mirror of
https://github.com/deepseek-ai/DeepSeek-R1.git
synced 2026-11-05 00:47:42 +00:00
[GH-ISSUE #442] Confused about the 800k SFT (on DeepSeek-V3-Base or CKPT of Reasoning RL ) #353
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#353
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 @HaitaoWuTJU on GitHub (Feb 17, 2025).
Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-R1/issues/442
(CoT Cold Start) SFT -> Reasoning RL -> (Reject Sampling) SFT（600k Reasoning+200k Non-Reasoning）->All Scenarios RL
Page 3: “Upon nearing convergence in the RL process, we create new SFT data through rejection
sampling on the RL checkpoint, combined with supervised data from DeepSeek-V3 in domains
such as writing, factual QA, and self-cognition, and then retrain the DeepSeek-V3-Base model.”
Page 11: “We fine-tune DeepSeek-V3-Base for two epochs using the above curated dataset of about
800k samples.”
Is "(CoT Cold Start) SFT -> Reasoning RL" only for generating 600k Reasoning CoT data?
@github-actions[bot] commented on GitHub (Mar 21, 2025):
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 (Apr 24, 2025):
false