Search results

Setting for testing Nemotron-H throughput

Status: Open.

#13962 In NVIDIA/NeMo;

· iamtonymwt opened on Jun 18, 2025

Issue with Model Evaluation: Metrics of Finetuned Model Are Identical to Pretrained Model

Status: Open.

#13948 In NVIDIA/NeMo;

· tsugumi-sys opened on Jun 17, 2025

Incorrect text output in BeamCTCInfer.flashlight_beam_search due to using tokens instead of words

bugSomething isn't workingSomething isn't working

Status: Open.

#13947 In NVIDIA/NeMo;

· khanh14ph opened on Jun 17, 2025

hf automodel train script fails with --enable-grad-ckpt or --sequence-parallel option enabled

bugSomething isn't workingSomething isn't working

Status: Open.

#13930 In NVIDIA/NeMo;

· SoundProvider opened on Jun 16, 2025

Megatron FSDP doesn't work with HF checkpoint

bugSomething isn't workingSomething isn't working

Status: Open.

#13913 In NVIDIA/NeMo;

· jiuqiant opened on Jun 13, 2025

Questions about logged throughput metrics

Status: Open.

#13903 In NVIDIA/NeMo;

· Proyag opened on Jun 12, 2025

DuplexS2SModel

Status: Open.

#13902 In NVIDIA/NeMo;

· sankar-mukherjee opened on Jun 12, 2025

hugging face saved model inference

Status: Open.

#13891 In NVIDIA/NeMo;

· riyajatar37003 opened on Jun 11, 2025

Adjusting "global_batch_size" and "micro_batch_size" has no impact on how long each training step takes when using HFAutoModel.

bugSomething isn't workingSomething isn't working

Status: Open.

#13887 In NVIDIA/NeMo;

· jiuqiant opened on Jun 11, 2025

Is Qwen3 pretraining architectural features fully supported now?

Status: Open.

#13878 In NVIDIA/NeMo;

· tjoymeed opened on Jun 10, 2025

Q: Forced alignment with streaming

Status: Open.

#13863 In NVIDIA/NeMo;

· WeiXiaoSummer opened on Jun 7, 2025

Loading .nemo model throws unhandled exception

bugSomething isn't workingSomething isn't working

Status: Open.

#13858 In NVIDIA/NeMo;

· ArthurBook opened on Jun 6, 2025