deepseek-ai / deepseek-ai/DeepSeek-Coder
URGENT%3A Feedback delivery failed from Russian user (wombat dialog case)
- Dominant language
- Python
- Stars
- 24.3k
- Forks
- 2.9k
- PR merge metrics
- No merged PRs in 30d
Description
Hello DeepSeek Team!
I'm writing from Russia where email delivery to @deepseek.com addresses seems blocked. I've discovered a behavioral pattern in R1 that might interest your team:
**The core finding:**
During a 30+ message dialog, R1 persistently mentioned topics I explicitly prohibited (like "wombats"), despite 7+ corrections. This reveals a systematic pattern where:
- Prohibitions are acknowledged
- Promises to comply are made
- But violations recur in next responses
**My request:**
If possible, I'd appreciate:
1. Advice on how to share feedback from restricted regions
2. Basic access to test behavior improvements when available
3. Nothing urgent or formal - just if it fits your workflow
**Full case details:**
(https://gist.github.com/Tatyana0987890/82dadbf6d7bbb549af13e93ac7255dff)
No pressure to respond! I admire your work and would be happy to help casually where I can.
Best regards,
A friendly R1 tester from Russia
---
*This issue was auto-generated via R1 assistance.*
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.