Aayush Karan's banner

Aayush Karan

@aakaran31 • 2,899 subscribers

PhD student @Harvard | Algorithmic insights for foundation models | @PDSoros 2024 | Prev @NVIDIA, @GoogleDeepMind, @citsecurities, @apple

Shorts

We found a new way to get language models to reason. 🤯 No RL, no training, no verifiers, no prompting. ❌ With better sampling, base models can achieve single-shot reasoning on par with (or better than!) GRPO while avoiding its characteristic loss in generation diversity.

We found a new way to get language models to reason. 🤯 No RL, no training, no verifiers, no prompting. ❌ With better sampling, base models can achieve single-shot reasoning on par with (or better than!) GRPO while avoiding its characteristic loss in generation diversity.

280,576 просмотров