https://t.me/einytech/131

On-Policy Self-Distillation without Any Supervision https://arxiv.org/abs/2608.06296 https://www.alphaxiv.org/ru/overview/2608.06296

Reversal Q-Learning https://arxiv.org/abs/2606.17551 https://www.alphaxiv.org/ru/overview/2606.17551

Да, попробовал v0.1.2 на Windows 11, RTX 3070 Laptop 8 GB, Qwen3.5-9B UD-Q3KXL. Приложение RTX 3070 определило и запустило llama-server с `--n-gpu-layers 999`,…

Learning to Disprove: Formal Counterexample Generation with Large Language Models https://arxiv.org/abs/2603.19514…

SFT Conflicts, RL Coexists: A Theoretical and Empirical Analysis of Multi-Task Learning for LLMs https://arxiv.org/abs/2608.03573…