Full Code of greydanus/mnist1d for AI

Repository: greydanus/mnist1d
Branch: master
Commit: 7878d96082ab
Files: 37
Total size: 3.8 MB

Directory structure:
gitextract_qf595l3o/

├── .github/
│   └── workflows/
│       └── tests.yml
├── .gitignore
├── CITATIONS.bib
├── LICENSE
├── MANIFEST.in
├── README.md
├── mnist1d/
│   ├── .gitignore
│   ├── __init__.py
│   ├── data.py
│   ├── transform.py
│   └── utils.py
├── mnist1d_data.pkl
├── notebooks/
│   ├── README.md
│   ├── benchmark-pooling.ipynb
│   ├── building-mnist1d.ipynb
│   ├── deep-double-descent.ipynb
│   ├── lottery-tickets.ipynb
│   ├── metalearn-activation-function.ipynb
│   ├── metalearn-learn-rate.ipynb
│   ├── mnist1d-classification.ipynb
│   ├── mnist1d-pip.ipynb
│   ├── models.py
│   ├── mpl_style.txt
│   ├── quickstart.ipynb
│   ├── self-supervised-learning.ipynb
│   ├── train.py
│   └── tsne-mnist-vs-mnist1d.ipynb
├── pypl_notes.md
├── pyproject.toml
├── static/
│   ├── human_q1.pkl
│   ├── human_q2.pkl
│   ├── human_q3.pkl
│   ├── human_q4.pkl
│   └── human_q5.pkl
└── tests/
    ├── test_data.py
    ├── test_transform.py
    └── test_utils.py

================================================
FILE CONTENTS
================================================

================================================
FILE: .github/workflows/tests.yml
================================================
name: MNIST1D tests

on: [push]

jobs:
  build: #building the library locally

    runs-on: ubuntu-latest
    strategy:
      matrix:
        python-version: ["3.7", "3.8", "3.9", "3.10", "3.11", "3.12"]

    steps:
      - uses: actions/checkout@v4
      - name: Set up Python ${{ matrix.python-version }}
        uses: actions/setup-python@v4
        with:
          python-version: ${{ matrix.python-version }}
      # You can test your matrix by printing the current Python version
      - name: Display Python version
        run: python -c "import sys; print(sys.version)"
      - name: Install dev version locally
        run: python -m pip install '.[dev]'
      - name: test dev version locally
        run: python -m pytest .
      - name: install notebooks
        run: python -m pip install notebook jupyter
      - name: run minimal example notebooks
        run: jupyter nbconvert --execute --to html "notebooks/mnist1d-pip.ipynb"
      
  install: #installing the library as if one would not want to git-clone it

    runs-on: ubuntu-latest
    strategy:
      matrix:
        python-version: ["3.7", "3.8", "3.9", "3.10", "3.11", "3.12"]

    steps:
      - uses: actions/checkout@v4
      - name: Set up Python ${{ matrix.python-version }}
        uses: actions/setup-python@v4
        with:
          python-version: ${{ matrix.python-version }}
      - name: Display Python version
        run: python -c "import sys; print(sys.version)"
      - name: Install dev version locally
        run: python -m pip install git+https://github.com/greydanus/mnist1d.git@master



================================================
FILE: .gitignore
================================================
# The MNIST-1D dataset | 2020
# Sam Greydanus

.DS_Store
*.egg-info
.ipynb_checkpoints
*/.ipynb_checkpoints/*
.pytest_cache
*.tar
*.zip
notebooks/*.pkl
notebooks/mnist1d
# *.pkl # allow pickle files
__pycache__
.vscode
*~

dist/*


================================================
FILE: CITATIONS.bib
================================================
@article{greydanus2024scaling,
  title={Scaling Down Deep Learning with MNIST-1D},
  author={Greydanus, Sam and Kobak, Dmitry},
  journal={ICML},
  year={2024}
}


================================================
FILE: LICENSE
================================================

                                 Apache License
                           Version 2.0, January 2004
                        http://www.apache.org/licenses/

   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION

   1. Definitions.

      "License" shall mean the terms and conditions for use, reproduction,
      and distribution as defined by Sections 1 through 9 of this document.

      "Licensor" shall mean the copyright owner or entity authorized by
      the copyright owner that is granting the License.

      "Legal Entity" shall mean the union of the acting entity and all
      other entities that control, are controlled by, or are under common
      control with that entity. For the purposes of this definition,
      "control" means (i) the power, direct or indirect, to cause the
      direction or management of such entity, whether by contract or
      otherwise, or (ii) ownership of fifty percent (50%) or more of the
      outstanding shares, or (iii) beneficial ownership of such entity.

      "You" (or "Your") shall mean an individual or Legal Entity
      exercising permissions granted by this License.

      "Source" form shall mean the preferred form for making modifications,
      including but not limited to software source code, documentation
      source, and configuration files.

      "Object" form shall mean any form resulting from mechanical
      transformation or translation of a Source form, including but
      not limited to compiled object code, generated documentation,
      and conversions to other media types.

      "Work" shall mean the work of authorship, whether in Source or
      Object form, made available under the License, as indicated by a
      copyright notice that is included in or attached to the work
      (an example is provided in the Appendix below).

      "Derivative Works" shall mean any work, whether in Source or Object
      form, that is based on (or derived from) the Work and for which the
      editorial revisions, annotations, elaborations, or other modifications
      represent, as a whole, an original work of authorship. For the purposes
      of this License, Derivative Works shall not include works that remain
      separable from, or merely link (or bind by name) to the interfaces of,
      the Work and Derivative Works thereof.

      "Contribution" shall mean any work of authorship, including
      the original version of the Work and any modifications or additions
      to that Work or Derivative Works thereof, that is intentionally
      submitted to Licensor for inclusion in the Work by the copyright owner
      or by an individual or Legal Entity authorized to submit on behalf of
      the copyright owner. For the purposes of this definition, "submitted"
      means any form of electronic, verbal, or written communication sent
      to the Licensor or its representatives, including but not limited to
      communication on electronic mailing lists, source code control systems,
      and issue tracking systems that are managed by, or on behalf of, the
      Licensor for the purpose of discussing and improving the Work, but
      excluding communication that is conspicuously marked or otherwise
      designated in writing by the copyright owner as "Not a Contribution."

      "Contributor" shall mean Licensor and any individual or Legal Entity
      on behalf of whom a Contribution has been received by Licensor and
      subsequently incorporated within the Work.

   2. Grant of Copyright License. Subject to the terms and conditions of
      this License, each Contributor hereby grants to You a perpetual,
      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
      copyright license to reproduce, prepare Derivative Works of,
      publicly display, publicly perform, sublicense, and distribute the
      Work and such Derivative Works in Source or Object form.

   3. Grant of Patent License. Subject to the terms and conditions of
      this License, each Contributor hereby grants to You a perpetual,
      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
      (except as stated in this section) patent license to make, have made,
      use, offer to sell, sell, import, and otherwise transfer the Work,
      where such license applies only to those patent claims licensable
      by such Contributor that are necessarily infringed by their
      Contribution(s) alone or by combination of their Contribution(s)
      with the Work to which such Contribution(s) was submitted. If You
      institute patent litigation against any entity (including a
      cross-claim or counterclaim in a lawsuit) alleging that the Work
      or a Contribution incorporated within the Work constitutes direct
      or contributory patent infringement, then any patent licenses
      granted to You under this License for that Work shall terminate
      as of the date such litigation is filed.

   4. Redistribution. You may reproduce and distribute copies of the
      Work or Derivative Works thereof in any medium, with or without
      modifications, and in Source or Object form, provided that You
      meet the following conditions:

      (a) You must give any other recipients of the Work or
          Derivative Works a copy of this License; and

      (b) You must cause any modified files to carry prominent notices
          stating that You changed the files; and

      (c) You must retain, in the Source form of any Derivative Works
          that You distribute, all copyright, patent, trademark, and
          attribution notices from the Source form of the Work,
          excluding those notices that do not pertain to any part of
          the Derivative Works; and

      (d) If the Work includes a "NOTICE" text file as part of its
          distribution, then any Derivative Works that You distribute must
          include a readable copy of the attribution notices contained
          within such NOTICE file, excluding those notices that do not
          pertain to any part of the Derivative Works, in at least one
          of the following places: within a NOTICE text file distributed
          as part of the Derivative Works; within the Source form or
          documentation, if provided along with the Derivative Works; or,
          within a display generated by the Derivative Works, if and
          wherever such third-party notices normally appear. The contents
          of the NOTICE file are for informational purposes only and
          do not modify the License. You may add Your own attribution
          notices within Derivative Works that You distribute, alongside
          or as an addendum to the NOTICE text from the Work, provided
          that such additional attribution notices cannot be construed
          as modifying the License.

      You may add Your own copyright statement to Your modifications and
      may provide additional or different license terms and conditions
      for use, reproduction, or distribution of Your modifications, or
      for any such Derivative Works as a whole, provided Your use,
      reproduction, and distribution of the Work otherwise complies with
      the conditions stated in this License.

