vime
High-Performance Training
Supports efficient training in various modes by connecting Megatron with vLLM.
Megatron⇄vLLM
A LITTLE SLIME. A LOT OF POSSIBILITY.
Built for
scaling RL.
vime is an LLM post-training framework for RL scaling, providing two core capabilities:
Build your first experiment ↗ Explore. Undo. Try again.Flexible Data Generation
Enables arbitrary training data generation workflows through custom data generation interfaces and server-based engines.
Your data & environments→Data Buffer
Training, rollout, custom generation, rewards, verifiers, and environment interaction flow through the same training / rollout / Data Buffer path.
BUILT IN THE OPEN. USED AT SCALE.
From GLM-4.5 to GLM-5.3 & GLM-5.3-Flash.
slime has powered RL post-training across Z.ai’s GLM generations. Explore the model releases, then use the same core capabilities in your own experiment.
01 / THE EXPERIMENT BUILDER
Make it yours.
A starting point?
02 / FROM PLAY TO RUN
Your experiment, to go.
A starting recipe grounded in this repository. Prepare the environment, review the assumptions, then run.
Three steps to your first reward
GPU counts describe placement, not a memory-fit guarantee. These generated combinations have not been benchmarked on your hardware. Start with a short run and compare reward, log-prob mismatch, clipping, and rollout time.
04 / WHEN YOU WANT TO GO FURTHER
A good experiment asks good questions.
Open a question. Find the next measurement, then the right tool.
01 / DEBUGThe reward looks wrong. Where do I start?＋
Inspect generated text, checkpoint loading, and the chat template first. Save a rollout, replay a fixed batch on the training side, then compare log-probs before changing the algorithm.
Debug & replay →Understand mismatch →02 / PROFILETraining is slow. What am I waiting for?＋
Separate rollout, reward, weight transfer, and training time. Trace the critical path, then profile the busy engine. Use tools/profile_rollout.py to collect vLLM traces and inspect prefill/decode behavior.
Profiling →Timeline & tracing →Metrics →03 / SCALEHow do I grow beyond one cluster?＋
Split GPU ownership, then choose a weight transport that matches the network. Add PD when prefill and decode need different resources; use persistent rollout storage when jobs need recovery.
External engines →Delta sync →Partial / distributed async →straw →Recovery →04 / EXTENDMy agent needs its own world.＋
Start with a custom per-sample generation function and reward. Replace rollout orchestration only when your data scheduling or cross-step state needs it.
Agent RL →Extension interfaces →vLLM topology →