paperHugging Face Daily Papers·22 Oct 20264.1
Traces the Wait token in reasoning models back to pretraining data rather than RL.
paperarXiv (systems, data, ML)·23 Oct 20263.8
Optimizes pass@k directly to prune reasoning candidates while keeping diversity at test time.
blogSebastian Raschka·24 Oct 20263.8
A full video course on building reasoning LLMs from scratch, covering inference-time scaling and RL training.
paperarXiv (systems, data, ML)·23 Oct 20263.5
Isolates context in chain of thought to stop attention leakage.
paperarXiv (systems, data, ML)·22 Oct 20263.2
Multi-token prediction and alternative tokenization help LMs learn latent graph shortest-path search.