Mostofa Patwary

Nemotron-CC-Math: A 133 Billion-Token-Scale High Quality Math Pretraining Dataset

RLP: Reinforcement as a Pretraining Objective

Front-Loading Reasoning: The Synergy between Pretraining and Post-Training Data

FusionFactory: Fusing LLM Capabilities with Multi-LLM Log Data

Multi-Agent Evolve: LLM Self-Improve through Multi-Agent Co-evolution

Prismatic Synthesis: Gradient-based Data Diversification Boosts Generalization in LLM Reasoning

Efficient Hybrid Language Model Compression through Group-Aware SSM Pruning

MIND: Math Informed syNthetic Dialogues for Pretraining LLMs

LLM Pruning and Distillation in Practice