AI & ML
TIDE: Reclaiming Lost Performance In On-Policy Reinforcement Learning
arXiv cs.DC0 views
arXiv:2611.17924v1 Announce Type: cross Abstract: Synchronous, colocated RL post-training of large language models (LLMs) today suffers from severe under-utilization due to long-tail rollouts. Inference engines fall idle waiting for long-tail rollouts or decode at low, inefficient batch sizes. To address this, we present TIDE, an RL training system that preserves on-policy training semantics while recovering resource utilization. TIDE does so by two complementary techniques: (1) dynamically repurposing rollout GPUs for training while long-tail rollouts complete and (2) dynamically migrating rollouts between inference engines to boost decoding throughput. TIDE automatically determines the right time to repurpose rollout resources through an adaptive migration policy, obtaining performance improvements that generalize across multiple applications and hardware configurations without requiring bespoke configuration. In our evaluation, TIDE provides end-to-end speedups of up to 1.36x over colocated systems, outperforming all evaluated on-policy baselines and achieving up to 97% of optimal throughput.
Source:https://arxiv.org/abs/2611.17924PreviousSIMTurn: Lightweight Divergent Subgroup Scheduling for Intra-Warp SynchronizationNextHetCCL: Efficient LLM Training on Heterogeneous Vendor GPUs
Related posts
Your Manager's Manager Has Booked a Skip Level Meeting With You
0dev.to

How to Make a Character Reference Sheet with AI (Prompts Included)
0dev.to Design
How Virtual Assistants Help Businesses Manage Their Work More Efficiently
0dev.to Productivity
DiffusionSSD: Achieving Extreme Visual Compression with Dual-stage Diffusion and Cross-level Near-Data Processing
0arXiv eess.IV
D2ToF: One-Step Diffusion for Depth Super-Resolution from Lightweight Single-Photon LiDAR Waveforms
0arXiv eess.IV
Few-Shot Open-Set Hyperspectral Pixel-Wise Classification with Spatial Aware Label Propagation
0arXiv eess.IV