Devel update #18
Merged
Devel update #18
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
* clean library and header dependencies Fix #926. * fix typo in rocm * set INSTALL_RPATH for libraries
* add model compression support for models without training script * fix line changes * update doc for model compression * make the error message more reasonable * update doc for model compression
* convert `decay_rate` to `stop_lr` from old inputs close #919. * fix tests
* add LAMMPS plugin - Breaking changes: update fix_dplr to support lammps/lammps#2560. Old version may be not supported. No idea how to support both versions. - Feature: support LAMMPS's new plugin package. See document for details. - update document and examples for LAMMPS plugin. The old installation method will be still supported, although it's removed from documents. - LAMMPS hasn't released a stable version, but I think it will be released in one or two months. * fix include path * add compatibility * add `make install` to lammps document * update document for built-in mode * revert troublesome * update docs * Update source/install/build_cc.sh
Although no one really enables and uses it, I still make it compatible with v2.0. Also fix compilation errors. Seems there is hard-code `DOUBLE` in the code, so I comment the float precision.
(cherry picked from commit ceb07e5) Co-authored-by: tuoping <[email protected]>
Clang doesn't support this flag, cause an error on osx: > ld: unknown option: --no-as-needed This commit should be cherry-picked to `r1.2` branch. (cherry picked from commit ee8c419)
In conda-forge's recent released tensorflow, protobuf is external from tensorflow and not inside the tensorflow directory. (cherry picked from commit f66c78e)
* rename for cherry-pick * fix an error in stress by ase interface (cherry picked from commit a24971f) * move back * fix lint error * fix lint warnings Co-authored-by: hsulab <[email protected]>
* add more badges * fix the document badge link * remove conda badge as it doesn't work
* add code comment for InputNlist struct * use doxygen style for comments * fix line changes
*Total -- 2,328.36kb -> 1,804.25kb (22.51%) /doc/images/tensorboard-distribution.png -- 265.82kb -> 175.13kb (34.12%) /doc/images/tensorboard-graph.png -- 366.55kb -> 257.79kb (29.67%) /doc/images/l2_loss.png -- 136.16kb -> 97.31kb (28.53%) /doc/images/tensorboard-scalar.png -- 345.15kb -> 250.00kb (27.57%) /doc/images/l2_energy_loss.png -- 223.01kb -> 167.06kb (25.09%) /doc/images/l2_force_loss.png -- 176.92kb -> 133.02kb (24.81%) /doc/images/tensorboard-histograms.png -- 814.76kb -> 723.96kb (11.15%) Signed-off-by: ImgBotApp <[email protected]> Co-authored-by: ImgBotApp <[email protected]>
Co-authored-by: Han Wang <[email protected]>
* refactor docs * Update type-embedding.md * fix typos * Update lammps.md * Update model-deviation.md * fix typos * fix links in readme; rewrite training * refactor model part * fix several typos; move sections * Update doc/third-party/lammps-command.md Co-authored-by: tuoping <[email protected]> * create markdown files * revert api_cc * update developer toxtree * remove unexpected api_cc.rst Co-authored-by: Han Wang <[email protected]> Co-authored-by: tuoping <[email protected]>
* generate Python API document automatically * add `ase` to generate `deepmd.calculator` * fix `ase` intersphinx_mapping
* unpin tensorflow version As tensorflow API v1 has been stable, it's safe not to pin the tensorflow version. * set `tf` to 2 * use 2.* instead * correct `setup.py`
Also regenerates `train-input-auto.rst`.
* fix incorrect type hints and docstrings * fix a typo
* add model compression training support * fix UT error * address comments * address comments * rm fitting_net_variables from class DPTabulate * clean class DPTabulate * fix typo * add doc for init-frz-model add doc for init-frz-model at training-advanced.md * fix rocm error
* add examples to `DeepPot`'s docstring * add an example to `calc_model_devi`'s docstring
This can fixes some wired styles.
GitHub just supported this feature: https://github.blog/2021-08-19-enhanced-support-citations-github/
* Instantiate RunOptions first when training. * Add an end-to-end test for parallel training. * Allow CPU parallel training.
* add docstring to `DescrptSeA` Contains other cleanups. * fix the equation of `s(r)`
* add `InputNlist` into API doc This struct is necessary for C++ APIs. * Apply suggestions from code review Co-authored-by: Denghui Lu <[email protected]> * update comment for `firstneigh` Co-authored-by: Denghui Lu <[email protected]>
* add docstring for `EnerFitting` * Apply suggestions from code review Co-authored-by: Han Wang <[email protected]> Co-authored-by: Han Wang <[email protected]>
* Use a local Numpy random generator. * Fix import error. * Replace random generator in test_gen_stat_data.py.
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 #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.
* speedup cuda kernel scan_nlist * fix no-pbc error
* use TF's built-in method to get numpy dtype
I got a way to get the numpy type from a int. Take an example
```py
>>> tf.dtypes.as_dtype(19).as_numpy_dtype
<class 'numpy.float16'>
```
`PRECISION_MAPPING` is not used any more, as it's actually not a public API.
By the way, it also supports `str`
```py
>>> tf.dtypes.as_dtype("float16")
tf.float16
```
* sadly only `tf.as_dtype` is supported in TF 1.8
The flag won't work without language specified. See https://cmake.org/cmake/help/v3.4/variable/CMAKE_COMPILER_IS_GNULANG.html
…in SE_A, now tanh & gelu is both available. (#1020) * commit-message: decouple activation function's type from model compression's process in SE_A, now tanh & gelu is both available. * commit-message: modified code and passed unittest * commit-message: Format Document * commit-message :Format revert * commit-message: format change * commit-message: Format change Co-authored-by: HLA <[email protected]>
* fix model compression error * add doc for model compression limitation
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.
No description provided.