Jian-Chao Tan

LLM agents trained with MCGPO achieve up to 5.16% higher success rates by intelligently reusing rollout data and constructing context-aware action preferences.

INT4 KV caching for MLA models is now possible, achieving 3.59x compression at 128K context and enabling 5.168x higher throughput in cache-pressure serving workloads.

LSA slashes indexing overhead while maintaining full attention performance, enabling efficient long-context processing for models with up to one million tokens.

MONA unlocks faster LLM pretraining and superior downstream performance by turbocharging the Muon optimizer with Nesterov-style acceleration, leaving AdamW in the dust.

Channel-wise adaptive learning rates in Gated Delta Networks unlock superior long-context recall, rivaling softmax attention without the quadratic cost.

Training long-context sparse attention models doesn't have to be a slow, imbalanced mess: SparseBalance achieves 1.33x speedup while *improving* accuracy.

Achieve full-attention accuracy with 10x operator speedup and 4.7x throughput improvement in long-context LLM inference by overlapping KV cache transfers with computation.