Skip to content

Sync/publish v0.2.1 GitHub#366

Merged
PanAndy merged 33 commits intomainfrom
sync/publish_v0.2.1_github
Mar 9, 2026
Merged

Sync/publish v0.2.1 GitHub#366
PanAndy merged 33 commits intomainfrom
sync/publish_v0.2.1_github

Conversation

@PanAndy
Copy link
Collaborator

@PanAndy PanAndy commented Mar 9, 2026

大家好!感谢大家对ROLL的关注。
ROLL近期更新了大量新功能,以下是近期更新的一些梳理,我们将持续对ROLL进行迭代更新,欢迎加入ROLL的社区。

🚀亮点:

  • rollout 重构为由router调度,支持sglang-router
  • 新增[On-Policy Distillation](docs_roll/i18n/zh-Hans/docusaurus-plugin-content-docs/current/User Guides/Pipeline/on_policy_distill_pipeline_start.md)训练支持
  • 支持Qwen3.5 Dense / MoE 系列模型

🚀主要新特性:

  • Rollout
    • 重构router调度支持
      • sglang strategy重构,同时支持engine、server两种模式。
      • Scheduler重构(rlvr的DynamicScheduler/ agentic的Rolloutscheduler),统一由Router提供调度
      • 迁移原LoadBalancer、RequestScheduler为PromptAffinityRouter、EnvAffinityRouter
      • 新增支持sglang-router
  • pipeline recipe
    • 新增On-Policy Distillation训练支持
  • Models
    • 支持Qwen3.5 Dense/MoE系列模型
  • docker
    • torch2.10 、vllm 0.16.0 nightly、vllm0.15.1版本 、mcore 0.16.0
  • bug fix:
    • 默认设置vllm VLLM_USE_FLASHINFER_SAMPLER=0 for torch 280,解决reponse重复度过高
    • fix sglang & vllm 偶现port conflict
    • fix sglang multi-nodes fail when infer_dp > 1
    • fix reward worker metrics 透出能力
    • fix model download get_node_ip cache,可能导致死锁timeout
    • fix CPU offload时FSDP2 DCE save
    • fix FSDP2 model initialization casting

PanAndy and others added 30 commits March 6, 2026 15:22
@PanAndy
Copy link
Collaborator Author

PanAndy commented Mar 9, 2026

#360

@PanAndy
Copy link
Collaborator Author

PanAndy commented Mar 9, 2026

#363

@PanAndy PanAndy merged commit 2eba7c3 into main Mar 9, 2026
6 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

7 participants