Issues

• [RFC]: Training support and verl/vime integration for looped-transformer inference optimization

  RFC
  ThinkFlowLab/vllm-rlt#70 · Filter by author bjf-frz opened on Sep 29, 2026
  13

• [RFC]: Model support and feature roadmap for Ouro, Nanbeige4.2, and Huginn

  RFC
  ThinkFlowLab/vllm-rlt#69 · Filter by author bjf-frz opened on Sep 29, 2026
  14

  

• [FIX] Batch-selection protection leaks across batches and the Scheduler boundary

  ThinkFlowLab/vllm-rlt#67 · Filter by author kunkunblueberry opened on Sep 28, 2026

• [RFC]: Wavefront Chunked Prefill Batching (WCPB) implementation

  ThinkFlowLab/vllm-rlt#65 · Filter by author LyxWxj opened on Sep 28, 2026
  1

• [RFC]: Adaptive early exit on Ouro-1.4B: accuracy–throughput evaluation and exact-policy async exit

  ThinkFlowLab/vllm-rlt#56 · Filter by author MinhaoLi0318 opened on Sep 24, 2026
  3

• [RFC] Speculative Decoding for rlt

  ThinkFlowLab/vllm-rlt#43 · Filter by author bjf-frz opened on Sep 23, 2026
  21

• [RFC] Module boundaries and incremental refactoring of the RLT inference runtime

  RFC
  ThinkFlowLab/vllm-rlt#32 · Filter by author bjf-frz opened on Sep 18, 2026
  22

• feat: add OpenAI-compatible serving and reuse vllm bench serve

  ThinkFlowLab/vllm-rlt#22 · Filter by author hsliuustc0106 opened on Sep 12, 2026