Search results

Training stops after long checkpointing

bugSomething isn't workingSomething isn't working

community-request

Status: Open.

#1847 In NVIDIA-NeMo/RL;

· xavier-owkin opened on Jan 29, 2026

Training stalls after several hours with no noticeable error

bugSomething isn't workingSomething isn't working

Status: Open.

#1846 In NVIDIA-NeMo/RL;

· roclark opened on Jan 29, 2026

Low performance of a multi-node SFT example from README.md

bugSomething isn't workingSomething isn't working

Status: Open.

#1844 In NVIDIA-NeMo/RL;

· OlegSudakov opened on Jan 29, 2026

Sequence packing + sequence parallel causes size mismatch in DTensorPolicyWorkerV2 logprobs

bugSomething isn't workingSomething isn't working

Status: Open.

#1835 In NVIDIA-NeMo/RL;

· RayenTian opened on Jan 28, 2026

MLFLow: Metrics are not reported to <code>experiment_name</code> consistently.

bugSomething isn't workingSomething isn't working

community-request

needs-follow-upIssue needs follow-upIssue needs follow-up

Status: Open.

#1832 In NVIDIA-NeMo/RL;

· dhineshkumar-r opened on Jan 27, 2026

Save checkpoint without optimizer states

community-request

enhancementNew feature or requestNew feature or request

Status: Open.

#1828 In NVIDIA-NeMo/RL;

· odedovadia opened on Jan 26, 2026

On-policy distillation with different tokenizers

enhancementNew feature or requestNew feature or request

x-shop

Status: Open.

#1827 In NVIDIA-NeMo/RL;

· snowmanwwg opened on Jan 26, 2026

Bump cudnn to 9.14 to support packed sequence fused attention

enhancementNew feature or requestNew feature or request

Status: Open.

#1819 In NVIDIA-NeMo/RL;

· guyueh1 opened on Jan 23, 2026

Online DPO Support

community-request

enhancementNew feature or requestNew feature or request

needs-follow-upIssue needs follow-upIssue needs follow-up

Status: Open.

#1816 In NVIDIA-NeMo/RL;

· dhineshkumar-r opened on Jan 23, 2026

Async data collector is single-epoch

bugSomething isn't workingSomething isn't working

Status: Open.

#1814 In NVIDIA-NeMo/RL;

· smahdavi4 opened on Jan 23, 2026

config option to disable monotonicity checks and on policy token id correction

enhancementNew feature or requestNew feature or request

Status: Open.

#1812 In NVIDIA-NeMo/RL;

· cmunley1 opened on Jan 22, 2026

GPT-OSS on-policy distillation fails during checkpoint load (Missing key: model.layers.0.mlp.experts.down_proj)

bugSomething isn't workingSomething isn't working

community-request

t-gptoss

Status: Open.

#1810 In NVIDIA-NeMo/RL;

· uygarmv opened on Jan 22, 2026