save checkpoint files with step and keep recent files #1031
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This commit saves checkpoint to `save_ckpt-step` (e.g. `model.ckpt-100`) instead of `save_ckpt` (e.g. `model.ckpt`), and keeps 5 recent checkpoint files (this is a default value of `tf.Saver`). Such thing is conducted by `tf.Saver`. To not break any behaviors, a symlink will then be made from `model.ckpt-100` to `model.ckpt`. (Usually such thing should be controlled by `checkpoint` file, but deepmd-kit doesn't read this file.) This can fix deepmodeling#1023, as (1) we made symlink after a checkpoint has been already saved; (2) if something is still wrong, one can use a previous checkpoint instead.
Codecov Report
@@ Coverage Diff @@
## devel #1031 +/- ##
===========================================
- Coverage 75.01% 64.28% -10.73%
===========================================
Files 86 5 -81
Lines 6924 14 -6910
===========================================
- Hits 5194 9 -5185
+ Misses 1730 5 -1725 Continue to review full report at Codecov.
|
amcadmus
approved these changes
Aug 25, 2021
gzq942560379
pushed a commit
to HPC-AI-Team/deepmd-kit
that referenced
this pull request
Sep 2, 2021
) This commit saves checkpoint to `save_ckpt-step` (e.g. `model.ckpt-100`) instead of `save_ckpt` (e.g. `model.ckpt`), and keeps 5 recent checkpoint files (this is a default value of `tf.Saver`). Such thing is conducted by `tf.Saver`. To not break any behaviors, a symlink will then be made from `model.ckpt-100` to `model.ckpt`. (Usually such thing should be controlled by `checkpoint` file, but deepmd-kit doesn't read this file.) This can fix deepmodeling#1023, as (1) we made symlink after a checkpoint has been already saved; (2) if something is still wrong, one can use a previous checkpoint instead.
njzjz
added a commit
to njzjz/deepmd-kit
that referenced
this pull request
Sep 21, 2023
Fixes deepmodeling#1029. Co-authored-by: Han Wang <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This commit saves checkpoint to
save_ckpt-step(e.g.model.ckpt-100) instead ofsave_ckpt(e.g.model.ckpt), and keeps 5 recent checkpoint files (this is a default value oftf.Saver). Such thing is conducted bytf.Saver. To not break any behaviors, a symlink will then be made frommodel.ckpt-100tomodel.ckpt. (Usually such thing should be controlled bycheckpointfile, but deepmd-kit doesn't read this file.)This can fix #1023, as (1) we made symlink after a checkpoint has been already saved; (2) if something is still wrong, one can use a previous checkpoint instead.