Comments (3)
I believe this is because these structures do not have threebody interactions in 4 Å. By design, M3GNet should still work for these structures, but current implementation is buggy in using the @tf.functions for functions of BasePotential and Potential. I have a temporary solution of commenting out the @tf.functions, which is discussed in #64, while this might not be the best solution. It would be nice if you can check it out and share your feedback.
from m3gnet.
No the current implementation is not "buggy".
@liuxiaotong15 after loading the model run on a normal structure first to compile and then subsequent runs will be ok. If the problem persists, that's a problem of tensorflow version. See PR 52 for this and also a test case "test_single_atoms" in "m3gnet/models/tests/test_model.py"
@shyuep for awareness
from m3gnet.
I think we shouldn't expect that a "normal" structure has to be loaded before making a prediction on a structure without 3-body terms. In any case, I am fine with leaving this problem as is since we will move to the matgl version soon. @JiQi535 make sure the matgl version does not have such problems.
from m3gnet.
Related Issues (20)
- Transfer learning with m3gnet
- m3gnet for 1 or 2 atoms ValueError with tensorflow==2.10.0 HOT 4
- Only isolated atom calculation is difficult for m3gnet? HOT 1
- Get energy above hull through the m3gnet model HOT 1
- segmentation issue with energy calculation for some structures HOT 3
- Feature request: relaxation under pressure
- Python 3.11 HOT 6
- Cannot read 'structure' of the figshare training data HOT 2
- Error in loss function during training HOT 1
- Inconsistent results when predicting on batches with GPU HOT 1
- a suggestion to change predict_structure function name of m3gNet to predict_energy HOT 1
- fix atoms during MD/relax HOT 2
- Derivatives of forces are NaN when there's a right angle in a structure
- is this model can use multi cpu for relaxing?
- Reproduce IAP training results
- Potential fails to predict structure without threebody interaction in threebody cutoff of 4 Å HOT 5
- Model restored from checkpoint predicts wrong energies
- Recovering MatBench results from weights for submission
- Validation Structures of None results in termination for PotentialTrainer
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from m3gnet.