7 papers
F-TIS: Harnessing Diverse Models in Collaborative GRPO
Nikolay Blagoev, OÄuzhan Ersoy, Wendelin Boehmer +1
Reinforcement learning methods such as GRPO have seen great popularity in LLM post-training. In GRPO, models produce completions to a set of prompts, which are rewarded, and the po…
Hail to the Thief: Exploring Attacks and Defenses in Decentralised GRPO
Nikolay Blagoev, OÄuzhan Ersoy, Lydia Yiyu Chen
Group Relative Policy Optimization (GRPO) has demonstrated wide adoption in the post-training of Large Language Models (LLMs). In GRPO, prompts are answered by the model and prefer…
All is Not Lost: LLM Recovery without Checkpoints
Nikolay Blagoev, OÄuzhan Ersoy, Lydia Yiyu Chen
Training LLMs on decentralized nodes or on-spot instances, lowers the training cost and enables model democratization. The inevitable challenge here is the transient churns of node…
Backdoor Attacks on Decentralised Post-Training
OÄuzhan Ersoy, Nikolay Blagoev, Jona te Lintelo +3
Decentralised post-training of large language models utilises data and pipeline parallelism techniques to split the data and the model. Unfortunately, decentralised post-training c…
Go With The Flow: Churn-Tolerant Decentralized Training of Large Language Models
Nikolay Blagoev, Bart Cox, Jérémie Decouchant +1
Motivated by the emergence of large language models (LLMs) and the importance of democratizing their training, we propose GWTF, the first crash tolerant practical decentralized tra…
NoLoCo: No-all-reduce Low Communication Training Method for Large Models
Jari Kolehmainen, Nikolay Blagoev, John Donaghy +2
Training large language models is generally done via optimization methods on clusters containing tens of thousands of accelerators, communicating over a high-bandwidth interconnect…