Issues

• Question: can scaling efficiency be estimated before running at larger scale?

  huggingface/nanotron#417 · Filter by author alexbuiko-sketch opened on Aug 24, 2026

• The qwen model is throwing an error when performing pipeline parallel processing

  huggingface/nanotron#413 · Filter by author liguodongiot opened on Jul 7, 2026

• When the pipeline parallelism is enabled, the loss_avg log reports an error.

  huggingface/nanotron#405 · Filter by author liguodongiot opened on Jul 3, 2026

• ZeRO-3 (full parameter + gradient sharding) support

  huggingface/nanotron#398 · Filter by author NeuroCoder47 opened on Mar 5, 2026

• Segmentation fault on nanotron import

  huggingface/nanotron#391 · Filter by author Godofnothing opened on Nov 9, 2025

• Log Error？

  huggingface/nanotron#390 · Filter by author JunkFood436 opened on Oct 18, 2025

• Quick start and Installation errors in docker container

  huggingface/nanotron#389 · Filter by author scoleri-mr opened on Sep 24, 2025

• Can you Support Flash Attention 3?

  huggingface/nanotron#388 · Filter by author LZY-the-boys opened on Aug 8, 2025

• Why only S3 paths are supported for consumption stats

  huggingface/nanotron#387 · Filter by author LZY-the-boys opened on Aug 6, 2025

• [Bug] Error when running example: AttributeError: 'ModelArgs' object has no attribute 'model'

  huggingface/nanotron#386 · Filter by author JessePrince opened on Aug 5, 2025
  5

• Small Issue in the Installation Instructions

  huggingface/nanotron#384 · Filter by author EliMCosta opened on Jul 29, 2025

• Context Parallelism

  huggingface/nanotron#383 · Filter by author kd303 opened on Jul 18, 2025
  2

• Getting UnionMatchError error when trying to run official examples in SmolLM

  huggingface/nanotron#371 · Filter by author rekcu opened on May 30, 2025
  5

• [BUG] Bug in Llama inference

  huggingface/nanotron#369 · Filter by author duynht opened on May 26, 2025

• Request for Clarification on Validation in Nanotron

  huggingface/nanotron#351 · Filter by author tpavankalyan opened on Apr 16, 2025

• Confusing variable name in PP configuration assert fail

  huggingface/nanotron#347 · Filter by author wirthFelix opened on Apr 11, 2025
  1

• supporting Qwen-2.5 series ？

  huggingface/nanotron#333 · Filter by author JoshonSmith opened on Apr 2, 2025
  1

• Error when initializing DistributedTrainer

  huggingface/nanotron#302 · Filter by author manuelbrack opened on Mar 24, 2025

• Any plans for supporting Qwen-2.5 series models?

  huggingface/nanotron#296 · Filter by author SinclairCoder opened on Mar 18, 2025

• why my nanoset is empty

  huggingface/nanotron#293 · Filter by author ziyanxzy opened on Mar 11, 2025

• Make sure nanotron works with lighteval

  huggingface/nanotron#276 · Filter by author xrsrke opened on Jan 28, 2025

• What is the purpose of "task"

  huggingface/nanotron#273 · Filter by author laiviet opened on Jan 23, 2025
  1

• half grads don't necessarily get released before the next forward

  help wantedHigh Priority
  huggingface/nanotron#272 · Filter by author nmntz opened on Jan 22, 2025
  1

• Memory leak with Nanoset

  huggingface/nanotron#271 · Filter by author eliebak opened on Jan 20, 2025
  1

  

• Missing v0.5 release

  huggingface/nanotron#268 · Filter by author thomas-bouvier opened on Jan 15, 2025