1 Pull request opened by 1 person

• Missing thinking tokens flag for no budget forcing eval

  #100 opened Mar 17, 2025

7 Issues closed by 7 people

• Are incorrect responses are used for training?

  #105 closed Mar 25, 2025

• Question about Idavidrein/gpqa data source in s1k-1.1 dataset

  #99 closed Mar 17, 2025

• [Evaluating trained model] safetensors_rust.SafetensorError: Error while deserializing header: MetadataIncompleteBuffer

  #98 closed Mar 15, 2025

• Distilling from Claude 3.7

  #90 closed Mar 12, 2025

• Special tag in "text" column of simplescaling/s1K-1.1_tokenized

  #94 closed Mar 10, 2025

• Diff for lm-evaluation-harness changes?

  #88 closed Mar 5, 2025

• Any plan about s1.1 data release

  #86 closed Mar 3, 2025

16 Issues opened by 15 people

• Issue Reproducing s1.1-32B Training Loss (Observed vs. WandB)

  #108 opened Mar 31, 2025

• Can I specify the thinking context of the query when s1 inferencing with vllm?

  #107 opened Mar 24, 2025

• qwen2.5-32B can not fine tuning on 80GB A100 in max_seq_length=32768

  #106 opened Mar 21, 2025

• why the model file size is 129g?

  #104 opened Mar 20, 2025

• The minimum GPU resources needed to fine-tune the 32B model?

  #103 opened Mar 19, 2025

• Error in Evaluating on Other Dataset (AssertionError: min_tokens_thinking only supports until_thinking tokens that are 1 token long)

  #102 opened Mar 19, 2025

• Exploring LoRA Adapters & Low Precision Training in S1 for Enhanced Test-Time Scaling

  #101 opened Mar 19, 2025

• s1-32b keeps generating same trajectories and final answer regardless of changing of seeds

  #97 opened Mar 12, 2025

• Experimental communication

  #96 opened Mar 12, 2025

• ValueError: please provide at least one prompt

  #95 opened Mar 11, 2025

• Token-conditonal control code?

  #93 opened Mar 10, 2025

• How is the AIME24 dataset of 30 rows created ?

  #92 opened Mar 9, 2025

• About Figure4 (a) and Table 1

  #91 opened Mar 5, 2025

• Question on rejection sampling and Fig 6 in paper.

  #89 opened Mar 4, 2025

• Is OPENAI_API_KEY necessary?

  #87 opened Mar 4, 2025

• Gemini thinking flash API no longer returns thoughts and response separately

  #85 opened Mar 3, 2025

4 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.

• Ollama ??

  #57 commented on Mar 4, 2025 • 0 new comments

• KeyError: 'text'

  #39 commented on Mar 5, 2025 • 0 new comments

• Did you try using budget forcing directly on Qwen-32B-Instruct?

  #45 commented on Mar 10, 2025 • 0 new comments

• Performance on Qwen2.5-7B-Instruct

  #42 commented on Mar 17, 2025 • 0 new comments