torchmd/torchmd-net issues and pull requests

#101 - Dataloader for SPICE

Pull Request - State: closed - Opened by raimis almost 2 years ago

#100 - Fix the force unit in QM9q

Pull Request - State: closed - Opened by raimis about 2 years ago

#99 - Loader for QM9q dataset

Pull Request - State: closed - Opened by raimis about 2 years ago

#98 - reduce the dependencies of torchmd-net

Pull Request - State: closed - Opened by stefdoerr about 2 years ago

#97 - Attempt at upgrading pytorch-lightning

Pull Request - State: closed - Opened by PhilippThoelke about 2 years ago - 3 comments

#96 - Ways to reduce memory use

Issue - State: open - Opened by peastman about 2 years ago - 15 comments

#95 - Question about adding missing atoms in protein

Issue - State: closed - Opened by psp3dcg about 2 years ago - 10 comments

#94 - Handle when load_model is specified twice

Pull Request - State: closed - Opened by PhilippThoelke about 2 years ago

#93 - `test_hdf5_multiprocessing` fails randomly

Issue - State: closed - Opened by raimis about 2 years ago - 2 comments

#92 - Periodic boundary conditions

Issue - State: closed - Opened by peastman about 2 years ago - 19 comments

#91 - Fix the dataset import

Pull Request - State: closed - Opened by raimis about 2 years ago - 1 comment

#90 - Update the dependencies

Pull Request - State: closed - Opened by raimis about 2 years ago - 2 comments

#89 - Added option for which metric to base learning rate decay on

Pull Request - State: closed - Opened by peastman about 2 years ago - 3 comments

#88 - see if we can use conda-forge NNPOps

Pull Request - State: closed - Opened by mikemhenry about 2 years ago - 2 comments

#87 - Fix CI

Pull Request - State: closed - Opened by raimis about 2 years ago - 1 comment

#86 - How to continue training with new settings?

Issue - State: closed - Opened by peastman about 2 years ago - 5 comments

#85 - NNPOps tests broken

Issue - State: closed - Opened by PhilippThoelke about 2 years ago - 5 comments

#84 - Avoid multiple processes reading from the same HDF5 file

Pull Request - State: closed - Opened by PhilippThoelke about 2 years ago

#83 - COMP6 dataset

Pull Request - State: closed - Opened by raimis about 2 years ago

#82 - Unable to fit model

Issue - State: closed - Opened by peastman about 2 years ago - 18 comments

#81 - Throw error if max_num_neighbors is too small

Pull Request - State: closed - Opened by PhilippThoelke about 2 years ago

#80 - Make sure self loops are kept with lower cutoff

Pull Request - State: closed - Opened by PhilippThoelke about 2 years ago

#79 - Skip gradients of atoms outside cutoff

Pull Request - State: closed - Opened by PhilippThoelke about 2 years ago - 3 comments

#78 - Dynamic max num neighbors

Pull Request - State: closed - Opened by PhilippThoelke about 2 years ago - 2 comments

#77 - NaN when fitting with derivative

Issue - State: closed - Opened by peastman about 2 years ago - 15 comments

#76 - Unify the ANI datasets

Pull Request - State: closed - Opened by raimis about 2 years ago - 2 comments

#75 - Reproducing SoTA results for QM9

Issue - State: closed - Opened by XJTUNR about 2 years ago - 5 comments

#74 - POC: the graph network optimization with the custom kernels

Pull Request - State: closed - Opened by raimis about 2 years ago - 1 comment

#73 - Memory-efficient ANI-1 dataset with MemMap

Pull Request - State: closed - Opened by raimis about 2 years ago - 5 comments

#72 - Memory-efficient ANI-1 dataset

Pull Request - State: closed - Opened by raimis about 2 years ago - 1 comment

#71 - Broken links in README.md

Issue - State: closed - Opened by yqshao about 2 years ago - 1 comment

#70 - adds ANI1x as dataset option

Pull Request - State: closed - Opened by sebastianmdick about 2 years ago

#69 - POC: message passing

Pull Request - State: closed - Opened by raimis about 2 years ago - 1 comment

#68 - Speed-up neighbors calculation

Pull Request - State: closed - Opened by claudi about 2 years ago - 6 comments

#67 - fix bugs for the newest version of pytorch_lightning

Pull Request - State: closed - Opened by lsnty5190 about 2 years ago - 4 comments

#66 - how to get the test result?

