Training fails because of accelerate configure settings.

[AdamLouly]

System Info

nightly optimum
nightly transformers
accelerate

Who can help?

No response

Information

• The official example scripts
• My own modified scripts

Tasks

• An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
• My own task or dataset (give details below)

Reproduction (minimal, reproducible, runnable)

args.accelerator_config does not get initialized.

Make sure accelerate is installed (even without using it)

python -m torch.distributed.launch --nproc_per_node=8 --use-env run_glue.py --model_name_or_path microsoft/deberta-large --task_name MRPC --max_seq_length 128 --learning_rate 3e-6 --do_train --output_dir /dev/shm --overwrite_output_dir --max_steps 200 --logging_steps 20 --per_device_train_batch_size 32 --fp16

Expected behavior

Error:

[rank4]: Traceback (most recent call last): [rank4]: File "/workspace/optimum/examples/onnxruntime/training/text-classification/run_glue.py", line 649, in <module> [rank4]: main() [rank4]: File "/workspace/optimum/examples/onnxruntime/training/text-classification/run_glue.py", line 540, in main [rank4]: trainer = ORTTrainer( [rank4]: File "/opt/conda/envs/ptca/lib/python3.8/site-packages/optimum-1.18.0.dev0-py3.8.egg/optimum/onnxruntime/trainer.py", line 227, in __init__ [rank4]: super().__init__( [rank4]: File "/opt/conda/envs/ptca/lib/python3.8/site-packages/transformers-4.39.0.dev0-py3.8.egg/transformers/trainer.py", line 369, in __init__ [rank4]: self.create_accelerator_and_postprocess() [rank4]: File "/opt/conda/envs/ptca/lib/python3.8/site-packages/transformers-4.39.0.dev0-py3.8.egg/transformers/trainer.py", line 4094, in create_accelerator_and_postprocess [rank4]: **self.args.accelerator_config.to_dict(), [rank4]: AttributeError: 'NoneType' object has no attribute 'to_dict'

[AdamLouly]

This should fix it:
#1750