If you are a data scientist or a ml engineer there are probably very high chances that you have waited for long hours sat in front of your laptops to watch your model train This experience can be really tiring and cumbersome.
To solve this problem there is a way through which we can monitor our models performance through a mobile phone. we just need to install an app called as tensordash and make some necessary changes to our code and provide the required credentials. Dont believe me?? I will give you a quick demo….
Let me give you some advantages of monitoring your model remotely :
- Watch our model train in real-time.
- Supports all major deep learning frameworks like tensorflow
- Remotely get details on the training and validation metrics.
- Get notified when your model has completed training or when it has crashed.
- Get detailed graphs on your model’s metrics.
Step 1 Head over to this link
TensorDash is an application that lets you remotely monitor your deep learning model's metrics and notifies you when…
Step 2 Clone this repository and download the tensordash app from the playstore
Step 3 Make sure you have tensorflow installed in your system. I had done this using version 2.2
Step 4 Note down the email id and password which you had used for signing into the app from the playstore. Both of them should be same which you will enter in your jupyter notebook
NOTE Had to make some changes in the script while installing tensordash
If you have cloned the repository inside tensordash folder there would be 4 files namely fastdash.py, firebasedata.py ,tensordash.py and torchdash.py rename tensordash.py to tensordash_test.py and firebasedata.py to firebasedata_v1.py. all these files should be present in the same directory to avoid dependency issues
Enter the model name and the email id you used while creating account on the app on playstore,you will be prompted to enter the password .Make sure the email id and password are same with the app.
In the tensordash_test.py make sure to include this line
The output as seen throgh the app look phenominal and it sends you a notification as well as soon as the model finishes training
If you want the complete code head over to my github repository.Hope you have learnt something new while following this blog. Happy learning :)