Merge branch 'update-readme-2022-06-05' into 'main'

corrected some typos + reformulation of README

See merge request !1

README.md

@@ -6,10 +6,10 @@ Interpretable deep learning was used to identify structure-property relationship
 </p>
 
 ## preparations
-1. Download repo.  
-2. Download dataset (`csv` + `xyz`s) from [COMPAS](https://gitlab.com/porannegroup/compas)
-3. Update `csv` + `xyz`s paths in `utiles/args.py`
-4. Install conda environment according to the instruction below
+1. Download repository  
+2. Download dataset (`csv` and `xyzs`) from [COMPAS](https://gitlab.com/porannegroup/compas)
+3. Update `csv` and `xyzs` paths in `utiles/args.py`
+4. Install conda environment according to the instructions below
 
 ## Dependencies
 ```
@@ -23,35 +23,32 @@ pip install requests
 
 ## Usage
 ### Training
-The training script will train the model using the train and validation datasets. 
-When training finish will run evaluation on the test set and will print the results. 
-The saved model and the some plots will be save in `summary/{exp_name}`. 
+During training, the script `train.py` trains the model using the train and validation datasets. After the training is done, the script runs an evaluation on the test set and prints the results. 
+The saved model and the plots of the prediction vs the target values are saved in `summary/{exp_name}`. The tensorboard log files are also saved in this directory.
 ```
 python train.py --name 'exp_name' --target_features 'GAP_eV, Erel_eV' 
 ```
 
-target_features should be separate with `,`.  
-Full list of possible arguments can be seen in `utiles/args.py`.  
+target_features should be separate with '`,`'.  
+A full list of possible arguments can be found in `utiles/args.py`.  
 
 ### Evaluation
-Run only the evaluation on trained model.
+To only run the evaluation on a previously trained model, run the following command.
 ```
 python eval.py --name 'exp_name'
 ```
 
 ### Interpretability
-Running the interpretability algorithm. Will save all the molecules in the dataset 
-with their GradRAM weights in the same directory as the logs and models.
+To run the interpretability algorithm, use `interpretability_from_names.py`. This script saves all the molecules in the dataset with their GradRAM weights in the same directory as the models and logs.
 ```
 python interpretability_save_all.py --name 'exp_name'
 ```
-To run only on subset of molecules run `interpretability_from_names.py` and change the molecules name
-list in line 27.
+To run only on a subset of molecules, run `interpretability_from_names.py` and change the names of the molecules in the list in line 27.
 
 ## Repo structure
 - `data` -  Dataset and data related code.
-- `se3_transformers` - SE3 model use as a predictor
-- `utils` - helper function. 
-- `summary` - Logs, trained models and interpretability figures for each experiment. 
+- `se3_transformers` - SE3 model used as a predictor.
+- `utils` - Helper functions. 
+- `summary` - Logs, trained models, plots, and interpretability figures for each experiment.