[docs] update miles description and chinese version

README.md

@@ -112,9 +112,9 @@ See the [Customization Guide](docs/en/get_started/customization.md) for which in
 
 These are not just demos. They are independent systems that use slime as a reusable RL substrate for production-scale post-training, agentic RL, domain RL, and rollout-system research.
 
-### ⛵ Miles: Production-Focused Reinforcement Learning Framework Built on slime
+### ⛵ Miles: Enterprise-Grade Reinforcement Learning for Large-Scale Model Training
 
-[**Miles**](https://github.com/radixark/miles) builds on the foundation of slime to provide a production-focused reinforcement learning framework for large-scale model post-training. It stays closely aligned with slime's upstream development while extending it with enterprise-oriented features: deeper [SGLang](https://github.com/sgl-project/sglang) integration, operational tooling, deployment support, and optimizations for new [models](https://www.radixark.com/miles/docs/models) and [hardware](https://www.radixark.com/miles/docs/platforms). Miles also adds production features such as LoRA, TITO, and low-precision training.
+[Miles](https://github.com/radixark/miles) is an RL post-training framework for large-scale models, built on slime by [RadixArk](https://github.com/radixark). It stays closely aligned with slime's upstream development while extending it with enterprise-oriented features: deeper [SGLang](https://github.com/sgl-project/sglang) integration, operational tooling, deployment support, and optimizations for new [models](https://www.radixark.com/miles/docs/models) and [hardware](https://www.radixark.com/miles/docs/platforms). Miles also adds a growing set of production features, including LoRA, TITO, and low-precision training.
 
 ### 🔷 vime: vLLM-Native RL Post-Training Built on slime
 

README_zh.md

@@ -114,6 +114,10 @@ slime 被当作 RL 基础设施来开发，因为“脚本能跑起来”远远
 
 这些项目不只是 demo。它们是把 slime 作为可复用 RL substrate 的独立系统，覆盖生产级 post-training、agentic RL、domain RL 和 rollout-system research。
 
+### ⛵ Miles：面向大规模模型训练的企业级强化学习框架
+
+[Miles](https://github.com/radixark/miles) 是 [RadixArk](https://github.com/radixark) 基于 slime 构建的大模型 RL 后训练框架。它与 slime 上游开发保持紧密同步，同时在此基础上针对企业场景做了一系列扩展：更深度的 [SGLang](https://github.com/sgl-project/sglang) 集成、配套的运维与部署工具和服务，以及针对[新模型](https://www.radixark.com/miles/docs/models)和[新硬件](https://www.radixark.com/miles/docs/platforms)的优化。Miles 也在持续围绕真实生产环境需求迭代和进化，例如加入对 LoRA、TITO、低精度训练的支持。
+
 ### 🔷 vime: 基于 slime 的 vLLM-Native RL Post-Training 框架
 
 [**vime**](https://github.com/vllm-project/vime) 是由 vLLM 项目维护的、基于 slime 的后训练框架。它保留 slime 的 Megatron 训练栈、Data Buffer 数据流与自定义 data generation 设计，主要特点是将 rollout 后端替换为 [**vLLM**](https://github.com/vllm-project/vllm)（配合 [vllm-router](https://github.com/vllm-project/router)）。在现有 slime 启动脚本基础上仅调整 rollout 相关参数，即可快速适配 vime 进行训练。