google-research / google-research/vmoe

Loading Saved Checkpoint After Finetuning

Open
#148 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Jupyter Notebook
Stars
728
Forks
58
Avg merge
4h 9m
Merged PRs (30d)
3

Description

Hi,
After finetuning one of your pre-trained models (vmoe_b16_imagenet21k_randaug_strong) using the provided config file, it saves checkpoint in a folder named "1000" within which I have two subfolders named "dataset_iterator" and "state". Within the "state" directory there are files with names "checkpoint", "manifest.ocdbt", "_sharding" and a folder named "d". Within the "d" directory there are many randomly named files. How can I convert this checkpoint to generate the ".data-00000-of-00001" and the ".index" file in this scenario? Can you please give me some directions on that?

Many thanks.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.