question-mark
Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

suggest methods could include a batch size.

See original GitHub issue

suggest_ methods could include an optional batch_size (default 1) value.

Motivation

We have an optimization problem in which it is faster to operate in batch. We would like to have 256 trials at once.

Description

There should be a way to generate multiple trials at once. Alternately, to create multiple trials before receiving the objective for the previous trial.

Alternatives (optional)

  • Slow down our optimization by using batch_size of 1.
  • It is not clear to me that I can create multiple trial in Optuna without returning the objective score from a previous trial.

Additional context (optional)

Issue Analytics

  • State:closed
  • Created 2 years ago
  • Reactions:1
  • Comments:8 (5 by maintainers)

github_iconTop GitHub Comments

1reaction
turiancommented, Apr 28, 2021

ask-and-tell is interesting, I was completely unaware of it. I would prefer a more straightforward interface, but this also looks pretty good.

1reaction
not522commented, Apr 28, 2021

The ask-and-tell interface enables batch optimization. https://optuna.readthedocs.io/en/stable/tutorial/20_recipes/009_ask_and_tell.html#batch-optimization

But, I think that a more straightforward interface like your suggestion will be helpful to write simple code.

Read more comments on GitHub >

github_iconTop Results From Across the Web

neural networks - How do I choose the optimal batch size?
Here are a few guidelines, inspired by the deep learning specialization course, to choose the size of the mini-batch:.
Read more >
How to use Different Batch Sizes when Training and ...
In this tutorial, you will discover how you can address this problem and even use different batch sizes during training and predicting. After ......
Read more >
Effect of batch size on training dynamics | by Kevin Shen
In this experiment, I investigate the effect of batch size on training dynamics. The metric we will focus on is the generalization gap...
Read more >
Understanding Optimal Batch Size / Troy Magennis - Observable
The batch setup cost is computed simply by amortizing that cost over the batch size. Batch size of one means total cost for...
Read more >
A batch too large: Finding the batch size that fits on GPUs
In addition, the batch size can greatly affect the performance of the model. For instance, a large batch size can lead to poor...
Read more >

github_iconTop Related Medium Post

No results found

github_iconTop Related StackOverflow Question

No results found

github_iconTroubleshoot Live Code

Lightrun enables developers to add logs, metrics and snapshots to live code - no restarts or redeploys required.
Start Free

github_iconTop Related Reddit Thread

No results found

github_iconTop Related Hackernoon Post

No results found

github_iconTop Related Tweet

No results found

github_iconTop Related Dev.to Post

No results found

github_iconTop Related Hashnode Post

No results found