Для локального использования - сколько vram оптимально?

Language Models Can Learn from Verbal Feedback Without Scalar Rewards https://arxiv.org/abs/2509.22638 https://www.alphaxiv.org/ru/overview/2509.22638v1…

Vision-Zero: Scalable VLM Self-Improvement via Strategic Gamified Self-Play https://arxiv.org/abs/2509.25541 https://www.alphaxiv.org/ru/overview/2509.25541v1…

From f(x) and g(x) to f(g(x)): LLMs Learn New Skills in RL by Composing Old Ones https://arxiv.org/abs/2509.25123…

ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory https://arxiv.org/abs/2509.25140

Не зря нефтяной проект делаю