Log2DB

DeepLearning Log Send to DB Module

Introdution

Log2DB is a tool that uploads, manages, and visualizes simple deep learning logs to your own DB.

How To Install

• Download manually:
  https://github.com/Oldentomato/Log2DB/releases

Download PyPI:

pip install log2db

How To Use

• Basic preparation

import log2db as lgdb

#Set the DB Server 
db = lgdb.SetMongoDB(db_url = 'db_url',
                    db_document_name='doc',
                    db_collection_name='coll',
                    port=12345)
coll = db.Connect_DB()
sendlog = lgdb.SendLog(coll)

• example

sendlog.on_train_start(
    model_name = 'pytorch_test',
    experiment_count = 3,
    datas_count = len(x_data),
    epoch = 100,
    batch_size = 8,
    learning_rate = 0.01,
    criterion = 'CrossEntropyLoss',
    optimizer = 'SGD(momentum=0.5)',
    model_shape = model,
    LR_scheduler = 'None',
    etc = 'None'
)

...

sendlog.on_epoch_end(epoch=epoch, loss=loss, val_loss=val_loss, acc=acc, val_acc=val_acc)

...

sendmoel.on_train_end(save_graph_url=True)

Dependency

• Pandas https://github.com/pandas-dev/pandas
• Matplotlib https://github.com/matplotlib/matplotlib
• pymongo https://github.com/mongodb/mongo-python-driver
• scikit-learn https://github.com/scikit-learn/scikit-learn

License

MIT