I installed the accelerated version to 0.28.0 (by ‘pip install accelerate -U’) but the same error repeats. The accelerate version seems to change midway through, is there a workaround? I’m using a Jupyter-Lab running in a virtualized environment with Python version 3.11. This is my Trainer code. !…

Accelerate version errors in Trainer

marcsun13 April 15, 2024, 1:06pm 4

Hi @BBIBIBBIBI , could you share a minimum reproducer with colab by calling is_accelerate_available since this seems to fail here.

Topic		Replies	Views
Error to import transformers[torch] or accelerate -U 🤗Transformers	0	416	April 25, 2024
Accelerate: command not found 🤗Accelerate	6	21536	July 15, 2024
Importing TrainingArguments gives error 🤗Transformers	2	2537	August 7, 2023
ImportError: Using the Trainer with PyTorch: Seq2SeqTrainingArguments 🤗Transformers	4	5338	June 17, 2023
Can't use Trainer( ) in Colab Beginners	1	1093	August 16, 2023