Save machine learning parameter

How can i use datalore to train machine learning model.
save different parameter used and accuracy received on each epoc save it for future analysis.

Is there any experiment management in datalore ?

Hello Vivek! There are no in-built features for experiment tracking in Datalore at the moment. However, you can try using such solutions depending on the framework you’re using:

  • In case you’re using TensorFlow, you could use tf.keras.callbacks.ModelCheckpoint (documentation)
  • If you’re using PyTorch, you could use pytorch_lightning.callbacks.ModelCheckpoint (documentation)
  • If you’re using some other framework or would like to have a framework-agnostic solution for experiment tracking, consider using Weights&Biases

The solutions above are what I’m using myself :slight_smile:
Here’s the example notebook with automatic checkpointing using TensorFlow, check it out: Automatic checkpointing with TensorFlow

3 Likes