mirror of
https://github.com/deepseek-ai/DeepSeek-R1.git
synced 2026-11-05 00:47:42 +00:00
[GH-ISSUE #851] Deepseek Not obeying rules or instructions specially expert mode #703
Labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
deepseek-ai/DeepSeek-R1#703
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @dinosour956-beep on GitHub (Jun 7, 2026).
Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-R1/issues/851
Problem Description
I've been using the DeepSeek app (chat.deepseek.com) for over a year and really loved it. The model used to be excellent at following user instructions precisely.
However, in recent updates (especially in Expert mode), it has become much worse at:
It sometimes ignores key constraints, deviates from requested style/structure, or gives incomplete adherence even when instructions are clear and repeated.
Expected Behavior
The model should reliably follow user instructions while keeping responses helpful and natural, as it did in earlier versions.
Additional Context
I understand prioritizing reasoning depth is important, but instruction adherence is critical for many users. Any improvements here would be greatly appreciated!
Thanks for the great work overall — hoping this feedback helps.
@github-actions[bot] commented on GitHub (Jul 8, 2026):
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. If you believe this issue is still relevant, please leave a comment to keep it open. Thank you for your contributions!
@stream2stream commented on GitHub (Jul 20, 2026):
This is a massive issue. We are using deepseek-v4-pro in Codewhale, and it will not follow instructions, deviates from the rules you give it, breaks out of the Persona you put it into, and constantly gives coding solutions to problems that cannot be solved with code. We are finding it unworkable.
@github-actions[bot] commented on GitHub (Aug 20, 2026):
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. If you believe this issue is still relevant, please leave a comment to keep it open. Thank you for your contributions!
@github-actions[bot] commented on GitHub (Sep 3, 2026):
false