   5. Submission of Contributions. Unless You explicitly state otherwise,
      any Contribution intentionally submitted for inclusion in the Work
      by You to the Licensor shall be under the terms and conditions of
      this License, without any additional terms or conditions.
      Notwithstanding the above, nothing herein shall supersede or modify
      the terms of any separate license agreement you may have executed
      with Licensor regarding such Contributions.

   6. Trademarks. This License does not grant permission to use the trade
      names, trademarks, service marks, or product names of the Licensor,
      except as required for reasonable and customary use in describing the
      origin of the Work and reproducing the content of the NOTICE file.

   7. Disclaimer of Warranty. Unless required by applicable law or
      agreed to in writing, Licensor provides the Work (and each
      Contributor provides its Contributions) on an "AS IS" BASIS,
      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
      implied, including, without limitation, any warranties or conditions
      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
      PARTICULAR PURPOSE. You are solely responsible for determining the
      appropriateness of using or redistributing the Work and assume any
      risks associated with Your exercise of permissions under this License.

   8. Limitation of Liability. In no event and under no legal theory,
      whether in tort (including negligence), contract, or otherwise,
      unless required by applicable law (such as deliberate and grossly
      negligent acts) or agreed to in writing, shall any Contributor be
      liable to You for damages, including any direct, indirect, special,
      incidental, or consequential damages of any character arising as a
      result of this License or out of the use or inability to use the
      Work (including but not limited to damages for loss of goodwill,
      work stoppage, computer failure or malfunction, or any and all
      other commercial damages or losses), even if such Contributor
      has been advised of the possibility of such damages.

   9. Accepting Warranty or Additional Liability. While redistributing
      the Work or Derivative Works thereof, You may choose to offer,
      and charge a fee for, acceptance of support, warranty, indemnity,
      or other liability obligations and/or rights consistent with this
      License. However, in accepting such obligations, You may act only
      on Your own behalf and on Your sole responsibility, not on behalf
      of any other Contributor, and only if You agree to indemnify,
      defend, and hold each Contributor harmless for any liability
      incurred by, or claims asserted against, such Contributor by reason
      of your accepting any such warranty or additional liability.

   END OF TERMS AND CONDITIONS

   APPENDIX: How to apply the Apache License to your work.

      To apply the Apache License to your work, attach the following
      boilerplate notice, with the fields enclosed by brackets "[]"
      replaced with your own identifying information. (Don't include
      the brackets!)  The text should be enclosed in the appropriate
      comment syntax for the file format. We also recommend that a
      file or class name and description of purpose be included on the
      same "printed page" as the copyright notice for easier
      identification within third-party archives.

   Copyright [yyyy] [name of copyright owner]

   Licensed under the Apache License, Version 2.0 (the "License");
   you may not use this file except in compliance with the License.
   You may obtain a copy of the License at

       http://www.apache.org/licenses/LICENSE-2.0

   Unless required by applicable law or agreed to in writing, software
   distributed under the License is distributed on an "AS IS" BASIS,
   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
   See the License for the specific language governing permissions and
   limitations under the License.


================================================
FILE: MANIFEST.in
================================================
global-exclude notebooks/*
global-exclude .git*
global-exclude static/*
global-exclude *png
global-exclude *pkl


================================================
FILE: README.md
================================================
The MNIST-1D Dataset
=======

ICML 2024 | [Blog post](https://greydanus.github.io/2020/12/01/scaling-down/) | [Paper at arXiv](https://arxiv.org/abs/2011.14439) | [Paper at OpenReview](https://openreview.net/forum?id=n9pru4bJU9) | [GitHub](https://github.com/greydanus/mnist1d)

Most machine learning models get around the same ~99% test accuracy on MNIST. Our dataset, MNIST-1D, is 100x smaller (default sample size: 4000+1000; dimensionality: 40) and does a better job of separating between models with/without nonlinearity and models with/without spatial inductive biases.

_**Dec 5, 2023**: MNIST-1D is now a core teaching dataset in Simon Prince's [Understanding Deep Learning](https://udlbook.github.io/udlbook/) textbook_

Citation:
```
@inproceedings{greydanus2024scaling,
  title={Scaling down deep learning with {MNIST}-{1D}},
  author={Greydanus, Sam and Kobak, Dmitry},
  booktitle={Proceedings of the 41st International Conference on Machine Learning},
  year={2024}
}
```

![overview.png](static/overview.png)

Quickstart and use cases
--------
* Getting started
  * [Quickstart](https://github.com/greydanus/mnist1d/blob/master/notebooks/quickstart.ipynb) ([Colab](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/quickstart.ipynb))
  * [Building MNIST-1D](https://github.com/greydanus/mnist1d/blob/master/notebooks/building-mnist1d.ipynb) ([Colab](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/building-mnist1d.ipynb))
  * [Pip installation (3 lines)](https://github.com/greydanus/mnist1d/blob/master/notebooks/mnist1d-pip.ipynb) ([Colab](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/mnist1d-pip.ipynb))
* Example use cases
  *  [Quantifying CNN spatial priors](https://github.com/greydanus/mnist1d/blob/master/notebooks/mnist1d-classification.ipynb) ([Colab](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/mnist1d-classification.ipynb))
  * [Self-supervised learning](https://github.com/greydanus/mnist1d/blob/master/notebooks/self-supervised-learning.ipynb) ([Colab](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/self-supervised-learning.ipynb))
  * [Finding lottery tickets](https://github.com/greydanus/mnist1d/blob/master/notebooks/lottery-tickets.ipynb) ([Colab](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/lottery-tickets.ipynb))
  * [Observing deep double descent](https://github.com/greydanus/mnist1d/blob/master/notebooks/deep-double-descent.ipynb) ([Colab](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/deep-double-descent.ipynb))
  * [Metalearning a learning rate](https://github.com/greydanus/mnist1d/blob/master/notebooks/metalearn-learn-rate.ipynb) ([Colab](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/metalearn-learn-rate.ipynb))
  * [Metalearning an activation function](https://github.com/greydanus/mnist1d/blob/master/notebooks/metalearn-activation-function.ipynb) ([Colab](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/metalearn-activation-function.ipynb))
  * [Benchmarking pooling methods](https://github.com/greydanus/mnist1d/blob/master/notebooks/benchmark-pooling.ipynb) ([Colab](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/benchmark-pooling.ipynb))
  * [t-SNE visualisations of MNIST-1D and MNIST](https://github.com/greydanus/mnist1d/blob/master/notebooks/tsne-mnist-vs-mnist1d.ipynb) ([Colab](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/tsne-mnist-vs-mnist1d.ipynb))
* Community contributions
  * [A from-scratch, Numpy-only MLP with handwritten backprop](https://colab.research.google.com/drive/1E4w9chTkK-rPK-Zl-D0t4Q3FrdpQrHRQ?usp=sharing)
  * [Simon Prince's _Understanding Deep Learning_](https://udlbook.github.io/udlbook/) textbook uses MNIST1D as a core teaching example
  * [A matching GoLang implementation](https://github.com/mdcfrancis/gomnist1d) including matching random seeds, etc. by [mdcfrancis](https://github.com/mdcfrancis/)
  * This dataset is also on [HuggingFace](https://huggingface.co/datasets/christopher/mnist1d) (thanks to [@cakiki](https://github.com/cakiki))
  * ...send me a Colab link to your experiment and I'll feature it here.


Installing with `pip`
--------

``` shell
pip install mnist1d
```

This allows you to build the default dataset locally:

```python
from mnist1d.data import make_dataset, get_dataset_args

defaults = get_dataset_args()
data = make_dataset(defaults)
x, y, t = data['x'], data['y'], data['t']
```

If you want to play around with this, see [notebooks/mnist1d-pip.ipynb](https://github.com/greydanus/mnist1d/blob/master/notebooks/mnist1d-pip.ipynb).


Alternatively, you can always `pip install` via the GitHub repo:

``` shell
python -m pip install git+https://github.com/greydanus/mnist1d.git@master
```


Comparing MNIST and MNIST-1D
--------

| Dataset		| Logistic regression		| MLP 	| CNN 	| GRU* | Human expert |
| ------------- 			| :---------------: | :---------------: | :---------------: | :---------------: | :---------------: |
| MNIST 					    | 94% | 99+% | 99+% | 99+% | 99+% |
| MNIST-1D 					  | 32% | 68%  | 94%  | 91%  | 96%  |
| MNIST-1D (shuffle**)	| 32% | 68%  | 56%  | 57%  | ~30% |

*Training the GRU takes at least 10x the walltime of the CNN.

**The term "shuffle" refers to shuffling the spatial dimension of the dataset, as in [Zhang et al. (2017)](https://arxiv.org/abs/1611.03530).


-----------

According to Geoffrey Hinton, the original MNIST dataset is the [Drosophila of machine learning](https://twitter.com/ivanukhov/status/639122460722528257). But we argue that it has a few drawbacks:
* **Discrimination between models.** The difference between major ML models comes down to a few percentage points.
* **Dimensionality.** Examples are 784-dimensional vectors so training ML models can take non-trivial compute and memory (think neural architecture search and metalearning).
* **Hard to hack.** MNIST is not procedurally generated so it's hard to change the noise distribution, the scale/rotation/translation/shear/etc of the digits, or the resolution.

