Skip to content

Pull requests: huggingface/accelerate

New pull request New
24 Open 1,604 Closed
24 Open 1,604 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

latest bnb no longer has optim_args attribute on optimizer
#3311 opened Dec 23, 2024 by winglian Loading…
5 tasks
1
[FSDP] Make env var and dataclass flag consistent for cpu_ram_efficient_loading
#3307 opened Dec 19, 2024 by SumanthRH Loading…
1 of 5 tasks
Fix 'AdamW' object has no attribute 'optim_bits' error when deepspeed zero2 is enabled
#3305 opened Dec 19, 2024 by bursteratom Loading…
[tests] make cuda-only test work on other hardware accelerators
#3302 opened Dec 18, 2024 by faaany Loading…
2
[tests] remove require_non_xpu test markers
#3301 opened Dec 18, 2024 by faaany Loading…
2
Integrate DeepSpeed's new HF interface for sequence parallelism
#3299 opened Dec 16, 2024 by ronald-d-rogers Loading…
1 of 5 tasks
[FSDP][Example] Fix FSDP example for use_orig_params=False
#3298 opened Dec 16, 2024 by SumanthRH Loading…
2 of 5 tasks
2
[docs] fix typo, change "backoff_filter" to "backoff_factor"
#3296 opened Dec 16, 2024 by suchot Loading…
1 of 5 tasks
1
Add keep_torch_compile param to unwrap_model and extract_model_from_parallel for distributed compiled model.
#3282 opened Dec 9, 2024 by ggoggam Loading…
5 tasks done
23
Fix logging logic when in_order is set to True
#3280 opened Dec 9, 2024 by yuxinyuan Loading…
2 of 5 tasks
5
Fix : get_balanced_memory when using multi gpus with small models or quantized models with a large vocabulary
#3244 opened Nov 18, 2024 by MekkCyber Loading…
1 of 5 tasks
1
7
[RFC] Support FSDP2
#3231 opened Nov 8, 2024 by kmehant Loading…
1 of 5 tasks
1
4
Fix slurm multinode example
#3229 opened Nov 8, 2024 by ffrancesco94 Loading…
3 of 5 tasks
1
3
create _preprare_fsdp to pre- prepare fsdp model training
#3213 opened Nov 3, 2024 by eljandoubi Loading…
1
14
Distributed inference example for llava_next
#3179 opened Oct 20, 2024 by VladOS95-cyber Loading…
3 of 5 tasks
43
feat: support tensor parallel & Data loader
#3173 opened Oct 16, 2024 by kmehant Loading…
1 of 5 tasks
27
🚀 [Feature] Add deprecated Decorator
#3161 opened Oct 12, 2024 by yhna940 Loading…
17
add support for custom function for reducing the batch size wip Work in progress
#3071 opened Sep 3, 2024 by winglian Loading…
5 tasks
8
Some adjustment for supporting Deepspeed-Ulysses wip Work in progress
#2877 opened Jun 20, 2024 by zeyugao Loading…
1 of 5 tasks
12
handle weight sharing with init_on_device wip Work in progress
#2737 opened May 3, 2024 by aws-rhsoln Loading…
5 tasks
6
[DO NOT MERGE] add all level buffer support when computing infer_auto_device_map enhancement New feature or request feature request Request for a new feature to be added to Accelerate wip Work in progress
#2663 opened Apr 12, 2024 by SunMarc Loading…
3
make big model inference compatible with torch.compile wip Work in progress
#2609 opened Apr 2, 2024 by SunMarc • Draft
1
2
allocate 80% for cpu is unset in max_memory enhancement New feature or request feature request Request for a new feature to be added to Accelerate wip Work in progress
#2219 opened Dec 5, 2023 by SunMarc Loading…
2 tasks
6
Enable cpu offload with weights inside the module wip Work in progress
#2214 opened Dec 4, 2023 by SunMarc • Draft
10
ProTip! Adding no:label will show everything without a label.