GREAMolecularPredictor Model

Model Description

Model Type: GREAMolecularPredictor
Framework: torch_molecule
Last Updated: 2024-11-05

Task Summary

Task	Version	Last Updated	Parameters	Metrics
CH4	0.0.1	2024-11-05	2,887,305	mae_logscale: 0.2346, rmse_logscale: 0.3129, r2_logscale: 0.9538, mae_original: 345.2733, rmse_original: 2421.8583, r2_original: 0.4823
CO2	0.0.1	2024-11-05	7,202,505	mae_logscale: 0.2183, rmse_logscale: 0.2986, r2_logscale: 0.9391, mae_original: 632.9703, rmse_original: 2853.8322, r2_original: 0.6528
H2	0.0.1	2024-11-05	5,467,643	mae_logscale: 0.2322, rmse_logscale: 0.3212, r2_logscale: 0.8981, mae_original: 451.6692, rmse_original: 2023.6836, r2_original: 0.7175
He	0.0.1	2024-11-05	3,664,329	mae_logscale: 0.2213, rmse_logscale: 0.3189, r2_logscale: 0.8598, mae_original: 235.3638, rmse_original: 990.1159, r2_original: 0.7079
N2	0.0.1	2024-11-05	6,066,825	mae_logscale: 0.2182, rmse_logscale: 0.2939, r2_logscale: 0.9488, mae_original: 120.5673, rmse_original: 765.6535, r2_original: 0.6790
O2	0.0.1	2024-11-05	8,272,009	mae_logscale: 0.2044, rmse_logscale: 0.2793, r2_logscale: 0.9470, mae_original: 174.0675, rmse_original: 985.5281, r2_original: 0.6549

Usage

from torch_molecule import GREAMolecularPredictor

# Load model for specific task
model = GREAMolecularPredictor()
model.load_model(
    "local_model_dir/GREA_O2.pt",
    repo="liuganghuggingface/torch-molecule-ckpt-GREA-gas-separation"
)

# Make predictions
predictions = model.predict(smiles_list)

Tasks Details

CH4 Task

Current Version: 0.0.1
Last Updated: 2024-11-05
Parameters: 2,887,305
Configuration:

{
  "gamma": 0.6499660870109112,
  "num_tasks": 1,
  "task_type": "regression",
  "num_layer": 5,
  "emb_dim": 292,
  "gnn_type": "gin",
  "drop_ratio": 0.051185452872413995,
  "norm_layer": "batch_norm",
  "graph_pooling": "max",
  "batch_size": 512,
  "epochs": 500,
  "learning_rate": 0.00026256493386201594,
  "grad_clip_value": null,
  "weight_decay": 0.0,
  "patience": 50,
  "evaluate_name": "r2",
  "evaluate_higher_better": true,
  "use_lr_scheduler": true,
  "scheduler_factor": 0.5,
  "scheduler_patience": 5,
  "device": {
    "_type": "unknown",
    "repr": "cuda:0"
  },
  "fitting_epoch": 482
}

CO2 Task

Current Version: 0.0.1
Last Updated: 2024-11-05
Parameters: 7,202,505
Configuration:

{
  "gamma": 0.7410363852617605,
  "num_tasks": 1,
  "task_type": "regression",
  "num_layer": 5,
  "emb_dim": 466,
  "gnn_type": "gin",
  "drop_ratio": 0.06177456455268606,
  "norm_layer": "batch_norm",
  "graph_pooling": "max",
  "batch_size": 512,
  "epochs": 500,
  "learning_rate": 0.00013874595577115532,
  "grad_clip_value": null,
  "weight_decay": 0.0,
  "patience": 50,
  "evaluate_name": "r2",
  "evaluate_higher_better": true,
  "use_lr_scheduler": true,
  "scheduler_factor": 0.5,
  "scheduler_patience": 5,
  "device": {
    "_type": "unknown",
    "repr": "cuda:0"
  },
  "fitting_epoch": 490
}

H2 Task

Current Version: 0.0.1
Last Updated: 2024-11-05
Parameters: 5,467,643
Configuration:

{
  "gamma": 0.6971165575657507,
  "num_tasks": 1,
  "task_type": "regression",
  "num_layer": 3,
  "emb_dim": 467,
  "gnn_type": "gin",
  "drop_ratio": 0.05045878948729124,
  "norm_layer": "batch_norm",
  "graph_pooling": "max",
  "batch_size": 512,
  "epochs": 500,
  "learning_rate": 0.00016488103933540608,
  "grad_clip_value": null,
  "weight_decay": 0.0,
  "patience": 50,
  "evaluate_name": "r2",
  "evaluate_higher_better": true,
  "use_lr_scheduler": true,
  "scheduler_factor": 0.5,
  "scheduler_patience": 5,
  "device": {
    "_type": "unknown",
    "repr": "cuda:0"
  },
  "fitting_epoch": 496
}

He Task

Current Version: 0.0.1
Last Updated: 2024-11-05
Parameters: 3,664,329
Configuration:

{
  "gamma": 0.671456137815321,
  "num_tasks": 1,
  "task_type": "regression",
  "num_layer": 5,
  "emb_dim": 330,
  "gnn_type": "gin",
  "drop_ratio": 0.07591468822202135,
  "norm_layer": "batch_norm",
  "graph_pooling": "max",
  "batch_size": 512,
  "epochs": 500,
  "learning_rate": 0.0005543898679116785,
  "grad_clip_value": null,
  "weight_decay": 0.0,
  "patience": 50,
  "evaluate_name": "r2",
  "evaluate_higher_better": true,
  "use_lr_scheduler": true,
  "scheduler_factor": 0.5,
  "scheduler_patience": 5,
  "device": {
    "_type": "unknown",
    "repr": "cuda:0"
  },
  "fitting_epoch": 499
}

N2 Task

Current Version: 0.0.1
Last Updated: 2024-11-05
Parameters: 6,066,825
Configuration:

{
  "gamma": 0.46518791970221784,
  "num_tasks": 1,
  "task_type": "regression",
  "num_layer": 5,
  "emb_dim": 427,
  "gnn_type": "gin",
  "drop_ratio": 0.05797774282594118,
  "norm_layer": "batch_norm",
  "graph_pooling": "max",
  "batch_size": 512,
  "epochs": 500,
  "learning_rate": 0.00010332984008227585,
  "grad_clip_value": null,
  "weight_decay": 0.0,
  "patience": 50,
  "evaluate_name": "r2",
  "evaluate_higher_better": true,
  "use_lr_scheduler": true,
  "scheduler_factor": 0.5,
  "scheduler_patience": 5,
  "device": {
    "_type": "unknown",
    "repr": "cuda:0"
  },
  "fitting_epoch": 489
}

O2 Task

Current Version: 0.0.1
Last Updated: 2024-11-05
Parameters: 8,272,009
Configuration:

{
  "gamma": 0.5720495153161845,
  "num_tasks": 1,
  "task_type": "regression",
  "num_layer": 5,
  "emb_dim": 500,
  "gnn_type": "gin",
  "drop_ratio": 0.06047414588643303,
  "norm_layer": "batch_norm",
  "graph_pooling": "max",
  "batch_size": 512,
  "epochs": 500,
  "learning_rate": 0.00028758290377145013,
  "grad_clip_value": null,
  "weight_decay": 0.0,
  "patience": 50,
  "evaluate_name": "r2",
  "evaluate_higher_better": true,
  "use_lr_scheduler": true,
  "scheduler_factor": 0.5,
  "scheduler_patience": 5,
  "device": {
    "_type": "unknown",
    "repr": "cuda:0"
  },
  "fitting_epoch": 496
}