 We developed MNIST-1D to address these issues. It is:
* **Discriminative between models.** There is a broad spread in test accuracy between key ML models.
* **Low dimensional.** Each MNIST-1D example is a 40-dimensional vector. This means faster training and less memory.
* **Easy to hack.** There's an API for adjusting max_translation, corr_noise_scale, shear_scale, final_seq_length and more. The code is clean and modular.
* **Still has some real-world relevance.** Though it's low-dimensional and synthetic, this task is arguably more interesting than [Sklearn's datasets](https://scikit-learn.org/stable/modules/classes.html#module-sklearn.datasets) such as two_moons, two_circles, or gaussian_blobs.

Dimensionality reduction
--------

Visualizing the MNIST and MNIST-1D datasets with t-SNE. The well-defined clusters in the MNIST plot indicate that the majority of the examples are separable via a kNN classifier in pixel space. The MNIST-1D plot, meanwhile, reveals a lack of well-defined clusters which suggests that learning a nonlinear representation of the data is much more important to achieve successful classification.

<img src="notebooks/figures/tsne.png" width=500>

Downloading the dataset
--------

Here's a minimal example of how to download the frozen dataset. This is arguably worse than installing this repo with `pip` and generating it from scratch. But it does have its uses. It can also be used for double-checking that the procedurally generated dataset exactly matches the one used in the paper and blog post:

```python
from urllib.request import urlopen
import pickle

url = 'https://github.com/greydanus/mnist1d/raw/master/mnist1d_data.pkl'
data = pickle.load(urlopen(url))

data.keys()

>>> dict_keys(['x', 'x_test', 'y', 'y_test', 't', 'templates'])  # these are NumPy arrays
```


Constructing the dataset
--------

This is a synthetically-generated dataset which, by default, consists of 4000 training examples and 1000 testing examples (you can change this as you wish). Each example contains a template pattern that resembles a handwritten digit between 0 and 9. These patterns are analogous to the digits in the original [MNIST dataset](http://yann.lecun.com/exdb/mnist/).

**Original MNIST digits**

![mnist1d_black.png](static/mnist.png)

**1D template patterns**

![mnist1d_black.png](static/mnist1d_black_small.png)

**1D templates as lines**

![mnist1d_white.png](static/mnist1d_white_small.png)

In order to build the synthetic dataset, we pass the templates through a series of random transformations. This includes adding random amounts of padding, translation, correlated noise, iid noise, and scaling. We use these transformations because they are relevant for both 1D signals and 2D images. So even though our dataset is 1D, we can expect some of our findings to hold for 2D (image) data. For example, we can study the advantage of using a translation-invariant model (eg. a CNN) by making a dataset where signals occur at different locations in the sequence. We can do this by using large padding and translation coefficients. Here's an animation of how those transformations are applied.

![mnist1d_tranforms.gif](static/mnist1d_transforms.gif)

Unlike the original MNIST dataset, which consisted of 2D arrays of pixels (each image had 28x28=784 dimensions), this dataset consists of 1D timeseries of length 40. This means each example is ~20x smaller, making the dataset much quicker and easier to iterate over. Another nice thing about this toy dataset is that it does a good job of separating different types of deep learning models, many of which get the same 98-99% test accuracy on MNIST.


Example use cases
--------

### [Quantifying CNN spatial priors](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/mnist1d-classification.ipynb)
For a fixed number of training examples, we show that a CNN achieves far better test generalization than a comparable MLP. This highlights the value of the inductive biases that we build into ML models.

<img src="notebooks/figures/benchmark.png" width=500>

### [Finding lottery tickets](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/lottery-tickets.ipynb)
We obtain sparse "lottery ticket" masks as described by [Frankle & Carbin (2018)](https://arxiv.org/abs/1803.03635). Then we perform some ablation studies and analysis on them to determine exactly what makes these masks special (spoiler: they have spatial priors including local connectivity). One result, which contradicts the original paper, is that lottery ticket masks can be beneficial even under different initial weights. We suspect this effect is present but vanishingly small in the experiments performed by Frankle & Carbin.

![lottery.png](static/lottery.png)

![lottery_summary.png](static/lottery_summary_small.png)

### [Observing deep double descent](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/deep-double-descent.ipynb)
We replicate the "deep double descent" phenomenon described by [Belkin et al. (2018)](https://arxiv.org/abs/1812.11118) and more recently studied at scale by [Nakkiran et al. (2019)](https://openai.com/blog/deep-double-descent/).

<img src="notebooks/figures/double-descent.png" width=500>

### [Metalearning a learning rate](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/metalearn-learn-rate.ipynb)
A simple notebook that introduces gradient-based metalearning, also known as "unrolled optimization." In the spirit of [Maclaurin et al (2015)](http://proceedings.mlr.press/v37/maclaurin15.pdf) we use this technique to obtain the optimal learning rate for an MLP.

![metalearn_lr.png](static/metalearn_lr.png)

### [Metalearning an activation function](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/metalearn-activation-function.ipynb)
This project uses the same principles as the learning rate example, but tackles a new problem that (to our knowledge) has not been tackled via gradient-based metalearning: how to obtain the perfect nonlinearity for a neural network. We start from an ELU activation function and parameterize the offset with an MLP. We use unrolled optimization to find the offset that leads to lowest training loss, across the last 200 steps, for an MLP classifier trained on MNIST-1D. Interestingly, the result somewhat resembles the Swish activation described by [Ramachandran et al. (2017)](https://arxiv.org/abs/1710.05941); the main difference is a positive regime between -4 and -1.

![metalearn_afunc.png](static/metalearn_afunc.png)

### [Benchmarking pooling methods](https://githubtocolab.com/greydanus/mnist1d/blob/master/notebooks/benchmark-pooling.ipynb)
We investigate the relationship between number of training samples and usefulness of pooling methods. We find that pooling is typically very useful in the low-data regime but this advantage diminishes as the amount of training data increases.

![pooling.png](static/pooling.png)


Dependencies
--------
 * NumPy
 * SciPy
 * PyTorch
 * (others)


================================================
FILE: mnist1d/.gitignore
================================================
py3*/


================================================
FILE: mnist1d/__init__.py
================================================
# The MNIST-1D dataset | 2024
# Sam Greydanus, Peter Steinbach

from .data import get_dataset, get_dataset_args, get_templates
from .utils import to_pickle, from_pickle, ObjectView, set_seed, plot_signals
from .transform import transform

================================================
FILE: mnist1d/data.py
================================================
# The MNIST-1D dataset | 2020
# Sam Greydanus

import numpy as np
import os
import requests
import mnist1d
from mnist1d.transform import transform
from mnist1d.utils import from_pickle, to_pickle, ObjectView, set_seed

def get_dataset_args(as_dict=False):
    """ Generate dictionary with dataset properties

    Parameters
    ----------
    as_dict : bool, optional
        if true, return the dataset properties as dictionary; if false, return an ObjectView, by default False

    Returns
    -------
    _type_
        _description_
    """
    arg_dict = {'num_samples': 5000,
            'train_split': 0.8,
            'template_len': 12,
            'padding': [36,60],
            'scale_coeff': .4, 
            'max_translation': 48,
            'corr_noise_scale': 0.25,
            'iid_noise_scale': 2e-2,
            'shear_scale': 0.75,
            'shuffle_seq': False,
            'final_seq_length': 40,
            'seed': 42,
            'url': 'https://github.com/greydanus/mnist1d/raw/master/mnist1d_data.pkl'}
    return arg_dict if as_dict else ObjectView(arg_dict)


# basic 1D templates for the 10 digits
def get_templates():
    d0 = np.asarray([5,6,6.5,6.75,7,7,7,7,6.75,6.5,6,5])
    d1 = np.asarray([5,3,3,3.4,3.8,4.2,4.6,5,5.4,5.8,5,5])
    d2 = np.asarray([5,6,6.5,6.5,6,5.25,4.75,4,3.5,3.5,4,5])
    d3 = np.asarray([5,6,6.5,6.5,6,5,5,6,6.5,6.5,6,5])
    d4 = np.asarray([5,4.4,3.8,3.2,2.6,2.6,5,5,5,5,5,5])
    d5 = np.asarray([5,3,3,3,3,5,6,6.5,6.5,6,4.5,5])
    d6 = np.asarray([5,4,3.5,3.25,3,3,3,3,3.25,3.5,4,5])
    d7 = np.asarray([5,7,7,6.6,6.2,5.8,5.4,5,4.6,4.2,5,5])
    d8 = np.asarray([5,4,3.5,3.5,4,5,5,4,3.5,3.5,4,5])
    d9 = np.asarray([5,4,3.5,3.5,4,5,5,5,5,4.7,4.3,5])
    
    x = np.stack([d0,d1,d2,d3,d4,d5,d6,d7,d8,d9])
    x -= x.mean(1,keepdims=True) # whiten
    x /= x.std(1,keepdims=True)
    x -= x[:,:1]  # signal starts and ends at 0
    
    templates = {'x': x/6., 't': np.linspace(-5, 5, len(d0))/6.,
                 'y': np.asarray([0,1,2,3,4,5,6,7,8,9])}
    return templates