Issue - State: closed - Opened by yzf-code about 2 years ago - 1 comment

#65 - The loss is NaN when ET models is set "distance_influence" to none

Issue - State: closed - Opened by XJTUNR over 2 years ago - 1 comment

#64 - Reproducing results for QM9

Issue - State: closed - Opened by shehzaidi over 2 years ago - 4 comments
Labels: help wanted

#63 - Add support for the Ace dataset and ability to pass the charge and spin state of molecules

Pull Request - State: closed - Opened by raimis over 2 years ago - 7 comments

#62 - Convert the unknown `load_model` parameters to warnings

Pull Request - State: closed - Opened by raimis over 2 years ago - 1 comment

#61 - POC: neighbor search

Pull Request - State: closed - Opened by raimis over 2 years ago - 3 comments

#60 - POC: optimize the graph network with CUDA Graph

Pull Request - State: closed - Opened by raimis over 2 years ago - 1 comment

#59 - Add testosterone as a benchmark system

Pull Request - State: closed - Opened by raimis over 2 years ago

#58 - Fix accidental bias in equivariant Transformer

Pull Request - State: closed - Opened by PhilippThoelke over 2 years ago

#57 - Bias terms in the Equivariant Block and Attention Mechanism Implementation

Issue - State: closed - Opened by charliexchen over 2 years ago - 4 comments
Labels: bug

#56 - Add more systems for benchmarking

Pull Request - State: closed - Opened by raimis over 2 years ago - 1 comment

#55 - Update the installation instructions

Pull Request - State: closed - Opened by raimis over 2 years ago - 3 comments

#54 - Unable to reproduce the results of ET on QM9

Issue - State: closed - Opened by 20171130 over 2 years ago - 1 comment
Labels: help wanted

#53 - Make compatible with conda-forge

Pull Request - State: closed - Opened by raimis over 2 years ago - 1 comment

#52 - Enable more CI triggers

Pull Request - State: closed - Opened by raimis over 2 years ago - 1 comment

#51 - Fix the initialization of CFConv

Pull Request - State: closed - Opened by raimis over 2 years ago - 1 comment

#50 - Accelerate the limited TorchMD_GN with NNPOps

Pull Request - State: closed - Opened by raimis over 2 years ago - 1 comment

#49 - Accelerate the limited `TorchMD_GN` with NNPOps

Issue - State: closed - Opened by raimis over 2 years ago

#48 - Optimization of the graph network

Issue - State: open - Opened by raimis over 2 years ago - 14 comments

#47 - Initial benchmark systems

Pull Request - State: closed - Opened by raimis over 2 years ago

#46 - Molecules for bechmarking speed

Issue - State: closed - Opened by raimis over 2 years ago - 1 comment

#45 - Optimize equivariant transformer

Issue - State: open - Opened by peastman over 2 years ago - 23 comments

#44 - Repo to move to github.com/torchmd and become public

Issue - State: closed - Opened by giadefa over 2 years ago - 1 comment

#43 - ANI1-equivariant_transformer cannot be converted to TorchScirpt

Issue - State: closed - Opened by raimis almost 3 years ago - 3 comments

#42 - `CosineCutoff` has drastically different behavior when the lower cutoff is not zero

Issue - State: closed - Opened by yaoyic almost 3 years ago - 3 comments

#41 - Loading jittable models

Issue - State: closed - Opened by nec4 almost 3 years ago - 11 comments
Labels: help wanted

#40 - added in aggr attribute

Pull Request - State: closed - Opened by nec4 almost 3 years ago
Labels: bug

#39 - `aggr` attribute is not properly being set from `create_model`

Issue - State: closed - Opened by nec4 almost 3 years ago - 1 comment
Labels: bug

#38 - Second attempt at multihead implementation

Pull Request - State: closed - Opened by stefdoerr almost 3 years ago

#37 - Added support for multiple heads in the output network

Pull Request - State: closed - Opened by stefdoerr almost 3 years ago

#36 - Automated flake8 testing

Pull Request - State: closed - Opened by stefdoerr almost 3 years ago - 2 comments

