Miles v0.1: Production-Level Post-Training
Radix Ark open-sources Miles v0.1, a full-stack RL post-training framework demonstrated with asynchronous agentic RL on GLM-5.2 744B-A40B across 64 GB300 GPUs.
Miles v0.1 is a full-stack, open-source system for frontier-scale reinforcement-learning post-training, built on slime with rollout engines on SGLang and trainers supporting NVIDIA Megatron-LM and PyTorch FSDP backends plus three weight-synchronization transports. It supports full-parameter RL, LoRA RL, on-policy distillation, supervised fine-tuning, true-on-policy rollout-training alignment, and extends to diffusion models. The end-to-end case study ran fully asynchronous agentic RL on GLM-5.2 744B-A40B for terminal-use coding tasks on 64 NVIDIA GB300 GPUs with a median step time of 263 seconds over the first 30 measured steps. The code is open-sourced on GitHub.
- Rollout engines built on SGLang; trainers support Megatron-LM and PyTorch FSDP backends.
- Supports full-parameter RL, LoRA RL, on-policy distillation, SFT, and diffusion models.
- Case study: asynchronous agentic RL on GLM-5.2 744B-A40B over terminal-use coding tasks.
- Ran on 64 NVIDIA GB300 GPUs with a 263-second median step time.
- Open-sourced at github.com/radixark/miles.
- Tags
- agentic-rldistributed-trainingopen-sourcepost-trainingreinforcement-learningsglangtraining-infrastructure
- AI models
- GLM-5.2 744B-A40B
From the original article
1 min read at huggingface.co
We present Miles v0. 1, a full-stack, production-ready system for frontier post-training. Building upon the clean design of slime, Miles designs each stage of the reinforcement-learning (RL) training loop around a single principle: components should be verified, clean, and customizable.
By RadixArk, Tom Chen, Mao Cheng · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.