# make a dataset
def make_dataset(args=None, template=None, ):
    templates = get_templates() if template is None else template
    args = get_dataset_args() if args is None else args
    set_seed(args.seed) # reproducibility
    
    xs, ys = [], []
    samples_per_class = args.num_samples // len(templates['y'])
    for label_ix in range(len(templates['y'])):
        for example_ix in range(samples_per_class):
            x = templates['x'][label_ix]
            t = templates['t']
            y = templates['y'][label_ix]
            x, new_t = transform(x, t, args) # new_t transformation is same each time
            xs.append(x) ; ys.append(y)
    
    batch_shuffle = np.random.permutation(len(ys)) # shuffle batch dimension
    xs = np.stack(xs)[batch_shuffle]
    ys = np.stack(ys)[batch_shuffle]
    
    if args.shuffle_seq: # maybe shuffle the spatial dimension
        seq_shuffle = np.random.permutation(args.final_seq_length)
        xs = xs[...,seq_shuffle]
    
    new_t = new_t/xs.std()
    xs = (xs-xs.mean())/xs.std() # center the dataset & set standard deviation to 1

    # train / test split
    split_ix = int(len(ys)*args.train_split)
    dataset = {'x': xs[:split_ix], 'x_test': xs[split_ix:],
               'y': ys[:split_ix], 'y_test': ys[split_ix:],
               't':new_t, 'templates': templates}
    return dataset



# we'll cache the dataset so that it doesn't have to be rebuild every time
# args must not be a dict
def get_dataset(args, path=None, verbose=True, download=True, regenerate=False, **kwargs):
    if 'args' in kwargs.keys() and kwargs['args'].shuffle_seq:
        shuffle = "_shuffle"
    else:
        shuffle = ""
    path = './mnist1d_data{}.pkl'.format(shuffle) if path is None else path

    assert not (download and regenerate), "You can either download the o.g. MNIST1D dataset or generate your own - but not both"
    try:
        if regenerate:
            raise ValueError("Regenerating dataset") # yes this is hacky
        if download:
            if os.path.exists(path):
                if verbose:
                    print("File already exists. Skipping download.")
            else:
                print("Downloading MNIST1D dataset from {}".format(args.url))
                r = requests.get(args.url, allow_redirects=True)
                open(path, 'wb').write(r.content)
                print("Saving to {}".format(path))
        dataset = from_pickle(path)
        if verbose:
            print("Successfully loaded data from {}".format(path))
    except:
        if verbose:
            print("Did or could not load data from {}. Rebuilding dataset...".format(path))
        dataset = make_dataset(args, **kwargs)
        to_pickle(dataset, path)
    return dataset

================================================
FILE: mnist1d/transform.py
================================================
# The MNIST-1D dataset | 2024
# Sam Greydanus, Peter Steinbach

import numpy as np
import scipy
from scipy.interpolate import interp1d
from scipy.ndimage import gaussian_filter


# transformations of the templates which will make them harder to classify
def pad(x, padding: tuple):
    """pad signal x with random number of zeros. Note, the signal is only padded at indices given by the interval in padding

    Parameters
    ----------
    x : _type_
        signal
    padding : tuple
        (low, high) corresponds to (start,end) of padding

    Returns
    -------
    _type_
        a padded signal
    """
    low, high = padding
    p = low + int(np.random.rand() * (high - low + 1))
    if len(x.shape) == 1:
        return np.concatenate([x, np.zeros((p))])
    else:
        padding = np.zeros((x.shape[0], p))
        return np.concatenate([x, padding], axis=-1)


def shear(x, scale=10):
    # TODO: add docstring
    coeff = scale * (np.random.rand() - 0.5)
    return x - coeff * np.linspace(-0.5, 0.5, len(x))


def translate(x, max_translation):
    # TODO: add docstring
    k = np.random.choice(max_translation)
    return np.concatenate([x[-k:], x[:-k]])


def corr_noise_like(x, scale):
    # TODO: add docstring
    noise = scale * np.random.randn(*x.shape)
    return gaussian_filter(noise, 2)


def iid_noise_like(x, scale):
    # TODO: add docstring
    noise = scale * np.random.randn(*x.shape)
    return noise


def interpolate(x, N):
    # TODO: add docstring
    scale = np.linspace(0, 1, len(x))
    new_scale = np.linspace(0, 1, N)
    new_x = interp1d(scale, x, axis=0, kind="linear")(new_scale)
    return new_x


def transform(x, y, args, eps=1e-8):
    new_x = pad(x + eps, args.padding)  # pad
    new_x = interpolate(new_x, args.template_len + args.padding[-1])  # dilate
    new_y = interpolate(y, args.template_len + args.padding[-1])
    new_x *= 1 + args.scale_coeff * (np.random.rand() - 0.5)  # scale
    new_x = translate(new_x, args.max_translation)  # translate

    # add noise
    mask = new_x != 0
    new_x = mask * new_x + (1 - mask) * corr_noise_like(new_x, args.corr_noise_scale)
    new_x = new_x + iid_noise_like(new_x, args.iid_noise_scale)

    # shear and interpolate
    new_x = shear(new_x, args.shear_scale)
    new_x = interpolate(new_x, args.final_seq_length)  # subsample
    new_y = interpolate(new_y, args.final_seq_length)
    return new_x, new_y


================================================
FILE: mnist1d/utils.py
================================================
# The MNIST-1D dataset | 2024
# Sam Greydanus, Peter Steinbach

import numpy as np
import random
import pickle
import matplotlib.pyplot as plt
from mnist1d.transform import transform

def set_seed(seed):
    random.seed(seed)
    np.random.seed(seed)

def to_pickle(thing, path): # save something
    with open(path, 'wb') as handle:
        pickle.dump(thing, handle, protocol=3)


def from_pickle(path): # load something
    value = None
    with open(path, 'rb') as handle:
        value = pickle.load(handle)
    return value

class ObjectView(object):
    def __init__(self, d): self.__dict__ = d


def plot_signals(xs, t, labels=None, args=None, ratio=2.6, do_transform=False, dark_mode=False, zoom=1):
    rows, cols = 1, 10
    fig = plt.figure(figsize=[cols*1.5,rows*1.5*ratio], dpi=60)
    for r in range(rows):
        for c in range(cols):
            ix = r*cols + c
            x, t = xs[ix], t
            ax = plt.subplot(rows,cols,ix+1)

            # plot the data
            if do_transform:
                assert args is not None, "Need an args object in order to do transforms"
                x, t = transform(x, t, args)  # optionally, transform the signal in some manner
            if dark_mode:
                plt.plot(x, t, 'wo', linewidth=6)
                ax.set_facecolor('k')
            else:
                plt.plot(x, t, 'k-', linewidth=2)
            if labels is not None:
                plt.title("label=" + str(labels[ix]), fontsize=22)

            plt.xlim(-zoom,zoom) ; plt.ylim(-zoom,zoom)
            plt.gca().invert_yaxis() ; plt.xticks([], []), plt.yticks([], [])
    plt.subplots_adjust(wspace=0, hspace=0)
    plt.tight_layout() ; plt.show()
    return fig

================================================
FILE: notebooks/README.md
================================================
# Try out the MNIST1D dataset

## Installation instructions

1. clone the repo
``` shell
git clone https://github.com/greydanus/mnist1d.git
```
2. install all additional depdendencies
``` shell
cd mnist1d
python -m pip install '.[all]'
```
3. run the notebooks, [quickstart.ipynb](quickstart.ipynb) or [mnist1d_tiny.ipynb](mnist1d-pip.ipynb) are a good start.


