ByteDance-Seed/VeOmni

[Feature Dev Track] Enhancing VeRL Integration

Open

#506 opened on Feb 25, 2026

View on GitHub
 (2 comments) (0 reactions) (1 assignee)Python (236 forks)auto 404
enhancementhelp wantedhf_v5road map

Repository metrics

Stars
 (2,076 stars)
PR merge metrics
 (PR metrics pending)

Description

Feature request

As VeOmni engine has been integrated in VeRL and would gradually replace the legacy FSDP engine, we need the following features to be validated or developed for smooth transition.

  • Validate existing actor profiling pipeline in VeRL is working for VeOmni engine.
  • Validate Meta-Init in VeRL for VeOmni engine.
  • Support chunked linear cross entropy for PPO to avoid monkey-patch from VeRL side. - after #458 - done in #711
  • Support router replay.
  • Support LoRA
  • Tensor hash for true on-policy RL verification and debugging.

cc @piyifan123 @FoolPlayer @Coach257 @TimYangst

Motivation

As explained

Your contribution

I will work on

  • Chunked linear cross entropy for PPO
  • Router replay
  • Tensor hash

Looking forward to volunteers on them

Contributor guide

Research direction
Study the current VeRL integration code in VeOmni and identify where to add LoRA support. Refer to the existing chunked linear cross entropy implementation as a guide.
Tech stack
pythonpytorch
Domain
machine learningdistributed systems
Issue type
Feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-2 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
Active
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
Clear
Prerequisites
PythonPyTorch
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
30