[PR #810] [MERGED] [RL] Add dummy example of RL training on mooncake store #1838

New issue

Closed

opened 2026-10-29 21:30:45 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-29 21:30:45 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/kvcache-ai/Mooncake/pull/810
Author: @Risc-lt
Created: 9/4/2025
Status: ✅ Merged
Merged: 9/8/2025
Merged by: @stmatengss

Base: main ← Head: feat/rl

📝 Commits (3)

d3e15c7 feat: init rl training using store for rollout
5685d4b chore: update structure
03adab4 fix: object store setup and training

📊 Changes

1 file changed (+450 additions, -0 deletions)

View changed files

➕ mooncake-rl/examples/rl_samples.py (+450 -0)

📄 Description

Create a dummy example for RL training based on mooncake store. This example is intended to demonstrate the basic design of integrating mooncake into slime. In the next steps, we will extend this foundation to build a distributed runtime on mooncake, aiming to achieve lower latency and a more flexible architecture than Ray.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/kvcache-ai/Mooncake/pull/810 **Author:** [@Risc-lt](https://github.com/Risc-lt) **Created:** 9/4/2025 **Status:** ✅ Merged **Merged:** 9/8/2025 **Merged by:** [@stmatengss](https://github.com/stmatengss) **Base:** `main` ← **Head:** `feat/rl` --- ### 📝 Commits (3) - [`d3e15c7`](https://github.com/kvcache-ai/Mooncake/commit/d3e15c7db1f0accb16b9fb7a21570913d6911427) feat: init rl training using store for rollout - [`5685d4b`](https://github.com/kvcache-ai/Mooncake/commit/5685d4bb1513dc1761dfbe12aef13cab8ba51755) chore: update structure - [`03adab4`](https://github.com/kvcache-ai/Mooncake/commit/03adab41cb679a331e0d44e9b85827ac1fde54dc) fix: object store setup and training ### 📊 Changes **1 file changed** (+450 additions, -0 deletions) <details> <summary>View changed files</summary> ➕ `mooncake-rl/examples/rl_samples.py` (+450 -0) </details> ### 📄 Description Create a dummy example for RL training based on mooncake store. This example is intended to demonstrate the basic design of integrating mooncake into slime. In the next steps, we will extend this foundation to build a distributed runtime on mooncake, aiming to achieve lower latency and a more flexible architecture than Ray. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-29 21:30:45 +00:00