================================================
FILE: notebooks/benchmark-pooling.ipynb
================================================
{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dKUcDM76bHx3"
      },
      "source": [
        "# **MNIST-1D**: Benchmarking pooling methods\n",
        "Sam Greydanus\n",
        "\n",
        "In this notebook, I will examine the effect of pooling on model generalization and sample efficiency. It looks like pooling is better than striding (or no pooling) on small datasets, but the discrepancy decreases as the number of training examples increase. I'm not yet sure why this is, but my best hypothesis is that pooling is a mediocre architectural prior which is better than nothing in low-data regimes and then ends up restricting model expression in high-data regimes. Let me know if you have a better explanation.\n",
        "\n",
        "This case study is meant to show the convenience and computational savings of working with the low-dimensional MNIST-1D dataset. You can find details at https://github.com/greydanus/mnist1d."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "id": "Sg2i1QmhKW5d"
      },
      "outputs": [],
      "source": [
        "!python -m pip install git+https://github.com/greydanus/mnist1d.git@master\n",
        "\n",
        "# Download repo directly (gives access to notebooks/models.py and notebooks/train.py)\n",
        "!git clone https://github.com/greydanus/mnist1d"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KaQo7QhvXvid",
        "outputId": "1cb2675c-776e-4891-db05-1c4649c15d3c"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Using: cuda\n"
          ]
        }
      ],
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from scipy.ndimage import gaussian_filter\n",
        "import torch, os\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "\n",
        "import mnist1d\n",
        "\n",
        "# Try attaching to GPU\n",
        "DEVICE = str(torch.device('cuda' if torch.cuda.is_available() else 'cpu'))\n",
        "print('Using:', DEVICE)\n",
        "\n",
        "plt.style.use('https://github.com/greydanus/mnist1d/raw/master/notebooks/mpl_style.txt')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "id": "cxzRpVn1wO1h"
      },
      "outputs": [],
      "source": [
        "# from mnist1d.data import get_dataset, get_dataset_args\n",
        "# from mnist1d.utils import set_seed, to_pickle\n",
        "\n",
        "import sys ; sys.path.append('./mnist1d/notebooks')\n",
        "from train import get_model_args, train_model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CP-Pmxw3wO1h"
      },
      "source": [
        "## Only run this if you're in Google Colab"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JjJoQ1KAwO1i",
        "outputId": "39c80efe-59ce-49a1-f656-a26461e7e43a"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/gdrive\n"
          ]
        }
      ],
      "source": [
        "if True:\n",
        "    # Only run this in Colab\n",
        "    from google.colab import drive\n",
        "    drive.mount('/content/gdrive')\n",
        "    project_dir = \"/content/gdrive/My Drive/Research/mnist1d/\"\n",
        "else:\n",
        "    project_dir = './'"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "O2vy0FKjfDwr"
      },
      "source": [
        "## Systematically generate different pooling settings\n",
        "We'll need to define a custom model for this. It looks similar to the `ConvBase` model in [mnist1d/models.py](https://github.com/greydanus/mnist1d/blob/master/models.py)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "id": "uBx5gNW-mqH_"
      },
      "outputs": [],
      "source": [
        "\n",
        "class ConvPooling(nn.Module):\n",
        "    def __init__(self, output_size, channels=20, stride=1, pool_fun=None, linear_in=200, seed=0):\n",
        "        super(ConvPooling, self).__init__()\n",
        "        _ = torch.manual_seed(seed) # use same initialization each time\n",
        "        self.conv1 = nn.Conv1d(1, channels, 3, stride=stride, padding=1)\n",
        "        self.conv2 = nn.Conv1d(channels, channels, 3, stride=stride, padding=1)\n",
        "        self.linear = nn.Linear(linear_in, output_size)\n",
        "        max_pool = lambda x: F.max_pool1d(x, 2)\n",
        "        self.pool_fun = max_pool if pool_fun is None else pool_fun\n",
        "\n",
        "    def forward(self, x): # the print statements are for debugging\n",
        "        x = x.view(-1,1,x.shape[-1])\n",
        "        h1 = self.pool_fun(self.conv1(x).relu())\n",
        "        h2 = self.pool_fun(self.conv2(h1).relu())\n",
        "        h2_flat = h2.view(h2.shape[0], -1) # flatten the conv features\n",
        "        # print(h2_flat.shape)\n",
        "        return self.linear(h2_flat) # a linear classifier goes on top\n",
        "\n",
        "def get_models(args):\n",
        "  models = [ConvPooling(args.output_size, stride=1, pool_fun=lambda x: x, linear_in=800),\n",
        "            ConvPooling(args.output_size, stride=2, pool_fun=lambda x: x),\n",
        "            ConvPooling(args.output_size, stride=1, pool_fun=lambda x: F.max_pool1d(x, 2)),\n",
        "            ConvPooling(args.output_size, stride=1, pool_fun=lambda x: F.avg_pool1d(x, 2)),\n",
        "            ConvPooling(args.output_size, stride=1, pool_fun=lambda x: F.lp_pool1d(x, norm_type=2, kernel_size=2))]\n",
        "\n",
        "  for m, n in zip(models, ['no_pool','stride_2','max_pool','avg_pool','l2_pool']):\n",
        "    m.name = n\n",
        "  return models"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XH8nHKh0Ss5M"
      },
      "source": [
        "## Let's see how pooling affects learning in different data regimes"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BD4YKFwiSeah",
        "outputId": "f06e465c-7f17-4ada-9a37-efeb8bbd963e"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Did or could not load data from ./mnist1d_data.pkl. Rebuilding dataset...\n",
            "\n",
            "New dataset, train set size=1000, test set size=1000\n",
            "step 1000, dt 3.05s, train_loss 1.452e-03, test_loss 2.785e+00, train_acc 100.0, test_acc 59.9\n",
            "step 2000, dt 1.60s, train_loss 3.115e-04, test_loss 3.279e+00, train_acc 100.0, test_acc 60.1\n",
            "step 3000, dt 1.57s, train_loss 1.155e-04, test_loss 3.599e+00, train_acc 100.0, test_acc 59.9\n",
            "step 4000, dt 1.53s, train_loss 5.209e-05, test_loss 3.858e+00, train_acc 100.0, test_acc 59.6\n",
            "step 5000, dt 1.56s, train_loss 2.587e-05, test_loss 4.088e+00, train_acc 100.0, test_acc 59.0\n",
            "step 6000, dt 1.53s, train_loss 1.361e-05, test_loss 4.296e+00, train_acc 100.0, test_acc 59.2\n",
            "step 1000, dt 1.61s, train_loss 1.222e-02, test_loss 2.728e+00, train_acc 100.0, test_acc 57.4\n",
            "step 2000, dt 1.49s, train_loss 1.769e-03, test_loss 3.687e+00, train_acc 100.0, test_acc 57.2\n",
            "step 3000, dt 1.55s, train_loss 5.796e-04, test_loss 4.239e+00, train_acc 100.0, test_acc 57.0\n",
            "step 4000, dt 1.54s, train_loss 2.459e-04, test_loss 4.657e+00, train_acc 100.0, test_acc 56.9\n",
            "step 5000, dt 1.51s, train_loss 1.183e-04, test_loss 5.018e+00, train_acc 100.0, test_acc 57.0\n",
            "step 6000, dt 1.54s, train_loss 6.033e-05, test_loss 5.345e+00, train_acc 100.0, test_acc 57.4\n",
            "step 1000, dt 1.79s, train_loss 2.974e-03, test_loss 1.021e+00, train_acc 100.0, test_acc 78.2\n",
            "step 2000, dt 1.74s, train_loss 6.399e-04, test_loss 1.253e+00, train_acc 100.0, test_acc 78.0\n",
            "step 3000, dt 1.74s, train_loss 2.322e-04, test_loss 1.404e+00, train_acc 100.0, test_acc 78.2\n",
            "step 4000, dt 1.76s, train_loss 1.042e-04, test_loss 1.523e+00, train_acc 100.0, test_acc 78.4\n",
