近期论文15
- Generative Verifiers: Reward Modeling as Next-Token Prediction第 6 位,共 6 位作者ICLR 2025
- Speculative Knowledge Distillation: Bridging the Teacher-Student Gap Through Interleaved Sampling第 8 位,共 14 位作者ICLR 2025
- Training Language Models to Self-Correct via Reinforcement Learning第 3 位,共 19 位作者ICLR 2025
- On-Policy Distillation of Language Models: Learning from Self-Generated Mistakes第 1 位,共 8 位作者ICLR 2024 · 311 次引用
- DistillSpec: Improving Speculative Decoding via Knowledge Distillation第 8 位,共 9 位作者ICLR 2024 · 143 次引用
- Stop Regressing: Training Value Functions via Classification for Scalable Deep RL第 12 位,共 12 位作者ICML 2024 · 118 次引用
- The Dormant Neuron Phenomenon in Deep Reinforcement Learning第 2 位,共 4 位作者ICML 2023 · 153 次引用
- Offline Q-learning on Diverse Multi-Task Data Both Scales And Generalizes第 2 位,共 5 位作者ICLR 2023 · 3 次引用
- Reincarnating Reinforcement Learning: Reusing Prior Computation to Accelerate Progress第 1 位,共 5 位作者NeurIPS 2022 · 95 次引用
- DR3: Value-Based Deep Reinforcement Learning Requires Explicit Regularization第 2 位,共 7 位作者ICLR 2022 · 85 次引用
- Deep Reinforcement Learning at the Edge of the Statistical Precipice第 1 位,共 4 位作者NeurIPS 2021 · 1,067 次引用
- Neural Additive Models: Interpretable Machine Learning with Neural Nets第 1 位,共 7 位作者NeurIPS 2021 · 663 次引用
- Implicit Under-Parameterization Inhibits Data-Efficient Deep Reinforcement Learning第 2 位,共 5 位作者ICLR 2021 · 155 次引用
- Contrastive Behavioral Similarity Embeddings for Generalization in Reinforcement Learning第 1 位,共 4 位作者ICLR 2021 · 27 次引用
- An Optimistic Perspective on Offline Reinforcement Learning第 1 位,共 3 位作者ICML 2020 · 568 次引用
