# head2toe

**Repository Path**: mirrors_google-research/head2toe

## Basic Information

- **Project Name**: head2toe
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Apache-2.0
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2022-01-12
- **Last Updated**: 2026-02-22

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Head2Toe: Utilizing Intermediate Representations for Better OOD Generalization
**Paper**: [goo.gle/h2t-paper](https://goo.gle/h2t-paper)
**Video**: [goo.gle/h2t-video](https://goo.gle/h2t-video)

<img src="https://github.com/google-research/head2toe/blob/main/imgs/h2t.png" alt="Head2Toe " width="80%" align="middle">

Code for reproducing our results in the Head2Toe paper.

## Setup
First clone this repo.
```bash
git clone https://github.com/google-research/head2toe.git
cd head2toe
```

We need to download the pre-trained ImageNet checkpoints. If you use the code
below it will move the checkpoints under the correct folder. If you use a
different name you need to update paths in `head2toe/configs_eval/finetune.py`.
```bash
mkdir checkpoints
cd checkpoints
wget -c https://storage.googleapis.com/gresearch/head2toe/imagenetr50.tar.gz
wget -c https://storage.googleapis.com/gresearch/head2toe/imagenetvitB16.tar.gz
tar -xvf imagenetr50.tar.gz
tar -xvf imagenetvitB16.tar.gz
rm *.tar.gz
cd ../
```

Let's run some tests. The following script creates a virtual environment and
installs the necessary libraries. Finally, it runs a few tests.
```bash
bash run.sh
```

We need to activate the virtual environment before running an experiment. With
that, we are ready to run some trivial Caltech101 experiments.
```bash
source env/bin/activate
export PYTHONPATH=$PYTHONPATH:$PWD

python head2toe/evaluate.py \
--config=head2toe/configs_eval/finetune.py:imagenetr50 \
--config.eval_mode='test' --config.dataset='data.caltech101'
```

Note that running evaluation for each task requires downloading and
preparing multiple datasets, which can take up-to a day. Please check out
https://github.com/google-research/task_adaptation for more details on
installing the datasets.

## Running Head2Toe
Our results presented in Table-1 of our paper can be reproduced by running the
following command for Caltech-101 task. This takes 15-10mins on a single V100
gpu.
```bash
python head2toe/evaluate.py \
--config=head2toe/configs_eval/finetune_h2t.py:imagenetr50 \
--config.dataset='data.caltech101' \
--config.eval_mode='test' --config.learning.cached_eval=False \
--config.backbone.additional_features_target_size=8192 \
--config.learning.feature_selection.keep_fraction=0.01 \
--config.learning.feature_selection.learning_config_overwrite.group_lrp_regularizer_coef=0.00001 \
--config.learning.learning_rate=0.01 --config.learning.training_steps=5000 \
--config.learning.log_freq=1000
```
Hyper-parameters used for different tasks can be found in the appendix. Here is
the command for dSprites-Orientation task.
```bash
python head2toe/evaluate.py \
--config=head2toe/configs_eval/finetune_h2t.py:imagenetr50 \
--config.dataset='data.dsprites(predicted_attribute="label_orientation",num_classes=16)' \
--config.eval_mode='test' --config.learning.cached_eval=False \
--config.backbone.additional_features_target_size=512 \
--config.learning.feature_selection.keep_fraction=0.2 \
--config.learning.feature_selection.learning_config_overwrite.group_lrp_regularizer_coef=0.00001 \
--config.learning.learning_rate=0.01 --config.learning.training_steps=500 \
--config.learning.log_freq=1000
```

## Running other baselines.
- **Regularization Baselines**: Use `finetune_h2t.py` config together with
`l1_regularizer`, `l2_regularizer` or `group_lrp_regularizer_coef` flags.
- **Linear**: Use `finetune.py` config.

Set `config.learning.finetune_backbones` to true for enabling the finetuning of
the backbone for any experiment. If you like to run any other experiments or
if you have questions, feel free to create a new issue.

## Citation
```
@InProceedings{evci22h2t,
  title = 	 {{H}ead2{T}oe: Utilizing Intermediate Representations for Better Transfer Learning},
  author =       {Evci, Utku and Dumoulin, Vincent and Larochelle, Hugo and Mozer, Michael C},
  booktitle = 	 {Proceedings of the 39th International Conference on Machine Learning},
  pages = 	 {6009--6033},
  year = 	 {2022},
  editor = 	 {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan},
  volume = 	 {162},
  series = 	 {Proceedings of Machine Learning Research},
  month = 	 {17--23 Jul},
  publisher =    {PMLR},
  pdf = 	 {https://proceedings.mlr.press/v162/evci22a/evci22a.pdf},
  url = 	 {https://proceedings.mlr.press/v162/evci22a.html},
}

```

## Disclaimer
This is not an officially supported Google product.