# NeMo-DCR: Bit-Exact Delta-Compressed Refit for Scalable Agentic RL at Trillion-Parameter Scale

arXiv: https://arxiv.org/abs/2610.08430 · HTML: https://mlsanity.com/paper/2610.08430/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Reinforcement learning for LLM agents, one of 16 directions in LLM agents and tool use.
- Builds on 12 of its 41 references that are in the map.
- First posted on arXiv 6 Oct 2026.

## Placed in

- LLM agents and tool use / [Reinforcement learning for LLM agents](https://mlsanity.com/agents/reinforcement-learning-for-llm-agents/)

## Builds on (most cited first, 12 of 12)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Kimi K2: Open Agentic Intelligence](https://mlsanity.com/paper/2507.20534/) (2025)
- [Defeating the Training-Inference Mismatch via FP16](https://mlsanity.com/paper/2510.26788/) (2025)
- [StreamRL: Scalable, Heterogeneous, and Elastic RL for LLMs with Disaggregated Stream Generation](https://mlsanity.com/paper/2504.15930/) (2025)
- [INTELLECT-2: A Reasoning Model Trained Through Globally Decentralized Reinforcement Learning](https://mlsanity.com/paper/2505.07291/) (2025)
- [Composer 2 Technical Report](https://mlsanity.com/paper/2603.24477/) (2026)
- [ROSE: Rollout On Serving GPUs via Cooperative Elasticity for Agentic RL](https://mlsanity.com/paper/2605.06534/) (2026)
- [Understanding and Exploiting Weight Update Sparsity for Communication-Efficient Distributed RL](https://mlsanity.com/paper/2602.03839/) (2026)
- [AReaL-DTE: Sparse Policy-Weight Transfer for Online Agentic Reinforcement Learning](https://mlsanity.com/paper/2608.00455/) (2026)
- [AuroraRL: Fast, Fault-Tolerant, and Cost-Efficient Reinforcement Learning over Decentralized Network](https://mlsanity.com/paper/2602.11456/) (2026)
- [SparseRL-Sync: Lossless Weight Synchronization with ~100x Less Communication](https://mlsanity.com/paper/2605.07330/) (2026)
- [Nereus: Adaptive Parallelism for LLM Post-Training](https://mlsanity.com/paper/2609.34645/) (2026)

## Abstract (arXiv)

Agentic reinforcement learning (RL) disaggregates training from rollout, so each policy update must reach the rollout clusters before the next batch. Transferring a full 1T checkpoint for such weight synchronization (refit) takes 87.5 min between two AWS regions. Measurements of BF16 training show that about 1% of weights change their stored values per step. Recent systems exploit this sparsity but fall short on placement, exactness, or efficiency: they reimplement placement rules, assemble full tensors, rebuild values arithmetically, or use a cross-cluster collective, and none fully recovers from mid-refit failures. We present NeMo-DCR (Delta-Compressed Refit), which sends only changes yet is bit-exact: receivers obtain the same parameter and buffer bits as a dense refit. For placement, fixed affine mappings project changes from training shards into the checkpoint's canonical coordinates, residual conversion covers the other changes, and the serving runtime's native loader places all changes in receiver storage. For exactness, compressible XOR masks carry affine changes whose projection and loader preserve stored bits, and overwrites carry the others. Receivers apply both in place, retries overwrite partial writes, and a joint commit binds the policy to the baseline for the next delta. For efficiency, object storage or a relay tree streams payloads during delta construction, without a cross-cluster collective. Even at 3% and 5% change rates, NeMo-DCR refits of 30B-1T models are 12-40× faster than a transport-only full-checkpoint reference. A 1T relay-tree refit at 3% takes 150 s instead of 87.5 min, making refits practical for cross-cluster agentic RL at trillion-parameter scale.