            "step 5000, dt 1.74s, train_loss 5.147e-05, test_loss 1.626e+00, train_acc 100.0, test_acc 78.3\n",
            "step 6000, dt 1.74s, train_loss 2.692e-05, test_loss 1.724e+00, train_acc 100.0, test_acc 78.2\n",
            "step 1000, dt 1.67s, train_loss 3.266e-02, test_loss 1.082e+00, train_acc 99.9, test_acc 77.8\n",
            "step 2000, dt 1.69s, train_loss 1.565e-03, test_loss 1.417e+00, train_acc 100.0, test_acc 77.9\n",
            "step 3000, dt 1.67s, train_loss 5.245e-04, test_loss 1.634e+00, train_acc 100.0, test_acc 77.6\n",
            "step 4000, dt 1.67s, train_loss 2.236e-04, test_loss 1.808e+00, train_acc 100.0, test_acc 77.7\n",
            "step 5000, dt 1.66s, train_loss 1.063e-04, test_loss 1.961e+00, train_acc 100.0, test_acc 77.9\n",
            "step 6000, dt 1.73s, train_loss 5.385e-05, test_loss 2.103e+00, train_acc 100.0, test_acc 78.0\n",
            "step 1000, dt 2.70s, train_loss 2.322e-03, test_loss 1.150e+00, train_acc 100.0, test_acc 78.1\n",
            "step 2000, dt 2.49s, train_loss 4.466e-04, test_loss 1.389e+00, train_acc 100.0, test_acc 78.0\n",
            "step 3000, dt 2.46s, train_loss 1.593e-04, test_loss 1.544e+00, train_acc 100.0, test_acc 77.3\n",
            "step 4000, dt 2.41s, train_loss 7.082e-05, test_loss 1.665e+00, train_acc 100.0, test_acc 77.0\n",
            "step 5000, dt 2.46s, train_loss 3.477e-05, test_loss 1.772e+00, train_acc 100.0, test_acc 77.1\n",
            "step 6000, dt 2.45s, train_loss 1.819e-05, test_loss 1.870e+00, train_acc 100.0, test_acc 77.2\n",
            "Did or could not load data from ./mnist1d_data.pkl. Rebuilding dataset...\n",
            "\n",
            "New dataset, train set size=2000, test set size=1000\n",
            "step 1000, dt 1.62s, train_loss 6.647e-02, test_loss 1.052e+00, train_acc 96.3, test_acc 78.6\n",
            "step 2000, dt 1.62s, train_loss 6.551e-04, test_loss 1.108e+00, train_acc 100.0, test_acc 82.1\n",
            "step 3000, dt 1.56s, train_loss 2.801e-04, test_loss 1.226e+00, train_acc 100.0, test_acc 81.8\n",
            "step 4000, dt 1.59s, train_loss 1.443e-04, test_loss 1.319e+00, train_acc 100.0, test_acc 81.8\n",
            "step 5000, dt 1.65s, train_loss 7.967e-05, test_loss 1.407e+00, train_acc 100.0, test_acc 81.9\n",
            "step 6000, dt 1.60s, train_loss 4.415e-05, test_loss 1.487e+00, train_acc 100.0, test_acc 82.0\n",
            "step 1000, dt 1.55s, train_loss 1.372e-01, test_loss 9.035e-01, train_acc 90.2, test_acc 75.1\n",
            "step 2000, dt 1.53s, train_loss 2.271e-02, test_loss 9.773e-01, train_acc 98.2, test_acc 79.0\n",
            "step 3000, dt 1.53s, train_loss 6.885e-03, test_loss 1.190e+00, train_acc 100.0, test_acc 81.3\n",
            "step 4000, dt 1.54s, train_loss 2.373e-03, test_loss 1.408e+00, train_acc 100.0, test_acc 81.5\n",
            "step 5000, dt 1.54s, train_loss 1.016e-03, test_loss 1.599e+00, train_acc 100.0, test_acc 81.4\n",
            "step 6000, dt 1.53s, train_loss 5.105e-04, test_loss 1.771e+00, train_acc 100.0, test_acc 81.7\n",
            "step 1000, dt 1.81s, train_loss 1.631e-02, test_loss 5.726e-01, train_acc 99.4, test_acc 87.1\n",
            "step 2000, dt 1.76s, train_loss 1.546e-03, test_loss 6.931e-01, train_acc 100.0, test_acc 89.3\n",
            "step 3000, dt 1.78s, train_loss 5.004e-04, test_loss 7.743e-01, train_acc 100.0, test_acc 89.3\n",
            "step 4000, dt 1.77s, train_loss 2.421e-04, test_loss 8.333e-01, train_acc 100.0, test_acc 89.4\n",
            "step 5000, dt 1.77s, train_loss 1.294e-04, test_loss 8.888e-01, train_acc 100.0, test_acc 89.1\n",
            "step 6000, dt 1.76s, train_loss 7.083e-05, test_loss 9.415e-01, train_acc 100.0, test_acc 89.2\n",
            "step 1000, dt 1.74s, train_loss 8.820e-02, test_loss 5.222e-01, train_acc 96.2, test_acc 85.9\n",
            "step 2000, dt 1.73s, train_loss 1.877e-02, test_loss 5.738e-01, train_acc 99.2, test_acc 89.5\n",
            "step 3000, dt 1.70s, train_loss 1.401e-03, test_loss 5.621e-01, train_acc 100.0, test_acc 90.1\n",
            "step 4000, dt 1.71s, train_loss 6.315e-04, test_loss 6.110e-01, train_acc 100.0, test_acc 89.8\n",
            "step 5000, dt 1.72s, train_loss 3.401e-04, test_loss 6.533e-01, train_acc 100.0, test_acc 90.0\n",
            "step 6000, dt 1.74s, train_loss 1.949e-04, test_loss 6.957e-01, train_acc 100.0, test_acc 90.0\n",
            "step 1000, dt 2.47s, train_loss 6.138e-02, test_loss 4.424e-01, train_acc 98.0, test_acc 88.6\n",
            "step 2000, dt 2.50s, train_loss 1.116e-02, test_loss 4.875e-01, train_acc 100.0, test_acc 91.0\n",
            "step 3000, dt 2.45s, train_loss 7.532e-04, test_loss 5.874e-01, train_acc 100.0, test_acc 91.0\n",
            "step 4000, dt 2.44s, train_loss 3.389e-04, test_loss 6.328e-01, train_acc 100.0, test_acc 90.5\n",
            "step 5000, dt 2.45s, train_loss 1.825e-04, test_loss 6.732e-01, train_acc 100.0, test_acc 90.2\n",
            "step 6000, dt 2.45s, train_loss 1.038e-04, test_loss 7.133e-01, train_acc 100.0, test_acc 89.9\n",
            "Did or could not load data from ./mnist1d_data.pkl. Rebuilding dataset...\n",
            "\n",
            "New dataset, train set size=5000, test set size=1000\n",
            "step 1000, dt 1.60s, train_loss 6.357e-02, test_loss 3.726e-01, train_acc 97.9, test_acc 89.5\n",
            "step 2000, dt 1.55s, train_loss 6.281e-03, test_loss 4.866e-01, train_acc 99.6, test_acc 89.7\n",
            "step 3000, dt 1.58s, train_loss 2.777e-02, test_loss 5.479e-01, train_acc 99.2, test_acc 89.7\n",
            "step 4000, dt 1.58s, train_loss 1.460e-04, test_loss 4.633e-01, train_acc 100.0, test_acc 92.0\n",
            "step 5000, dt 1.58s, train_loss 9.093e-05, test_loss 4.866e-01, train_acc 100.0, test_acc 92.0\n",
            "step 6000, dt 1.61s, train_loss 6.229e-05, test_loss 5.048e-01, train_acc 100.0, test_acc 91.9\n",
            "step 1000, dt 1.62s, train_loss 1.056e-01, test_loss 4.257e-01, train_acc 94.5, test_acc 85.8\n",
            "step 2000, dt 1.59s, train_loss 1.069e-01, test_loss 5.056e-01, train_acc 96.3, test_acc 88.5\n",
            "step 3000, dt 1.57s, train_loss 3.146e-02, test_loss 5.121e-01, train_acc 97.8, test_acc 88.4\n",
            "step 4000, dt 1.59s, train_loss 5.262e-02, test_loss 5.659e-01, train_acc 98.6, test_acc 89.5\n",
            "step 5000, dt 1.54s, train_loss 4.487e-02, test_loss 6.898e-01, train_acc 98.4, test_acc 88.9\n",
            "step 6000, dt 1.51s, train_loss 1.630e-03, test_loss 7.076e-01, train_acc 99.6, test_acc 89.9\n",
            "step 1000, dt 1.74s, train_loss 6.580e-02, test_loss 2.896e-01, train_acc 97.3, test_acc 92.4\n",
            "step 2000, dt 1.78s, train_loss 7.009e-02, test_loss 4.034e-01, train_acc 96.5, test_acc 91.6\n",
            "step 3000, dt 1.77s, train_loss 1.755e-02, test_loss 3.655e-01, train_acc 99.7, test_acc 93.7\n",
            "step 4000, dt 1.74s, train_loss 1.276e-03, test_loss 4.159e-01, train_acc 100.0, test_acc 94.2\n",
            "step 5000, dt 1.73s, train_loss 4.149e-04, test_loss 4.464e-01, train_acc 100.0, test_acc 94.6\n",
            "step 6000, dt 1.73s, train_loss 2.434e-04, test_loss 4.736e-01, train_acc 100.0, test_acc 94.5\n",
            "step 1000, dt 1.69s, train_loss 1.472e-01, test_loss 3.249e-01, train_acc 94.3, test_acc 88.6\n",
            "step 2000, dt 1.68s, train_loss 5.878e-02, test_loss 2.491e-01, train_acc 97.4, test_acc 93.5\n",
            "step 3000, dt 1.72s, train_loss 3.734e-02, test_loss 2.645e-01, train_acc 98.1, test_acc 93.3\n",
            "step 4000, dt 1.75s, train_loss 8.816e-03, test_loss 2.959e-01, train_acc 98.7, test_acc 93.1\n",
            "step 5000, dt 1.70s, train_loss 2.268e-02, test_loss 3.870e-01, train_acc 98.9, test_acc 93.3\n",
            "step 6000, dt 1.73s, train_loss 2.471e-02, test_loss 3.406e-01, train_acc 99.2, test_acc 94.0\n",
