Tag: Post-Training

All the articles with the tag "Post-Training".

• How Should We Repair Reasoning Traces Before Distillation?

  1 Oct, 2026 · 23 min read

  A map of 26 research papers on reasoning distillation, trajectory correction, and student compatibility: where each method acts, what it costs to implement, and an experiment to test how much of an RL teacher trace to rewrite.

• What Makes an SFT Example Worth Learning? Quality, Novelty, and Learnability

  Updated: 27 Sep, 2026 · 23 min read

  From curriculum learning and DAgger to LIMA, LESS, and on-policy distillation: a source-backed framework for separating SFT data quality, novelty, and learnability, with executable examples and a small negative result.

• Can RFT Learn From Failure? Negative Signals for Agent Fine-Tuning

  20 Sep, 2026 · 16 min read

  A technical guide to learning from failed agent rollouts through step masking, unlikelihood, DPO, SimPO, and step-level credit assignment.

• From GRPO Outcome Rewards to Token-Level Advantage

  Updated: 6 Sep, 2026 · 22 min read

  A practical framework for turning GRPO-style sequence rewards into token-level advantages, including GAE-style estimators, credit assignment routes, and multi-reward training design.

• When Does Model Souping Work for LLMs?

  Updated: 6 Sep, 2026 · 21 min read

  A practical guide to when LLM weight averaging and model merging work, why they fail, how methods such as Task Arithmetic, TIES, DARE, and LoRA merging differ, and how to evaluate a merge before deployment.

• Do Verifier Errors Grow Superlinearly with Horizon? A Six-Stage Experiment

  Updated: 17 Aug, 2026 · 17 min read

  A preregistered 224-artifact study confirms a 66% hybrid-verifier improvement, does not support the universal superlinear headline, and finds that task structure can reverse the apparent horizon effect.