-
Notifications
You must be signed in to change notification settings - Fork 2.3k
Insights: huggingface/open-r1
Overview
-
0 Active pull requests
-
- 0 Merged pull requests
- 0 Open pull requests
- 1 Closed issue
- 3 New issues
There hasn’t been any commit activity on huggingface/open-r1 in the last week.
Want to help out?
1 Issue closed by 1 person
-
Potential bug: decoder prompt length longer than maximum model length
#672 closed
Jun 16, 2025
3 Issues opened by 3 people
-
Looking for someone with curating long-CoT dataset experience
#681 opened
Jun 21, 2025 -
about the model run in vllm server when resume from checkpoint in grpo?
#680 opened
Jun 20, 2025 -
Evaluate during Training
#679 opened
Jun 15, 2025
5 Unresolved conversations
Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.
-
After I trained for 500 steps, the length of think became smaller and smaller, and even disappeared.
#394 commented on
Jun 17, 2025 • 0 new comments -
`model.generate` produces right-padded completions, causing incompatibility with Flash Attention 2
#599 commented on
Jun 17, 2025 • 0 new comments -
Datasets for code
#28 commented on
Jun 18, 2025 • 0 new comments -
Does the default training script for step 1 sft phase, calculate the loss for the prompt part?
#648 commented on
Jun 18, 2025 • 0 new comments -
Hanging during move_model_to_vllm in GRPO on single node
#575 commented on
Jun 20, 2025 • 0 new comments