Hacker News Reader

New-comer to the field here with a few questions.

Do I understand correctly that a checkpoint is just a snapshot of the model at a point in time? i.e. "Here are the probabilities of each outcome given the characteristics I have observed already."

Also, what does "fully converged" signify? Are there points in the course of training the model at which it is more appropriate to "save" progress than at other times?

View on HN

> Also, what does "fully converged" signify?

In machine learning/deep learning, the decrease in training loss has major diminishing returns as training continues. Eventually, training the model hits a point where the loss barely improves each epoch/iteration. (fun visualization from one of my projects: http://minimaxir.com/img/char-embeddings/epoch-losses.png)

In some cases, the loss can stop improving entirely, or increase.

Any recs on choosing TensorFlow vs Keras-on-TensorFlow?

Start with Keras, if you run into something you want to do that is not supported by Keras drop into TensorFlow, they are not mutually exclusive and all of TensorFlow is availble.

Fantastic. Thanks to you and minimaxir for the guidance.

Keras does not add much overhead, if any, and yes, it is as easy as everyone claims.

Reply on news.ycombinator.com