updating data.py for simplicity and z-mass and running 3particle update #2
Conversation
| @@ -2,3 +2,5 @@ | |||
| *.DS_Store | |||
| **/.ipynb_checkpoints | |||
| *.out | |||
| batch**.sh | |||
There was a problem hiding this comment.
What is the reason for the batch**.sh scripts to be ignored by git? Are they autogenerated by another program?
There was a problem hiding this comment.
Not autogenerated but I figured since they were native to our specific casper environment, we don't necessarily want them on the git repo. What do you think?
| model_name: "cnn" | ||
| num_particles: 1 | ||
| random_seed: 328942 | ||
| output_cols: ["x", "y", "z", "d"] | ||
| input_scaler: "MinMaxScaler" | ||
| scaler_out: "MinMaxScaler" |
There was a problem hiding this comment.
Does scaler_out handle the scaling of both the inputs and outputs or just the outputs?
There was a problem hiding this comment.
Based on further review of the code, it looks like it handles the outputs, which is fine.
There was a problem hiding this comment.
sorry, yes, just outputs. I renamed scaler_vals to scaler_in to mimic the naming convention in my latest commit.
library/models.py
Outdated
| self.model.compile(optimizer=self.optimizer, loss=self.loss) | ||
| self.model.summary() | ||
|
|
||
| def fit(self, x, y): | ||
| def fit(self, x, y, xv, yv): |
There was a problem hiding this comment.
Can you change xv and yv to keyword arguments with the default equal to None (xv=None, yv=None) in case someone wants to fit data without using a validation set?
scripts/ggantos/train_conv2d.py
Outdated
| print("Saving results and config file..") | ||
| mod.model.save(join(path_save, config["model_name"]+".h5")) | ||
| name_save = open(join(path_save, "train_outputs_pred.pkl"), 'wb') | ||
| pickle.dump(train_outputs_pred, name_save) |
There was a problem hiding this comment.
For the prediction and error metric files, I would prefer they be saved in csv or parquet format instead of pickle. Pickle is not completely portable between different Python environments since it relies on environment code and libraries being the same to read the file.
There was a problem hiding this comment.
I didn't realize that. I'll revert back.
| print("Saving results and config file..") | ||
| mod.model.save(join(path_save, config["model_name"]+".h5")) | ||
| name_save = open(join(path_save, "train_outputs_pred.pkl"), 'wb') | ||
| pickle.dump(train_outputs_pred, name_save) |
There was a problem hiding this comment.
Same issue with pickle files here.
There was a problem hiding this comment.
So the reason I pickled was to save a dict. I've fixed in subsequent commit. Thanks!
|
I also added a few other things since I was working between the PR and this time. Sorry I hope that doesn't make things too complicated. |
No description provided.