update research section (#99)

This commit is contained in:
vintro 2025-04-23 18:24:27 -04:00 committed by GitHub
parent 12972056cb
commit 1f5c5417b4
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194

View File

@ -42,6 +42,15 @@ We're building systems that haven't been built before, solving problems that hav
## Research We're Excited About
[Inference-Time Scaling for Generalist Reward Modeling](https://arxiv.org/abs/2504.02495)
[All Roads Lead to Likelihood: The Value of Reinforcement Learning in Fine-Tuning](https://arxiv.org/abs/2503.01067)
[Cognitive Behaviors that Enable Self-Improving Reasoners, or, Four Habits of Highly Effective STaRs](https://arxiv.org/abs/2503.01307)
[Is a Good Foundation Necessary for Efficient Reinforcement Learning? The Computational Role of the Base Model in Exploration](https://arxiv.org/abs/2503.07453)
[Echo Chamber: RL Post-training Amplifies Behaviors Learned in Pretraining](https://arxiv.org/abs/2504.07912)
[Finding Flawed Fictions: Evaluating Complex Reasoning in Language Models via Plot Hole Detection](https://arxiv.org/abs/2504.11900)
[Self-playing Adversarial Language Game Enhances LLM Reasoning](https://arxiv.org/abs/2404.10642)
[Welcome to the Era of Experience](https://storage.googleapis.com/deepmind-media/Era-of-Experience%20/The%20Era%20of%20Experience%20Paper.pdf)
[TextArena](https://arxiv.org/abs/2504.11442)
[s1: Simple test-time scaling](https://arxiv.org/abs/2501.19393)
[Neural Networks Are Elastic Origami!](https://youtu.be/l3O2J3LMxqI?si=bhodv2c7GG75N2Ku)
[Titans: Learning to Memorize at Test Time](https://arxiv.org/abs/2501.00663v1)