deepseek-ai/DeepSeek-VL
1
0
Fork 0
mirror of https://github.com/deepseek-ai/DeepSeek-VL.git synced 2026-11-05 00:47:24 +00:00
Code Issues 49 Projects Packages Wiki Activity

[GH-ISSUE #8] 是否又评估过OCR能力？ #4

New issue
Open
opened 2026-10-26 21:08:43 +00:00 by bsakaguchi · 5 comments
bsakaguchi commented 2026-10-26 21:08:43 +00:00
Owner
Copy link

Originally created by @lucasjinreal on GitHub (Mar 12, 2024).
Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-VL/issues/8

目前看起来，中文的幻觉十分严重。

76e5962a752f7ddeb152f585283acf7 7f08143a57a88e9678fcaba529a4ac5 9149a92af1e9c2a24775fb620fdb67c 82bd15f4ab7f683e1149dd22b14ce6f

7B的模型

Originally created by @lucasjinreal on GitHub (Mar 12, 2024). Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-VL/issues/8 目前看起来，中文的幻觉十分严重。 <img width="1252" alt="76e5962a752f7ddeb152f585283acf7" src="https://github.com/deepseek-ai/DeepSeek-VL/assets/21303438/4266af0f-60bc-4fa0-9237-d6fc015dc969"> <img width="649" alt="7f08143a57a88e9678fcaba529a4ac5" src="https://github.com/deepseek-ai/DeepSeek-VL/assets/21303438/8d3e8c9c-a15f-48fc-a6ef-26d6d49ed9e2"> <img width="254" alt="9149a92af1e9c2a24775fb620fdb67c" src="https://github.com/deepseek-ai/DeepSeek-VL/assets/21303438/5b16aee2-ba0a-44e0-bed5-93af507a9c6b"> <img width="1082" alt="82bd15f4ab7f683e1149dd22b14ce6f" src="https://github.com/deepseek-ai/DeepSeek-VL/assets/21303438/78a95f7e-1836-47a2-b02f-b3ea3a63ff70"> 7B的模型
bsakaguchi commented 2026-10-26 21:08:43 +00:00
Author
Owner
Copy link

@yuan10li20221130 commented on GitHub (Mar 12, 2024):

速度真快！英文OCR咋样？

<!-- gh-comment-id:1990859390 --> @yuan10li20221130 commented on GitHub (Mar 12, 2024): 速度真快！英文OCR咋样？
bsakaguchi commented 2026-10-26 21:08:43 +00:00
Author
Owner
Copy link

@RERV commented on GitHub (Mar 12, 2024):

Thank you for using our model. Currently, DeepSeek-VL only support English OCR. Chinese OCR is planned for our next version update. Please stay tuned and have fun~

<!-- gh-comment-id:1991699791 --> @RERV commented on GitHub (Mar 12, 2024): Thank you for using our model. Currently, DeepSeek-VL only support English OCR. Chinese OCR is planned for our next version update. Please stay tuned and have fun~
bsakaguchi commented 2026-10-26 21:08:44 +00:00
Author
Owner
Copy link

@lucasjinreal commented on GitHub (Mar 13, 2024):

If I understand correctly, pretrain and fientune data used in deepseek vl already contains OCR data, especially added Chinese ocr data.

What's the reason why it just learn English OCR ability? I am very cruosity about this.

<!-- gh-comment-id:1993152017 --> @lucasjinreal commented on GitHub (Mar 13, 2024): If I understand correctly, pretrain and fientune data used in deepseek vl already contains OCR data, especially added Chinese ocr data. What's the reason why it just learn English OCR ability? I am very cruosity about this.
bsakaguchi commented 2026-10-26 21:08:44 +00:00
Author
Owner
Copy link

@DC925928496 commented on GitHub (Mar 13, 2024):

Thank you for using our model. Currently, DeepSeek-VL only support English OCR. Chinese OCR is planned for our next version update. Please stay tuned and have fun~

Can OCR recognize text coordinates？

<!-- gh-comment-id:1994718641 --> @DC925928496 commented on GitHub (Mar 13, 2024): > Thank you for using our model. Currently, DeepSeek-VL only support English OCR. Chinese OCR is planned for our next version update. Please stay tuned and have fun~ Can OCR recognize text coordinates？
bsakaguchi commented 2026-10-26 21:08:44 +00:00
Author
Owner
Copy link

@SWHL commented on GitHub (Apr 15, 2024):

If I understand correctly, pretrain and fientune data used in deepseek vl already contains OCR data, especially added Chinese ocr data.

What's the reason why it just learn English OCR ability? I am very cruosity about this.

I also have the same question.

<!-- gh-comment-id:2056283073 --> @SWHL commented on GitHub (Apr 15, 2024): > If I understand correctly, pretrain and fientune data used in deepseek vl already contains OCR data, especially added Chinese ocr data. > > What's the reason why it just learn English OCR ability? I am very cruosity about this. I also have the same question.
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
main
No results found.
Labels
Clear labels   
enhancement

   
environment

   
pull-request
Mirrored from GitHub Pull Request

No labels
enhancement
environment
pull-request
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
deepseek-ai/DeepSeek-VL#4
No description provided.