NVIDIA Releases Polar, a Token-Faithful Rollout Framework for GRPO Training Across Codex, Claude Code, and Qwen Code
Quick Answer
NVIDIA has launched Polar, a rollout framework that enhances GRPO training for language agents without altering their harnesses.
Quick Take
NVIDIA has launched Polar, a rollout framework that enhances GRPO training for language agents without altering their harnesses. By utilizing a model API proxy, Polar significantly boosts performance on , achieving a 22.6-point increase under Codex, 4.8 points under Claude Code, and 6.2 points under Pi, and is available as a NeMo Gym environment.
Key Points
- Polar captures token-level interactions for training language agents.
- Improves SWE-Bench Verified pass@1 by 22.6 points under Codex harness.
- Achieves 4.8-point increase under Claude Code and 6.2 points under Pi.
- Framework registered as a NeMo Gym environment.
- Released under the ProRL Agent Server repository.
Article Excerpt
From source RSS / original summaryNVIDIA researchers have introduced Polar, a rollout framework that trains language agents using reinforcement learning without modifying their agent harnesses. Polar places a model API proxy between the harness and the inference server, capturing token-level interactions and reconstructing trainer-ready trajectories. Using GRPO on a Qwen3. 5-4B base model, Polar improves Verified pass@1 by 22. 6 points under the Codex harness, 4. 8 points under Claude Code, and 6. 2 points under Pi.
The framework is registered as a NeMo Gym environment and released under the ProRL Agent Server repository. The post NVIDIA Releases Polar, a Token-Faithful Rollout Framework for GRPO Training Across Codex, Claude Code, and Qwen Code appeared first on MarkTechPost.
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from MarkTechPost
See more →Meet Flash-KMeans: An IO-Aware, Exact K-Means That Runs Over 200× Faster Than FAISS on GPUs
Flash-KMeans is an open-source, IO-aware k-means implementation that operates over 200× faster than FAISS on NVIDIA H200 GPUs. It achieves 17.9× end-to-end and 33× speedup over cuML by optimizing distance calculations and updating mechanisms without approximating results. This advancement significantly enhances performance for data scientists and machine learning practitioners.