            "step 1000, dt 2.49s, train_loss 5.895e-02, test_loss 2.953e-01, train_acc 96.1, test_acc 90.9\n",
            "step 2000, dt 2.47s, train_loss 2.253e-02, test_loss 2.380e-01, train_acc 99.2, test_acc 94.4\n",
            "step 3000, dt 2.51s, train_loss 3.262e-03, test_loss 2.879e-01, train_acc 99.9, test_acc 94.0\n",
            "step 4000, dt 2.48s, train_loss 2.204e-03, test_loss 3.246e-01, train_acc 99.9, test_acc 94.4\n",
            "step 5000, dt 2.50s, train_loss 1.358e-02, test_loss 4.344e-01, train_acc 99.1, test_acc 94.0\n",
            "step 6000, dt 2.45s, train_loss 5.408e-04, test_loss 3.288e-01, train_acc 100.0, test_acc 95.3\n",
            "Did or could not load data from ./mnist1d_data.pkl. Rebuilding dataset...\n",
            "\n",
            "New dataset, train set size=10000, test set size=1000\n",
            "step 1000, dt 1.61s, train_loss 6.490e-02, test_loss 2.128e-01, train_acc 96.6, test_acc 93.7\n",
            "step 2000, dt 1.59s, train_loss 3.926e-02, test_loss 2.481e-01, train_acc 97.6, test_acc 92.6\n",
            "step 3000, dt 1.57s, train_loss 2.231e-02, test_loss 3.569e-01, train_acc 96.2, test_acc 92.5\n",
            "step 4000, dt 1.60s, train_loss 2.069e-02, test_loss 2.970e-01, train_acc 98.7, test_acc 93.8\n",
            "step 5000, dt 1.62s, train_loss 3.838e-02, test_loss 2.727e-01, train_acc 99.0, test_acc 94.9\n",
            "step 6000, dt 1.61s, train_loss 5.584e-02, test_loss 3.307e-01, train_acc 98.3, test_acc 94.3\n",
            "step 1000, dt 1.57s, train_loss 1.529e-01, test_loss 2.215e-01, train_acc 94.1, test_acc 92.4\n",
            "step 2000, dt 1.56s, train_loss 8.962e-02, test_loss 2.316e-01, train_acc 95.8, test_acc 94.4\n",
            "step 3000, dt 1.54s, train_loss 1.006e-01, test_loss 2.705e-01, train_acc 95.5, test_acc 92.2\n",
            "step 4000, dt 1.53s, train_loss 6.084e-02, test_loss 3.377e-01, train_acc 95.5, test_acc 92.7\n",
            "step 5000, dt 1.53s, train_loss 6.458e-02, test_loss 3.558e-01, train_acc 96.8, test_acc 92.1\n",
            "step 6000, dt 1.55s, train_loss 3.079e-02, test_loss 4.216e-01, train_acc 96.4, test_acc 91.3\n",
            "step 1000, dt 1.74s, train_loss 3.803e-02, test_loss 1.283e-01, train_acc 96.1, test_acc 95.3\n",
            "step 2000, dt 1.77s, train_loss 5.471e-02, test_loss 1.476e-01, train_acc 96.3, test_acc 94.6\n",
            "step 3000, dt 1.79s, train_loss 2.909e-02, test_loss 1.753e-01, train_acc 96.2, test_acc 94.6\n",
            "step 4000, dt 1.75s, train_loss 1.382e-02, test_loss 1.087e-01, train_acc 97.9, test_acc 96.5\n",
            "step 5000, dt 1.73s, train_loss 1.656e-02, test_loss 1.651e-01, train_acc 97.5, test_acc 94.8\n",
            "step 6000, dt 1.74s, train_loss 9.807e-03, test_loss 1.033e-01, train_acc 98.9, test_acc 97.0\n",
            "step 1000, dt 1.73s, train_loss 6.617e-02, test_loss 1.560e-01, train_acc 94.0, test_acc 94.1\n",
            "step 2000, dt 1.75s, train_loss 1.983e-02, test_loss 1.080e-01, train_acc 96.8, test_acc 96.4\n",
            "step 3000, dt 1.72s, train_loss 2.736e-02, test_loss 1.253e-01, train_acc 97.2, test_acc 96.2\n",
            "step 4000, dt 1.77s, train_loss 9.317e-03, test_loss 1.427e-01, train_acc 96.8, test_acc 95.2\n",
            "step 5000, dt 1.71s, train_loss 4.564e-03, test_loss 1.508e-01, train_acc 97.6, test_acc 96.4\n",
            "step 6000, dt 1.71s, train_loss 2.449e-03, test_loss 1.681e-01, train_acc 97.7, test_acc 96.0\n",
            "step 1000, dt 2.46s, train_loss 5.464e-02, test_loss 1.262e-01, train_acc 95.9, test_acc 95.5\n",
            "step 2000, dt 2.48s, train_loss 2.815e-02, test_loss 9.669e-02, train_acc 97.7, test_acc 96.3\n",
            "step 3000, dt 2.52s, train_loss 5.754e-02, test_loss 1.387e-01, train_acc 97.2, test_acc 95.5\n",
            "step 4000, dt 2.51s, train_loss 3.447e-02, test_loss 1.724e-01, train_acc 96.1, test_acc 95.4\n",
            "step 5000, dt 2.47s, train_loss 1.558e-02, test_loss 1.057e-01, train_acc 98.3, test_acc 97.0\n",
            "step 6000, dt 2.50s, train_loss 5.077e-02, test_loss 1.476e-01, train_acc 98.0, test_acc 96.6\n",
            "Did or could not load data from ./mnist1d_data.pkl. Rebuilding dataset...\n",
            "\n",
            "New dataset, train set size=20000, test set size=1000\n",
            "step 1000, dt 1.63s, train_loss 1.885e-01, test_loss 1.926e-01, train_acc 94.7, test_acc 93.9\n",
            "step 2000, dt 1.61s, train_loss 1.854e-01, test_loss 1.803e-01, train_acc 94.7, test_acc 94.6\n",
            "step 3000, dt 1.68s, train_loss 9.691e-02, test_loss 1.736e-01, train_acc 96.0, test_acc 94.1\n",
            "step 4000, dt 1.69s, train_loss 4.889e-02, test_loss 1.545e-01, train_acc 96.5, test_acc 95.1\n",
            "step 5000, dt 1.63s, train_loss 8.644e-02, test_loss 1.541e-01, train_acc 96.5, test_acc 95.3\n",
            "step 6000, dt 1.62s, train_loss 5.165e-02, test_loss 1.904e-01, train_acc 96.3, test_acc 94.4\n",
            "step 1000, dt 1.55s, train_loss 4.043e-01, test_loss 2.679e-01, train_acc 91.4, test_acc 91.2\n",
            "step 2000, dt 1.56s, train_loss 1.870e-01, test_loss 2.070e-01, train_acc 94.1, test_acc 93.4\n",
            "step 3000, dt 1.56s, train_loss 1.352e-01, test_loss 2.134e-01, train_acc 94.0, test_acc 93.1\n",
            "step 4000, dt 1.56s, train_loss 1.081e-01, test_loss 2.186e-01, train_acc 94.7, test_acc 93.5\n",
            "step 5000, dt 1.57s, train_loss 8.404e-02, test_loss 2.497e-01, train_acc 95.1, test_acc 92.8\n",
            "step 6000, dt 1.61s, train_loss 9.615e-02, test_loss 2.977e-01, train_acc 94.7, test_acc 92.2\n",
            "step 1000, dt 1.79s, train_loss 1.787e-01, test_loss 1.251e-01, train_acc 96.6, test_acc 96.7\n",
            "step 2000, dt 1.77s, train_loss 8.813e-02, test_loss 1.334e-01, train_acc 97.3, test_acc 96.0\n",
            "step 3000, dt 1.79s, train_loss 6.658e-02, test_loss 1.014e-01, train_acc 97.9, test_acc 97.1\n",
            "step 4000, dt 1.79s, train_loss 2.576e-02, test_loss 9.413e-02, train_acc 98.2, test_acc 97.1\n",
            "step 5000, dt 1.81s, train_loss 3.511e-02, test_loss 1.047e-01, train_acc 98.0, test_acc 96.4\n",
            "step 6000, dt 1.84s, train_loss 2.144e-02, test_loss 1.194e-01, train_acc 98.2, test_acc 96.8\n",
            "step 1000, dt 1.77s, train_loss 1.431e-01, test_loss 1.761e-01, train_acc 94.2, test_acc 94.7\n",
            "step 2000, dt 1.74s, train_loss 8.928e-02, test_loss 1.217e-01, train_acc 97.2, test_acc 96.3\n",
            "step 3000, dt 1.73s, train_loss 4.217e-02, test_loss 1.089e-01, train_acc 97.9, test_acc 97.0\n",
            "step 4000, dt 1.71s, train_loss 1.961e-02, test_loss 1.013e-01, train_acc 98.6, test_acc 96.8\n",
            "step 5000, dt 1.75s, train_loss 3.450e-02, test_loss 1.004e-01, train_acc 98.4, test_acc 96.7\n",
            "step 6000, dt 1.74s, train_loss 1.617e-02, test_loss 1.172e-01, train_acc 97.7, test_acc 96.2\n",
            "step 1000, dt 2.56s, train_loss 1.061e-01, test_loss 1.580e-01, train_acc 95.2, test_acc 95.4\n",
            "step 2000, dt 2.50s, train_loss 6.005e-02, test_loss 1.012e-01, train_acc 97.7, test_acc 96.8\n",
            "step 3000, dt 2.50s, train_loss 8.669e-02, test_loss 9.806e-02, train_acc 98.0, test_acc 97.1\n",
            "step 4000, dt 2.48s, train_loss 4.814e-02, test_loss 7.269e-02, train_acc 98.6, test_acc 98.2\n",
            "step 5000, dt 2.50s, train_loss 5.166e-02, test_loss 9.254e-02, train_acc 98.3, test_acc 97.4\n",
            "step 6000, dt 2.56s, train_loss 4.138e-02, test_loss 8.727e-02, train_acc 98.6, test_acc 97.8\n",
            "Did or could not load data from ./mnist1d_data.pkl. Rebuilding dataset...\n",
            "\n",
            "New dataset, train set size=50000, test set size=1000\n",
            "step 1000, dt 1.70s, train_loss 1.733e-01, test_loss 1.549e-01, train_acc 96.4, test_acc 95.3\n",
