash_attention_2" if your GPU is Ampere architecture
    device_map="auto"
)

tokenizer =  AutoTokenizer.from_pretrained(model_path)
# processor = AutoProcessor.from_pretrained(model_path)

min_pixels = 256 * 28 * 28
max_pixels = 1280 * 28 * 28
processor_model = "Qwen/Qwen2-VL-7B-Instruct-GPTQ-Int4"
processor = AutoProcessor.from_pretrained(
     processor_model,
     min_pixels=min_pixels,
     max_pixels=max_pixels,
 )

image_path ="image.jpg"

with open(image_path, "rb") as f:
    encoded_image = base64.b64encode(f.read())
decoded_image_text = encoded_image.decode('utf-8')
base64_data = f"data:image;base64,{decoded_image_text}"

messages = [
    {
        "role": "user",
        "content": [
            {
                "type": "image",
                "image": base64_data,
            },
            {
                "type": "text",
                "text": "Diễn tả nội dung bức ảnh như 1 bác sỹ giỏi."
                # "Diễn tả nội dung bức ảnh này bằng định dạng json."
            },
        ],
    }
]

# Prepare prompt
tokenized_text = processor.apply_chat_template(
    messages, tokenize=False, add_generation_prompt=True
)

image_inputs, video_inputs = process_vision_info(messages)

inputs = processor(
    text=[ tokenized_text],
    images=image_inputs,
    videos=video_inputs,
    padding=True,
    return_tensors="pt",
)
inputs = inputs.to("cuda")

# Generation configs
generation_config =  model.generation_config
generation_config.do_sample   = True
generation_config.temperature = 0.2
generation_config.top_k       = 1
generation_config.top_p       = 0.001
generation_config.max_new_tokens     = 2048
generation_config.repetition_penalty = 1.1

# Inference
generated_ids = model.generate(**inputs, generation_config=generation_config)
generated_ids_trimmed = [
    out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
]
output_text = processor.batch_decode(
    generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
)

print(output_text[0])
```

## Acknowledgments👏
We thank Khang Đoàn ([5CD-AI](https://huggingface.co/5CD-AI)) for his invaluable support in order to train `EraX-VL-7B-V1`. Our appreciation also goes to AAA JS Company for their support and resources, which significantly contributed to this project. 

## Contact🤝
For correspondence regarding this work, please contact Nguyễn Anh Nguyên at [nguyen@erax.ai](nguyen@erax.ai).

## Citation📝
<!-- - title={EraX-VL-7B-V1: A Highly Efficient Multimodal LLM for Vietnamese, especially for medical forms and bills.},
- author={Nguyễn Anh Nguyên and Nguyễn Hồ Nam (BCG) and Dũng Hoàng and Thục Phạm and Nhật Phạm},
- helpers={Khang Đoàn and AAA JS Company},
- contact={nguyen@erax.ai},
- organization={EraX} -->
If you find our project useful, we would appreciate it if you could star our repository and cite our work as follows:
```
@article{EraX-VL-7B-V1,
  title={EraX-VL-7B-V1: A Highly Efficient Multimodal LLM for Vietnamese, especially for medical forms and bills},
  author={Nguyễn Anh Nguyên and Nguyễn Hồ Nam (BCG) and Dũng Hoàng and Thục Phạm and Nhật Phạm},
  organization={EraX},
  year={2024},
  url={https://huggingface.co/erax-ai/EraX-VL-7B-V1}
}
```

## References📑
[1] Qwen team. Qwen2-VL. 2024.

[2] Bai, Jinze, et al. "Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond." arXiv preprint arXiv:2308.12966 (2023).

[4] Yang, An, et al. "Qwen2 technical report." arXiv preprint arXiv:2407.10671 (2024).

[5] Chen, Zhe, et al. "Internvl: Scaling up vision foundation models and aligning for generic visual-linguistic tasks." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2024.

[6] Chen, Zhe, et al. "How far are we to gpt-4v? closing the gap to commercial multimodal models with open-source suites." arXiv preprint arXiv:2404.16821 (2024).

[7] Tran, Chi, and Huong Le Thanh. "LaVy: Vietnamese Multimodal Large Language Model." arXiv preprint arXiv:2404.07922 (2024).