#35 - Blackened the repo

Pull Request - State: closed - Opened by stefdoerr almost 3 years ago

#34 - Crashes while training

Issue - State: closed - Opened by peastman almost 3 years ago - 1 comment

#33 - TorchScript support

Issue - State: closed - Opened by raimis almost 3 years ago - 2 comments

#32 - added additional cfconv aggr options

Pull Request - State: closed - Opened by nec4 almost 3 years ago
Labels: enhancement

#31 - Additional aggregation options for message passing/cfconv

Issue - State: closed - Opened by nec4 almost 3 years ago
Labels: enhancement

#30 - added alpha scaling in expnorm and visualization function

Pull Request - State: closed - Opened by nec4 almost 3 years ago - 2 comments
Labels: enhancement

#29 - Ability to roll back an epoch

Issue - State: open - Opened by peastman almost 3 years ago - 3 comments

#28 - Distance scaling in ExpNorm RBFs

Issue - State: closed - Opened by nec4 almost 3 years ago - 6 comments
Labels: enhancement

#27 - Base learning rate decay on training loss

Pull Request - State: closed - Opened by peastman about 3 years ago - 28 comments

#26 - Physics based priors

Issue - State: open - Opened by peastman about 3 years ago - 72 comments

#25 - Added HDF5 dataset class

Pull Request - State: closed - Opened by peastman about 3 years ago - 1 comment

#24 - Redo train/val/test split generation

Pull Request - State: closed - Opened by PhilippThoelke about 3 years ago

#23 - Broken repr

Issue - State: closed - Opened by raimis about 3 years ago - 1 comment

#22 - Adjusting the maximum number of neighbor nodes in radius_graph

Pull Request - State: closed - Opened by nec4 about 3 years ago - 3 comments

#21 - radius graph defaults

Issue - State: closed - Opened by nec4 about 3 years ago - 11 comments

#20 - Improve model modularity

Pull Request - State: closed - Opened by PhilippThoelke about 3 years ago

#19 - Pre-trained model

Issue - State: closed - Opened by raimis about 3 years ago - 32 comments

#18 - Versioning

Issue - State: closed - Opened by raimis about 3 years ago - 2 comments

#17 - Move dataset handling into a Lightning Datamodule

Pull Request - State: closed - Opened by PhilippThoelke about 3 years ago

#16 - Move atomref calculation into the model

Pull Request - State: closed - Opened by PhilippThoelke about 3 years ago

#15 - Optional standardization step using the training data

Pull Request - State: closed - Opened by PhilippThoelke about 3 years ago

#14 - Redo test set evaluation

Pull Request - State: closed - Opened by PhilippThoelke about 3 years ago

#13 - fixed GraphNormMSE example size division error

Pull Request - State: closed - Opened by nec4 about 3 years ago

#12 - Feat/cgnet nec4

Pull Request - State: closed - Opened by nec4 about 3 years ago

#10 - AttributeError: can't set attribute

Issue - State: closed - Opened by peastman about 3 years ago - 1 comment

#9 - Rework custom dataset and allow training on energies + forces

Pull Request - State: closed - Opened by PhilippThoelke about 3 years ago

#8 - Implement atom filter in output network

Pull Request - State: closed - Opened by PhilippThoelke about 3 years ago

#7 - Reorganize datasets and implement ANI1

Pull Request - State: closed - Opened by PhilippThoelke about 3 years ago - 3 comments

#6 - Creating a custom dataset

Issue - State: closed - Opened by peastman about 3 years ago - 20 comments

#5 - How to create an environment?

Issue - State: closed - Opened by peastman about 3 years ago - 3 comments

#4 - Implement TorchMD-Transformer architecture

Pull Request - State: closed - Opened by PhilippThoelke about 3 years ago

#3 - Make testing compatible with different PyTorch Lightning versions

Pull Request - State: closed - Opened by PhilippThoelke about 3 years ago

#2 - Test loss not improving in most recent PyTorch Lightning version

Issue - State: closed - Opened by PhilippThoelke about 3 years ago

#1 - Open dataset in setup and/instead of init

Issue - State: closed - Opened by PhilippThoelke about 3 years ago - 6 comments

GitHub / torchmd/torchmd-net issues and pull requests