merge master #6
Merged
merge master #6
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
fix bugs in examples
Merge the latest devel branch from DP in Apr 20, 2021
Add what's new in v2.0 to readme.
fix bug of using old name se_a, changed to se_e2_a
devel update
use bit operations to construct nbor sort keys
Devel update
Close loss normalization
Devel update
Fix bug in coord.cu
update devel
Add lib coord unittest
Add type embedding
Currently, TF only initializes GPUs once, and never releases memory. As a result, we must set `allow_growth` in the first Session (or every session?), otherwise it doesn't work. Please confirm the `allow_growth` config in `trainer.py` is expected, otherwise we should remove it instead.
…' must occur in namespace 'deepmd' (#1071) Co-authored-by: Han Wang <wang_han@iapcm.ac.cn>
* change the default checkpoint path for model compression * update UT for model compression * clean up files
* Remove dupliated setting of `allow_growth` in trainer. * Make parallel training UT independent of its working folder. * Skip parallel-training tests when there is only 1 GPU card. * Enable parallel training UT in GitHub CI.
#1074) * Fix shape mismatch when type_embedding is enabled and type_one_side is disabled. * Add UT to cover the case where batch_size is larger than 1. * Fix random issue in unit tests.
* create cross-references in docstring * add typing alias
* add ABC for descriptors I'm going to add abstract base classes for different object, where a list of methods and attributes is defined to normalize classes and their external call by other classes. It's also useful to develop and extend new classes. The first one I did is the descriptor. * TYPE_CHECKING doesn't work in python 3.6 * fix warnings
* fix a potential slice bug in se_t * fix UT error * address comments
* make compress work for hybrid descriptor composed of se_e2_a * fix `get_embedding_net_nodes_from_graph_def` * fix lint warning
…l" (#1093) * done interface for deepmd with the centroids atoms * Update lammps-command.md Added documentation to the computation of the heat flux
* Fix gradient not averaged when parallel training. * Correct throughput metrics and explain CPU runtime in the parallel-training tutorial.
* enable init-frz-model support for the original model within the dp train interface * add init_variables method for ABC * add doc for embedding_net method
Note: this is a simple fix to resolve #1088, but I think we should design a clear architecture to call neighbor stat. This should reduce the half of the time, but it may be still too long. We can consider some better algorithm to calculate neighbour stat (like KDtree?) for further optimization.
* enable init-frz-model support for the original model within the dp train interface * add init_variables method for ABC * add doc for embedding_net method * update doc for init-frz-model
* Fix missing `std::numeric_limits` - Include `<limits>` header, fix missing `std::numeric_limits` when compiling with GCC 11.0.
* support init_frz_model for hybrid descriptor Refactors some methods to implement it. Also fixes some typos. * rename `graph_def` to `model_file` Co-authored-by: Denghui Lu <ludenghui.cs@gmail.com> Co-authored-by: Denghui Lu <ludenghui.cs@gmail.com>
Co-authored-by: Han Wang <wang_han@iapcm.ac.cn>
…es (#1124) Co-authored-by: Han Wang <wang_han@iapcm.ac.cn>
As detected by LGTM, here is an error: Call to method DeepTensor.__init__ with too many arguments; should be no more than 4. I agree with the automatic code analysis tool.
Merge devel into master
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.
No description provided.