Training-Free Dynamic Upcycling of Expert Language Models
Paper • 2603.29765 • Published • 10
Gensyn Papers
-
-
Sharing is Caring: Efficient LM Post-Training with Collective RL Experience Sharing
Paper • 2509.08721 • Published • 665 -
All is Not Lost: LLM Recovery without Checkpoints
Paper • 2506.15461 • Published • 40 -
NoLoCo: No-all-reduce Low Communication Training Method for Large Models
Paper • 2506.10911 • Published • 9 -
Verde: Verification via Refereed Delegation for Machine Learning Programs
Paper • 2502.19405 • Published • 9 -
HDEE: Heterogeneous Domain Expert Ensemble
Paper • 2502.19385 • Published • 5 -
SkipPipe: Partial and Reordered Pipelining Framework for Training LLMs in Heterogeneous Networks
Paper • 2502.19913 • Published • 6 -
Hail to the Thief: Exploring Attacks and Defenses in Decentralised GRPO
Paper • 2511.09780 • Published • 29 -
F-TIS: Harnessing Diverse Models in Collaborative GRPO
Paper • 2605.22537 • Published • 12 -
IR3DE: A Linear Router for Large Language Models
Paper • 2606.06098 • Published • 4 -
The Role of Feedback Alignment in Self-Distillation
Paper • 2606.11173 • Published • 3 -
DEI: Diversity in Evolutionary Inference for Quality-Diversity Search
Paper • 2605.27130 • Published • 14