Commits

Commits on Jun 30, 2026

Commits on Jun 25, 2026

lint fix
hrathina
committed

Commits on Jun 24, 2026

fix: ensure CallbackManager is closed in test_close
hrathina
committed

Commits on Jun 23, 2026

feat: add class-based callback system for training lifecycle hooks
hrathina
committed

Commits on Apr 30, 2026

Rename checkpoint trigger file to be product-agnostic (#704)

RobotSail
and
claude
authored

Commits on Apr 23, 2026

Relax numba pin from >=0.62.0 to >=0.61.2 (#702)
Maxusmusti
authored

Commits on Apr 20, 2026

Commits on Mar 31, 2026

Fix trust_remote_code and gradient checkpointing for custom models (#696)

authored

Commits on Mar 24, 2026

Fix Gemma 3 SFT training by detecting dual-registered VLM configs (#695)
RobotSail
authored

Commits on Mar 6, 2026

add support for qwen3.5 vl model (#693)

RobotSail
and
claude
authored

Commits on Feb 26, 2026

Adds Validation Adds validation loss + exposes it in the API (#685)
RobotSail
authored

Commits on Feb 25, 2026

Add backwards compatibility for transformers v4.57 (#684)
Maxusmusti
authored

Commits on Feb 11, 2026

fix _no_split_modules subscript error for transformers v5 (#683)
Maxusmusti
authored

Commits on Feb 5, 2026

Add MLflow support and expose logging configuration in TrainingArgs (#680)

RobotSail
and
claude
authored

Commits on Feb 3, 2026

Fix HYBRID_SHARD failure when world_size < available GPUs (#682)
rtj1
authored

Commits on Jan 30, 2026

Add transformers v5 compatibility (#681)
Maxusmusti
authored

Commits on Jan 27, 2026

adds integration for tox-uv and updates workflows to use tox-uv (#676)

RobotSail
and
claude
authored

Commits on Jan 8, 2026

Exposes API for processing pretraining data (#672)
RobotSail
authored

Commits on Dec 18, 2025

Commits on Dec 5, 2025

Handle granite 4 as MoE models in training (#669)
mtake
authored

Commits on Nov 18, 2025

Commits on Oct 14, 2025

fix(torchrun): Omit empty arguments and correct nproc_per_node type (#661)

szaher
and
RobotSail
authored

Commits on Oct 13, 2025