Quentin Gallouédec 4ebc726849 Skip _no_split_modules classes the model doesn't contain when auto-wrapping (#4384)

2026-10-09 17:02:48 +02:00

..

deepspeed

Fix nightly CI: tensor parallel size detection and DeepSpeed warmup steps (#4229)

2026-09-09 14:49:52 +02:00

fsdp

Skip _no_split_modules classes the model doesn't contain when auto-wrapping (#4384)

2026-10-09 17:02:48 +02:00

test_configs

Remove duplicate debug key from SageMaker config fixture (#4302)

2026-09-21 14:23:31 +02:00

test_samples

Introduce a pod-config command (#802)

2022-11-01 10:00:48 -04:00

tp

Fix all tests (#4072)

2026-06-10 23:38:57 +02:00

__init__.py

Torchao float8 training (#3348)

2025-02-17 11:51:47 -05:00

test_accelerator.py

Fix unwrap_model(keep_fp32_wrapper=False) breaking hooked models (#4255)

2026-09-25 18:30:16 +02:00

test_activation_checkpointing_offload.py

Add FSDP2 activation_checkpointing_offload: offload checkpointed layer inputs to pinned CPU memory (#4175)

2026-09-01 14:11:30 +02:00

test_big_modeling.py

Fix accelerator dispatch test on minimum PyTorch (#4336)

2026-09-25 18:32:08 +02:00

test_cli.py

Stop accelerate launch from forcing torch.compile dynamic=False (#4254)

2026-09-25 17:31:37 +02:00

test_compile.py

Fix compile_regions running the uncompiled module (#4188)

2026-09-07 18:32:51 +02:00

test_cpu.py

Add test for ops and fix reduce (#1122)

2023-02-28 09:18:09 -05:00

test_data_loader.py

Fix metric padding removal after skip_first_batches (#4235)

2026-10-01 15:31:42 +02:00

test_dataclasses.py

Deepspeed Ulysses/ALST integration (#3817)

2025-11-20 18:24:00 +01:00

test_examples.py

fixing ci (#3948)

2026-02-27 22:24:07 +01:00

test_fp8.py

Fix convert_model_to_fp8_ao converting the first and last linear layers (#4147)

2026-08-31 17:40:58 +02:00

test_grad_sync.py

HPU support (#3378)

2025-03-11 11:16:57 -04:00

test_hooks.py

enable 2 model hook ut cases on XPU (#3774)

2025-09-17 01:32:59 +02:00

test_imports.py

Do not import transformer_engine on import (#3056)

2024-08-28 09:06:13 -04:00

test_kwargs_handlers.py

HPU support (#3378)

2025-03-11 11:16:57 -04:00

test_launch.py

Fix notebook_launcher to propagate mixed_precision to the workers (#4331)

2026-09-29 16:37:55 +02:00

test_load_checkpoint_and_dispatch_with_broadcast.py

Fix gloo backend error in test_load_checkpoint_and_dispatch_with_broadcast on XPU (#4056)

2026-06-01 18:09:33 +02:00

test_logging.py

Fix logging logic when in_order is set to True (#3280)

2026-03-02 16:27:29 +01:00

test_memory_utils.py

enhance new codes to XPU, and make them be device agnostic (#3890)

2026-01-27 13:30:46 +01:00

test_metrics.py

HPU support (#3378)

2025-03-11 11:16:57 -04:00

test_modeling_utils.py

Fix loading checkpoints with modules offloaded to disk (#4352)

2026-09-29 17:50:25 +02:00

test_multidevice.py

switch XPU ccl backend to torch-builtin xccl in test_zero3_integration (#3773)

2025-09-18 11:50:32 +02:00

test_offload.py

fix: handle FP8 dtypes in disk offload write/read path (#4151)

2026-09-07 18:32:05 +02:00

test_optimizer.py

HPU support (#3378)

2025-03-11 11:16:57 -04:00

test_quantization.py

Fix VRAM leak from cached int8 weights during CPU/disk offload (#4325)

2026-09-25 15:55:21 +02:00

test_sagemaker.py

Add copyright + some ruff lint things (#2523)

2024-03-04 09:14:31 -05:00

test_scheduler.py

Bump ruff to 0.11.2 (#3471)

2025-04-01 11:57:06 +02:00

test_state_checkpointing.py

Bump ruff to 0.11.2 (#3471)

2025-04-01 11:57:06 +02:00

test_tpu.py

Remove unnecessary env=os.environ.copy()s (#2449)

2024-03-06 06:36:56 -05:00

test_tracking.py

Fix MLFLOW_NESTED_RUN never being able to turn nesting off (#4223)

2026-09-08 14:57:37 +02:00

test_utils.py

Do not import torch.distributed in model_has_dtensor on a non-distributed build (#4250)

2026-09-25 17:31:43 +02:00

xla_spawn.py

Bump ruff to 0.11.2 (#3471)

2025-04-01 11:57:06 +02:00