Publications
- Macaron-V1: Towards Open Continual Learning with Self-Improvement and Mixture-of-LoRARead
- LongStraw: Long-Context RL Beyond 2M Tokens under a Fixed GPU BudgetRead
- On the Scaling of PEFT: Towards Million Personal Models of Trillion ParametersRead
- Macaron-A2UI: A Model for Generative UI in Personal AgentsRead
- MinT: Managed Infrastructure for Training and Serving Millions of LLMsRead
- δ-mem: Efficient Online Memory for Large Language ModelsRead