Issues

• [support] Was length_penalty (DAPO overlong / ProRL stop-properly) intentionally left out of the OpenRLHF port?

  NVIDIA-NeMo/labs-molt#193 · Filter by author jasonchen505 opened on Oct 7, 2026
  1

• Your project is on StackMap — a curated map of the AI stack

  NVIDIA-NeMo/labs-molt#146 · Filter by author hoghweed opened on Sep 27, 2026

• [feature] OrcaRouter provider support for Molt

  NVIDIA-NeMo/labs-molt#104 · Filter by author nissrin2020ali-ux opened on Sep 7, 2026

• Please use MOLT/AutoModel with CUDA 13+

  NVIDIA-NeMo/labs-molt#101 · Filter by author hijkzzz opened on Sep 1, 2026

• [feature] Quickstart scripts missing prepare_math

  NVIDIA-NeMo/labs-molt#95 · Filter by author bogdansalyp opened on Aug 24, 2026

• [bug] Quickstart Qwen3.6 SFT crash with metadata shape mismatch

  bug
  NVIDIA-NeMo/labs-molt#90 · Filter by author bogdansalyp opened on Aug 21, 2026
  3

• [feature] DeepSeekV4 + FP8 RL

  NVIDIA-NeMo/labs-molt#59 · Filter by author hijkzzz opened on Jul 31, 2026

• Are community contributions welcome?

  NVIDIA-NeMo/labs-molt#51 · Filter by author NancyFyong opened on Jul 30, 2026
  3

• [feature] Roadmap for DeepSeek-V4 RL training support (DSA/MLA kernels, mixed precision, GPU footprint, and the agentic environment path)

  NVIDIA-NeMo/labs-molt#10 · Filter by author hyfine opened on Jul 16, 2026
  2