            "step 2000, dt 1.67s, train_loss 8.866e-02, test_loss 1.494e-01, train_acc 96.6, test_acc 96.5\n",
            "step 3000, dt 1.66s, train_loss 1.025e-01, test_loss 1.449e-01, train_acc 97.1, test_acc 96.4\n",
            "step 4000, dt 1.70s, train_loss 1.023e-01, test_loss 1.439e-01, train_acc 97.4, test_acc 96.1\n",
            "step 5000, dt 1.69s, train_loss 1.172e-01, test_loss 1.432e-01, train_acc 97.4, test_acc 96.6\n",
            "step 6000, dt 1.70s, train_loss 8.589e-02, test_loss 1.643e-01, train_acc 96.9, test_acc 96.3\n",
            "step 1000, dt 1.63s, train_loss 2.023e-01, test_loss 2.477e-01, train_acc 92.7, test_acc 93.0\n",
            "step 2000, dt 1.65s, train_loss 1.273e-01, test_loss 1.865e-01, train_acc 94.7, test_acc 94.4\n",
            "step 3000, dt 1.64s, train_loss 9.741e-02, test_loss 1.758e-01, train_acc 95.0, test_acc 94.8\n",
            "step 4000, dt 1.61s, train_loss 6.318e-02, test_loss 1.748e-01, train_acc 95.6, test_acc 94.7\n",
            "step 5000, dt 1.64s, train_loss 7.016e-02, test_loss 1.626e-01, train_acc 96.1, test_acc 95.4\n",
            "step 6000, dt 1.63s, train_loss 9.876e-02, test_loss 1.687e-01, train_acc 95.9, test_acc 95.4\n",
            "step 1000, dt 1.90s, train_loss 8.345e-02, test_loss 1.365e-01, train_acc 96.5, test_acc 95.8\n",
            "step 2000, dt 1.83s, train_loss 5.780e-02, test_loss 1.120e-01, train_acc 97.2, test_acc 97.1\n",
            "step 3000, dt 1.82s, train_loss 3.284e-02, test_loss 9.956e-02, train_acc 97.6, test_acc 97.3\n",
            "step 4000, dt 1.86s, train_loss 1.737e-02, test_loss 7.605e-02, train_acc 98.3, test_acc 97.8\n",
            "step 5000, dt 1.84s, train_loss 6.650e-02, test_loss 8.646e-02, train_acc 98.2, test_acc 98.1\n",
            "step 6000, dt 1.83s, train_loss 2.151e-02, test_loss 7.358e-02, train_acc 98.5, test_acc 98.1\n",
            "step 1000, dt 1.83s, train_loss 1.803e-01, test_loss 2.064e-01, train_acc 93.7, test_acc 93.6\n",
            "step 2000, dt 1.82s, train_loss 1.939e-01, test_loss 1.279e-01, train_acc 96.9, test_acc 96.1\n",
            "step 3000, dt 1.79s, train_loss 1.401e-01, test_loss 1.037e-01, train_acc 97.8, test_acc 96.9\n",
            "step 4000, dt 1.78s, train_loss 1.196e-01, test_loss 9.660e-02, train_acc 98.2, test_acc 97.1\n",
            "step 5000, dt 1.76s, train_loss 1.295e-01, test_loss 1.037e-01, train_acc 98.0, test_acc 97.4\n",
            "step 6000, dt 1.79s, train_loss 1.177e-01, test_loss 9.063e-02, train_acc 98.3, test_acc 97.3\n",
            "step 1000, dt 2.63s, train_loss 1.067e-01, test_loss 1.470e-01, train_acc 96.3, test_acc 95.6\n",
            "step 2000, dt 2.60s, train_loss 1.235e-01, test_loss 1.275e-01, train_acc 97.2, test_acc 96.6\n",
            "step 3000, dt 2.58s, train_loss 1.193e-01, test_loss 1.283e-01, train_acc 97.0, test_acc 96.6\n",
            "step 4000, dt 2.56s, train_loss 8.530e-02, test_loss 1.038e-01, train_acc 97.5, test_acc 97.1\n",
            "step 5000, dt 2.58s, train_loss 1.378e-01, test_loss 8.621e-02, train_acc 98.0, test_acc 97.7\n",
            "step 6000, dt 2.60s, train_loss 9.037e-02, test_loss 9.042e-02, train_acc 97.7, test_acc 97.1\n"
          ]
        }
      ],
      "source": [
        "dataset_sizes = [2000, 3000, 6000, 11000, 21000, 51000]\n",
        "train_splits = [1-(1000./ds) for ds in dataset_sizes]  # split so that there are always 1000 test examples\n",
        "regime_results = []\n",
        "for i, num_samples in enumerate(dataset_sizes):\n",
        "\n",
        "  # make dataset\n",
        "  data_args = mnist1d.get_dataset_args()\n",
        "  data_args.num_samples = num_samples\n",
        "  data_args.train_split = train_splits[i]\n",
        "  dataset = mnist1d.get_dataset(args=data_args, download=False, regenerate=True)\n",
        "\n",
        "  # build the models\n",
        "  model_args = get_model_args()\n",
        "  model_args.total_steps = 6000\n",
        "  model_args.eval_every = 100\n",
        "  model_args.device = DEVICE\n",
        "  models = get_models(model_args)\n",
        "\n",
        "  # train the models, save results\n",
        "  print('\\nNew dataset, train set size={}, test set size={}'\\\n",
        "        .format(dataset['x'].shape[0], dataset['x_test'].shape[0]))\n",
        "  result = [train_model(dataset, m, model_args) for m in models]\n",
        "  regime_results.append(result)"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "rows, cols = 1, 3\n",
        "fig = plt.figure(figsize=[cols*3.2,rows*2.4], dpi=120)\n",
        "\n",
        "t = range(0, model_args.total_steps+1, model_args.eval_every)\n",
        "styles = ['k-', 'k--', 'r-', 'g-', 'b-']\n",
        "ylims = [[30,85],[80,100],[90,100]]\n",
        "runs_to_show = [0, 2, 5]\n",
        "\n",
        "for r in range(rows):\n",
        "    for c in range(cols):\n",
        "        ix = r*cols + c\n",
        "        ax = plt.subplot(rows,cols,ix+1)\n",
        "\n",
        "        results = regime_results[runs_to_show[ix]]\n",
        "        for j, result in enumerate(results):\n",
        "          model = result['checkpoints'][-1]\n",
        "          plt.plot(t, gaussian_filter(result['test_acc'], 1), styles[j], label=model.name)\n",
        "\n",
        "        plt.title('{} examples'.format(dataset_sizes[runs_to_show[ix]]-1000), fontsize=10)\n",
        "        if ix==0:\n",
        "          plt.ylabel(\"Test accuracy\", fontsize=9)\n",
        "          plt.xlabel(\"Train step\", fontsize=9)\n",
        "        if ix==0:\n",
        "          plt.legend(fontsize=8, ncol=2)\n",
        "        plt.ylim(*ylims[ix])\n",
        "\n",
        "plt.show()\n",
        "os.makedirs(project_dir + 'figures/', exist_ok=True)\n",
        "fig.savefig(project_dir + 'figures/pooling.png')\n",
        "fig.savefig(project_dir + 'figures/pooling.pdf')"
      ],
      "metadata": {
        "id": "rxWZwLWlxMFv",
        "outputId": "1e61f48c-73d4-4c40-d31d-a170b9838181",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 318
        }
      },
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1152x288 with 3 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "rows, cols = 2, 3\n",
        "fig = plt.figure(figsize=[cols*3,rows*2.3], dpi=120)\n",
        "\n",
        "t = range(0, model_args.total_steps+1, model_args.eval_every)\n",
        "styles = ['k-', 'k--', 'r-', 'g-', 'b-']\n",
        "ylims = [[30,85],[70,92],[80,100],[90,100],[90,100],[90,100]]\n",
        "\n",
        "for r in range(rows):\n",
        "    for c in range(cols):\n",
        "        ix = r*cols + c\n",
        "        ax = plt.subplot(rows,cols,ix+1)\n",
        "\n",
        "        results = regime_results[ix]\n",
        "        for j, result in enumerate(results):\n",
        "          model = result['checkpoints'][-1]\n",
        "          plt.plot(t, gaussian_filter(result['test_acc'], 1), styles[j], label=model.name)\n",
        "\n",
        "        plt.title('{} examples'.format(dataset_sizes[ix]-1000), fontsize=10)\n",
        "        if ix==0:\n",
        "          plt.ylabel(\"Test accuracy\", fontsize=9)\n",
        "          plt.xlabel(\"Train step\", fontsize=9)\n",
        "        if ix==0:\n",
        "          plt.legend(fontsize=8, ncol=2)\n",
        "        plt.ylim(*ylims[ix])\n",
        "\n",
        "plt.show()\n",
        "fig.savefig(project_dir + 'pooling_all.png')\n",
        "fig.savefig(project_dir + 'pooling_all.pdf')"
      ],
      "metadata": {
        "id": "L4n_TyCVxfO9",
        "outputId": "6a517928-5cd0-46d8-c3dc-841197a162b1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 582
        }
      },
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1080x552 with 6 Axes>"
            ],
            "image/png": "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