Selecting Diverse SFT Traces Improves Post-RL Generalization Paper • 2609.33780 • Published 7 days ago • 38
CausaLab: A Scalable Environment for Interactive Causal Discovery Toward AI Scientists Paper • 2605.26029 • Published May 28 • 14
Useful Memories Become Faulty When Continuously Updated by LLMs Paper • 2605.12978 • Published May 13 • 19
Good SFT Optimizes for SFT, Better SFT Prepares for Reinforcement Learning Paper • 2602.01058 • Published Feb 1 • 45
Good SFT Optimizes for SFT, Better SFT Prepares for Reinforcement Learning Paper • 2602.01058 • Published Feb 1 • 45
is-sft-271828/synlogic_grpo_qwen3-4b-base-qwen3-8b-3e-5-seq-seqlen_8192_chunksize_8 4B • Updated Jan 27 • 8
is-sft-271828/synlogic_grpo_qwen3-4b-base-qwen3-8b-3e-5-seq-seqlen_8192_chunksize_8 4B • Updated Jan 27 • 8
is-sft-271828/synlogic_grpo_qwen3-4b-base-qwen3-8b-3e-5-seq-seqlen_8192_chunksize_4 4B • Updated Jan 27 • 9
is-sft-271828/synlogic_grpo_qwen3-4b-base-qwen3-8b-3e-5-seq-seqlen_8192_chunksize_4 4B • Updated Jan 27 • 9
is-sft-271828/math_grpo_qwen3-4b-base-qwen3-8b-3e-5-seq-seqlen_8192_chunksize_8 4B • Updated Jan 27 • 8
is-sft-271828/math_grpo_qwen3-4b-base-qwen3-8b-3e-5-seq-seqlen_8192_chunksize_8 4B • Updated Jan 27 • 8
is-sft-271828/math_grpo_qwen3-4b-base-qwen3-8b-3e-5-seq-seqlen_8192_chunksize_4 4B • Updated Jan 27 • 10
is-sft-271828/math_grpo_qwen3-4b-base-qwen3-8b-3e-5-seq-seqlen_8192_chunksize_4 4B • Updated Jan 27 • 10