Search results

[bug] Every shipped MoE recipe defaults to ep_size > 1 + gradient_checkpoint full, which fails at step 0

Status: Open.

#60 In NVIDIA-NeMo/labs-molt;

· NancyFyong opened on Jul 31, 2026

[feature] DeepSeekV4 + FP8 RL

Status: Open.

#59 In NVIDIA-NeMo/labs-molt;

· hijkzzz opened on Jul 31, 2026

Are community contributions welcome?

Status: Open.

#51 In NVIDIA-NeMo/labs-molt;

· NancyFyong opened on Jul 30, 2026

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

Status: Open.

#10 In NVIDIA-NeMo/labs-molt;

· hyfine opened on Jul 16, 2026