huggingface/accelerate

Pull Requests Commits

feat: allow device_map auto to work with XLA devices

tengomucho committed 1 year ago

396bb23e

Allow notebook_launcher to launch to multiple GPUs from Colab (#2561)

StefanTodoran committed 1 year ago

Verified 4d415f21

[docs] Fix kwarg docstring (#2590)

stevhliu committed 1 year ago

Verified 829171a9

Expound PartialState docstring (#2589)

muellerzr committed 1 year ago

Verified 5a232de2

Guard stateful objects (#2572)

muellerzr committed 1 year ago

Verified 5f8048cd

Fix `load_checkpoint_in_model` behavior when unexpected keys are in the checkpoint (#2588)

fxmarty committed 1 year ago

Verified 4378b560

Refactor and improve model estimator tool (#2581)

muellerzr committed 1 year ago

Verified 8644e23b

Refactor affinity and make it stateful (#2579)

muellerzr committed 1 year ago

Verified b2fc3a3b

Update data_loader.py to Ensure Reproducibility in Multi-Process Environments with Dataloader Shuffle (#2584)

universuen committed 1 year ago

Verified 290446d4

[docs] Missing functions from API (#2580)

stevhliu committed 1 year ago

Verified 85a75d4c

Allow for custom deepspeed env files (#2566)

muellerzr committed 1 year ago

Verified f94f0ff9

Rm uv install (#2577)

muellerzr committed 1 year ago

Verified 1b2e6349

Unpin deepspeed (#2570)

muellerzr committed 1 year ago

Verified dd62fc90

Allow for setting deterministic algorithms (#2569)

muellerzr committed 1 year ago

Verified 10b41849

Improve deepspeed env gen (#2565)

muellerzr committed 1 year ago

Verified c2f193a2

Add log message for RTX 4000 series when performing multi-gpu inference with device_map (#2557)

SunMarc committed 1 year ago

Verified 18121523

Add NUMA affinity control for NVIDIA GPUs (#2535)

muellerzr committed 1 year ago

Verified b8b353b7

Add Cambricon MLU accelerator support (#2552)

huismiling committed 1 year ago

Verified f2778d65

🚨🚨🚨Move to using tags rather than latest for docker images and consolidate image repos 🚨 🚨🚨 (#2554)

muellerzr committed 1 year ago

Verified 2ad42e77

Include working driver check (#2558)

muellerzr committed 1 year ago

Verified e8aaee5d

split_between_processes for Dataset (#2433)

geronimi73 committed 1 year ago

Verified 910c1b6a

Add mapping main_process_ip and ip-master_addr when not using standard as deepspeed launcher (#2495)

asdfry committed 1 year ago

Verified 92d3240b

Fix test_script.py on TPU v2/v3 (#2542)

vanbasten23 committed 1 year ago

Verified 02a8a9a3

muellerzr committed 1 year ago

ee163b66

Use uv instead of pip install for github CI (#2546)

muellerzr committed 1 year ago

Verified 354db5b5

Update FSDP mixed precision setter to enable fsdp+qlora (#2544)

pacman100 committed 1 year ago

Verified 92b1ad01

Allow Gradients to be Synced Each Data Batch While Performing Gradient Accumulation (#2531)

fabianlim committed 1 year ago

Verified 60bfdaa9

Remove extra double-dash in error message (#2541)

drscotthawley committed 1 year ago

Verified 16eb6d76

[docs] Troubleshoot (#2538)

stevhliu committed 1 year ago

Verified c8acfa70

Overdue email change... (#2534)

muellerzr committed 1 year ago

Verified e70e3c87

Older