Spaces:
Running
on
Zero
Running
on
Zero
khang119966
commited on
Update app.py
Browse files
app.py
CHANGED
@@ -226,10 +226,7 @@ button.svelte-1lcyrx4[aria-label="user's message: a file of type image/jpeg, "]
|
|
226 |
|
227 |
demo = gr.ChatInterface(
|
228 |
fn=chat,
|
229 |
-
description="""
|
230 |
-
To further enhance its performance in Vietnamese while maintaining good capabilities on existing English datasets, Vintern-1B-v3.5 has been fine-tuned using a vast amount of Vietnamese-specific data. This results in a model that is exceptionally powerful in text recognition, OCR, and understanding Vietnam-specific documents.
|
231 |
-
The model might have biases because it learned from data that could be biased.
|
232 |
-
Users should be cautious of these possible biases when using the model.""",
|
233 |
examples=[{"text": "Mô tả hình ảnh.", "files":["./demo_3.jpg"]},
|
234 |
{"text": "Trích xuất các thông tin từ ảnh trả về markdown.", "files":["./demo_1.jpg"]},
|
235 |
{"text": "Viết một bài quảng cáo dài trên mạng xã hội cho hình ảnh.", "files":["./demo_2.jpg"]}],
|
|
|
226 |
|
227 |
demo = gr.ChatInterface(
|
228 |
fn=chat,
|
229 |
+
description="""**Vintern-1B-v3.5** is the latest in the **Vintern** series, bringing **major improvements** over v2 across all benchmarks. 🚀 Fine-tuned from **InternVL-1B-2.5**, already strong in **Vietnamese 🇻🇳 tasks** thanks to **Viet-ShareGPT-4o-Text-VQA** data, this version further enhances **Vietnamese capabilities** while retaining **strong English performance**. It excels in **OCR**, **text recognition**, and **Vietnam-specific document understanding**. 📄 However, users should be mindful of **potential biases** due to the training data. ⚠️""",
|
|
|
|
|
|
|
230 |
examples=[{"text": "Mô tả hình ảnh.", "files":["./demo_3.jpg"]},
|
231 |
{"text": "Trích xuất các thông tin từ ảnh trả về markdown.", "files":["./demo_1.jpg"]},
|
232 |
{"text": "Viết một bài quảng cáo dài trên mạng xã hội cho hình ảnh.", "files":["./demo_2.jpg"]}],
|