PretrainZero: Reinforcement Active Pretraining
Paper • 2512.03442 • Published • 50
Papers
-
-
UniQL: Unified Quantization and Low-rank Compression for Adaptive Edge LLMs
Paper • 2512.03383 • Published • 6 -
ToolOrchestra: Elevating Intelligence via Efficient Model and Tool Orchestration
Paper • 2511.21689 • Published • 126 -
Nemotron-Flash: Towards Latency-Optimal Hybrid Small Language Models
Paper • 2511.18890 • Published • 35 -
Every Token Counts: Generalizing 16M Ultra-Long Context in Large Language Models
Paper • 2511.23319 • Published • 24 -
WUSH: Near-Optimal Adaptive Transforms for LLM Quantization
Paper • 2512.00956 • Published • 23 -
CUDA-L2: Surpassing cuBLAS Performance for Matrix Multiplication through Reinforcement Learning
Paper • 2512.02551 • Published • 13 -
Stabilizing Reinforcement Learning with LLMs: Formulation and Practices
Paper • 2512.01374 • Published • 106 -
LightRAG: Simple and Fast Retrieval-Augmented Generation
Paper • 2410.05779 • Published • 39 -
MinerU2.5: A Decoupled Vision-Language Model for Efficient High-Resolution Document Parsing
Paper • 2509.22186 • Published • 160 -
End-to-End Test-Time Training for Long Context
Paper • 2512.23675 • Published • 24 -
Hybrid Policy Distillation for LLMs
Paper • 2604.20244 • Published • 11 -
How to Fine-Tune a Reasoning Model? A Teacher-Student Cooperation Framework to Synthesize Student-Consistent SFT Data
Paper • 2604.14164 • Published • 34 -
SkillLearnBench: Benchmarking Continual Learning Methods for Agent Skill Generation on Real-World Tasks
Paper • 2604.20087 • Published • 14 -
Self-Distillation Enables Continual Learning
Paper • 2601.19897 • Published • 31