← Back to all news
arXiv Machine Learning By Dezhen Lu, Zixiang Ni, Runguo Li

ScaleRLT: Exact Log-Probabilities and Fast Deferred Rollouts for Reinforcement Learning on Looped Transformers

Read the original on arXiv Machine Learning →

ScaleRLT introduces a hybrid learner log‑probability method that reconciles differences between dense and deferred rollout computations in looped language models like Ouro. By using a stop‑gradient replay of the rollout schedule, the approach achieves bit‑for‑bit reproducibility across FP32 and bf16 real‑weight cells while maintaining the same gradient accuracy as dense implementations. The technique boosts rollout throughput by up to 1.56× on a single GPU and significantly speeds up multi‑node GRPO updates, leading to substantial accuracy gains on GSM8K in short training runs.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

Related stories

arXiv Machine Learning
6d ago

End-to-End MXFP8 Reinforcement Learning on Blackwell: What Matching Buys, When It Pays Off, and How It Fails

The paper presents an end‑to‑end MXFP8 reinforcement learning implementation on Blackwell GPUs, evaluating its benefits and limitations with Qwen3-8B and Qwen3-30B-A3B models. It finds that matching FP8 precision between sampler and learner improves consistency, that speed gains appear only for large micro‑batches, and that several weight‑synchronization correctness conditions can fail silently. The study also introduces seven explicit checks to detect such failures during training.

By Wengee Zhang
llmsreinforcement-learningefficiency
More like this →
arXiv Machine Learning
6d ago

Error Diffusion Replaces Latent Weights: Training Binary-Weight Language Models on a Few Bits of Dithered State

The paper introduces an error‑diffusion (ED) optimizer that replaces the full‑precision latent copy of binary‑weight language models with a small 2‑8‑bit residual buffer, reducing training state per weight from 97 bits to 3‑9 bits—up to 19× less. ED accumulates unrealized weight updates in the buffer and flips the binary weight when the accumulated value crosses a threshold aligned with the quantization grid, enabling efficient training on large transformers with minimal loss compared to latent‑weight baselines. The method scales to models with up to 19 M parameters on enwik8 and 70 M on wikitext‑103, maintaining performance within a few percent of reference while allowing on‑device adaptation within a 275 MB memory budget.

By Pierre-Olivier Bonin
llmsragdiffusionefficiencybenchmarkssafety
More like this →
arXiv Machine Learning
1d ago

bf16 Rounding Silently Discards Most RL Policy Updates: A Hidden Training-Inference Mismatch and an Unbiased One-Line Fix

The paper identifies a hidden mismatch in reinforcement‑learning fine‑tuning of large language models: when training uses fp32 master weights but rollouts are served from a bf16 copy produced by round‑to‑nearest conversion, most per‑step updates are too small to affect the served weights, causing a lag of 50–150 steps. The authors propose Fair Cast, a one‑line change that exports the master weights with an unbiased rounding rule (stochastic rounding or error feedback) and computes the loss on the served copy, eliminating the lag without adding hyperparameters and incurring only a modest computational cost. Across multiple models and optimizers, Fair Cast consistently outperforms round‑to‑nearest, improves checkpoint consistency, and reveals the implicit regularization effect of the lag.

By Andrew Gao, William Gao
llmsreinforcement-learning
More like this →