Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Aug 26, 2024

• Wrap CPU model init with megatron_lazy_init_context (NVIDIA#10219)

  akoumpa

  authored
  8210e9c

• [NeMo-UX[ Move nemotron imports inline (NVIDIA#10255)

  marcromeyn

  authored
  642c97a

• Revert torchrun fix for model import (NVIDIA#10251)

  akoumpa

  authored
  8d9cfee

Commits on Aug 24, 2024

• [Audio] SSL Pretraining framework for flow-matching model for audio processing (NVIDIA#10052)

  Kuray107

  authored
  7cc99e9

Commits on Aug 23, 2024

• Lower bound transformers to support nemotron (NVIDIA#10240)

  thomasdhc

  authored
  d415621

• Add MemoryProfileCallback (NVIDIA#10166)

  ShriyaPalsamudram

  authored
  6d1be93

• [NeMo-UX] Fix a serialization bug that prevents users from moving checkpoints (NVIDIA#9939)

  ashors1

  authored
  1c90b5e

• Add tests for LazyNeMoIterator and fix case with metadata_only=True and offsets in manifest (NVIDIA#10198)

  pzelasko

  authored
  d4d6a5b

• Move trt imports in nemo.collections.llm inside respective functions (NVIDIA#10234)

  hemildesai

  authored
  753c70e

Commits on Aug 22, 2024

• Tutorial: audio codec inference (NVIDIA#10186)

  anteju

  authored
  42c2910

• Update TRTLLM 0.12 (NVIDIA#10215)

  

  meatybobby

  and

  oyilmaz-nvidia

  authored
  5269caf

• force optimizer.param_groups to match mcore_optimizer.param_groups after restoring from a checkpoint (NVIDIA#10225)

  ashors1

  authored
  78f57fe

• akoumparouli/add_check_param_hashes_across_dp_replicas (NVIDIA#9811)

  

  4 peopleauthored
  d323022

• [Draft] Nemotron in Nemo-UX (NVIDIA#10138)

  suiyoubi

  authored
  08cddbf

• Set apply_query_key_layer_scaling=False in Llama3Config (NVIDIA#10216)

  hemildesai

  authored
  9612d04

Commits on Aug 21, 2024

• Optionally disable logging in the data sampler to support predict_step (NVIDIA#10127)

  jstjohn

  authored
  ff7c614

• fix mamba convert/ add test (NVIDIA#10224)

  JRD971000

  authored
  c87e542

• Enable DDP memory optimization by default (NVIDIA#10214)

  pzelasko

  authored
  b88b662

• 24.07 vboost numbers (NVIDIA#10221)

  

  malay-nagda

  and

  erhoo82

  authored
  3a8081f

• Riva and k2 ASR WFST decoding (2) (NVIDIA#9391)

  

  3 peopleauthored
  8606350

• <SFT> <Long Sequence Length> Remove Attention Mask in SFT Chat Dataset (NVIDIA#10210)

  tomlifu

  authored
  cef98db

• [NeMo-UX] Add LLM and Lightning READMEs (NVIDIA#9894)

  ashors1

  authored
  e0ebd6f

• Add Long Context Recipe for NeMo 2.0 (NVIDIA#10140)

  BoxiangW

  authored
  90e146e

• Precision plugin recipes (NVIDIA#10187)

  maanug-nv

  authored
  aa25424

• Fix train loss broadcasting (NVIDIA#10212)

  maanug-nv

  authored
  276aaf0

Commits on Aug 20, 2024

• Add blend dataset in NeVA (NVIDIA#10000)

  

  3 peopleauthored
  8d2e43a

• Akoumparouli/nemo ux precision plugin refactor (NVIDIA#10129)

  akoumpa

  authored
  60442c2

• Set parallelism correctly when using VPP (NVIDIA#10204)

  marcromeyn

  authored
  d4f02b5

Commits on Aug 19, 2024

• Enable specifying kwargs to DDPStrategy via config for most models (NVIDIA#10046)

  pzelasko

  authored
  54458fa

• [NeMo UX] Disable softmax fusion for non-te/non-apex test (NVIDIA#10205)

  ashors1

  authored
  05c5124

• Add Mistral-NeMo 2407 configs (NVIDIA#10184)

  akoumpa

  authored
  397ca25

• [Docs] Fixing warnings, focusing on SpeechLLM docs + copy editing (NVIDIA#10141)

  erastorgueva-nv

  authored
  9c86009

• Use packaging & fall-back to pkg_resources if not available (NVIDIA#9958)

  akoumpa

  authored
  4937667

• Gemma 2 (NVIDIA#9708)

  

  3 peopleauthored
  c73f677

• fix (NVIDIA#9759) (NVIDIA#9781)

  

  github-actions[bot]

  and

  cuichenx

  authored
  cbcc222

Pagination

PreviousNext