Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Avoid saving checkpoints in diffusers convergence test #1038

Merged

Conversation

yeounoh
Copy link
Contributor

@yeounoh yeounoh commented Jan 9, 2024

Description

Saving the checkpoints during training with full 833 batch size ran into OOM.

Tests

tested locally,

INFO:__main__:***** Running training *****
INFO:__main__:  Num examples = 833
INFO:__main__:  Num Epochs = 1
INFO:__main__:  Instantaneous batch size per device = 1
INFO:__main__:  Total train batch size (w. parallel, distributed & accumulation) = 1
INFO:__main__:  Gradient Accumulation steps = 1
INFO:__main__:  Total optimization steps = 1
Steps:   0%|                                                                                                              | 0/1 [00:00<?, ?it/s]/usr/local/lib/python3.8/site-packages/transformers/deepspeed.py:23: FutureWarning: transformers.deepspeed module is deprecated and will be removed in a future version. Please import deepspeed modules directly from transformers.integrations
  warnings.warn(
Steps: 100%|██████████████████████████████████████████████████████████████████████████| 1/1 [01:44<00:00, 104.08s/it, lr=1e-5, step_loss=0.0488]


INFO:root:Joining rendezvous 'accelerate.utils.wait_for_everyone'...
{'image_encoder', 'requires_safety_checker'} was not found in config. Values will be initialized to default values.
                                                                                                                                               Loaded feature_extractor as CLIPImageProcessor from `feature_extractor` subfolder of CompVis/stable-diffusion-v1-4.        | 0/7 [00:00<?, ?it/s]
Loaded tokenizer as CLIPTokenizer from `tokenizer` subfolder of CompVis/stable-diffusion-v1-4.
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["id2label"]` will be overriden.
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["bos_token_id"]` will be overriden.
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["eos_token_id"]` will be overriden.
Loaded safety_checker as StableDiffusionSafetyChecker from `safety_checker` subfolder of CompVis/stable-diffusion-v1-4.
                                                                                                                                               {'prediction_type', 'timestep_spacing'} was not found in config. Values will be initialized to default values.     | 6/7 [00:00<00:00, 13.69it/s]
Loaded scheduler as PNDMScheduler from `scheduler` subfolder of CompVis/stable-diffusion-v1-4.
Loading pipeline components...: 100%|█████████████████████████████████████████████████████████████████████████████| 7/7 [00:00<00:00, 15.89it/s]
Steps: 100%|██████████████████████████████████████████████████████████████████████████| 1/1 [05:20<00:00, 320.96s/it, lr=1e-5, step_loss=0.0488]

@will-cromar will-cromar merged commit 5aed632 into GoogleCloudPlatform:master Jan 9, 2024
3 checks passed
zpcore added a commit that referenced this pull request Jan 9, 2024
will-cromar pushed a commit that referenced this pull request Jan 9, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants