Comments (6)
@araquejc for historic reasons, the QM9 models were built by using atom types, so the model will first convert the atomic number to the corresponding type before passing it to the model.
Line 10 in 0047e04
from megnet.
@araquejc To be precise, our original qm9 models were initially trained on the processed qm9 dataset by Faber et. al. J. Chem. Theory Comput. 2017, 13, 11, 5255–5264, where they have atom types (integer) instead of atomic number. In later model developments, we decided to use the atomic number for graph construction, so we had to make this conversion to be compatible.
from megnet.
@chc273 Thanks for your quick reply!
Does that mean that in the code above embedding[1]
== "H", embedding[2]
== "C", embedding[4]
== "N", embedding[6]
== "O" and embedding[8]
== "F"?
from megnet.
Yes you are right @araquejc
from megnet.
Thanks again! That was my initial guess, but I wanted to cross check with you. I think it is useful to know this before trying transfer learning for MEGNet models reading pymatgen molecules directly (i.e., using atomic number).
from megnet.
BTW, you are welcome to close the issue.
from megnet.
Related Issues (20)
- Can MEGNet be made end-to-end differentiable wrt atomic positions of atoms? HOT 8
- import pybel error HOT 1
- Multi-output Regression HOT 4
- could you please provide an example to predict disordered structure? HOT 12
- Multi classification task HOT 5
- installation issue (RDKit and openbabel must be installed) HOT 3
- Might be nice to see MEGNet on matbench HOT 7
- unlockNN module for uncertainty quantification of MEGNet HOT 5
- code in pytorch? HOT 3
- Problem with load molecule data from ase HOT 4
- how does megnet paper address possible data leakage HOT 1
- Return history of the model using .train_from_graphs HOT 1
- Using megnet for systems having atoms apart from "H", "C", "N", "O", "F" HOT 4
- Breaking up a big molecule into two parts; making predictions from graphs of the two separate "molecular" parts
- Unable to understand the shape of the target/output tensor for example in "Customized Graph Network Models"
- List index out of range while training the NN
- Web App HOT 2
- Web App Access HOT 1
- Review request for disordered structure example HOT 2
- ImportError: cannot import name 'ReduceLRUponNan' from 'megnet.callbacks' in multifidelity/train.py
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 megnet.