Skip to content

Failed to load checkpoint to continue training. #6

Description

@seTalent

When I trained to step 51, an error occurred. I had saved the checkpoint at step 50 in the checkpoint-50 folder, but after loading it, the training still starts from step 0. Why?

from transformers.trainer_utils import get_last_checkpoint
...
    last_checkpoint = get_last_checkpoint(train_args.output_dir)
    if last_checkpoint is not None:

        trainer.train(resume_from_checkpoint=last_checkpoint)

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions