Kento Nozawa

Stabilise visualisation module

9

## Motivation Optuna has two visualisation backends: Plotly and Matplotlib. It would be great to minimise the difference between backends to provide consistent plots. Concretely, 1. https://github.com/optuna/optuna/issues/3226 2. Unify appearance...

When `BatchNorm` with `affine=True` is used, the previous layer's bias is cancelled (See the original BN paper for details). The current implementation's `Conv2D` layers are used with `bias=True`, so by...

Incorrect IOU definition

2

[The current IOU implementation](https://github.com/4uiiurz1/pytorch-nested-unet/blob/master/metrics.py#L15-L18) has calculated the union and intersection over samples in a mini-batch, then it has calculated the IOU score. But I think IOU should be calculated per...

How to recover the search space from `NAS-Bench-301`?

6

Hi, thank you for sharing these benchmark tools! Let me ask one question about a dataset provided by this repo. When I looked at data in `NAS-Bench-301` trained on CIFAR-10...

Decide the policy whether or not to run with pruning option in CIs

1

## Motivation Several Optuna example scripts have pruning option `--pruning`, more concretely, ```bash python pytorch_lightning_simple.py --pruning ``` as in [this example](https://github.com/optuna/optuna-examples/blob/6a6b20ad634627eebb3e7e104f73b70b45c6e624/pytorch/pytorch_lightning_simple.py). But currently CIs have not performed with such pruning...

Improve catboost example

6

Hyperparameter search space can improve by following the CatBoost's docs: https://catboost.ai/en/docs/concepts/parameter-tuning. In addition, it might be great to unify the code to the CatBoost example and the other boosting libraries....

Use truncated norm cdf instead of norm cdf

3

## Motivation When we run the following optuna optimisation with dynamic search space, ```python import optuna def objective1(trial): return trial.suggest_float("x", -1, 1) def objective2(trial): return trial.suggest_float("x", -100, -90) study =...

Drop python 3.6 from integration CIs

3

## Motivation https://github.com/optuna/optuna/issues/3021#issuecomment-1129545028: remove Python 3.6 tests from the CI of the integrations. Note that we keep using python 3.6 to test Optuna's core functionality. ## Description of the changes...

Remove GPyTorch version constraint

3

## Motivation Relax the version constraint by https://github.com/optuna/optuna/pull/3950 for GPyTorch. BoTorch 0.7.0 was released last week. It starts to support GPyTorch>=1.9.0. however, it also drops Python __3.7__ support because GPyTorch...