| Name | spline-based-transformer JSON |
| Version |
0.0.14
JSON |
| download |
| home_page | None |
| Summary | Spline Based Transformer |
| upload_time | 2024-11-09 00:35:50 |
| maintainer | None |
| docs_url | None |
| author | None |
| requires_python | >=3.8 |
| license | MIT License Copyright (c) 2024 Phil Wang Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. |
| keywords |
artificial intelligence
deep learning
transformers
attention mechanism
b-spline
latent trajectories
|
| VCS |
 |
| bugtrack_url |
|
| requirements |
No requirements were recorded.
|
| Travis-CI |
No Travis.
|
| coveralls test coverage |
No coveralls.
|
<img src="./spline-based-transformer.png" width="400px"></img>
## Spline-Based Transformer
Implementation of the proposed <a href="https://www.youtube.com/watch?v=AzolLlIbKhg">Spline-Based Transformer</a> ([paper](https://la.disneyresearch.com/wp-content/uploads/SBT.pdf)) from Disney Research
This is basically a transformer based autoencoder, but they cleverly use a set of latent tokens, where that set of tokens are the (high dimensional) control points for a spline.
## Install
```bash
$ pip install spline-based-transformer
```
## Usage
```python
import torch
from spline_based_transformer import SplineBasedTransformer
model = SplineBasedTransformer(
dim = 512,
enc_depth = 6,
dec_depth = 6
)
data = torch.randn(1, 1024, 512)
loss = model(data, return_loss = True)
loss.backward()
# after much training
recon, control_points = model(data, return_latents = True)
assert data.shape == recon.shape
# mess with the control points, which should preserve continuity better
control_points += 1
controlled_recon = model.decode_from_latents(control_points, num_times = 1024)
assert controlled_recon.shape == data.shape
```
For an example of an image autoencoder
```python
import torch
from spline_based_transformer import (
SplineBasedTransformer,
ImageAutoencoderWrapper
)
model = ImageAutoencoderWrapper(
image_size = 256,
patch_size = 32,
spline_transformer = SplineBasedTransformer(
dim = 512,
enc_depth = 6,
dec_depth = 6
)
)
images = torch.randn(2, 3, 256, 256)
loss = model(images, return_loss = True)
loss.backward()
# after much training
recon_images, control_points = model(images, return_latents = True)
assert images.shape == recon_images.shape
# changing the control points
control_points += 1
controlled_recon_images = model.decode_from_latents(control_points)
assert controlled_recon_images.shape == images.shape
```
## Citations
```bibtex
@misc{Chandran2024,
author = {Prashanth Chandran, Agon Serifi, Markus Gross, Moritz Bächer},
url = {https://la.disneyresearch.com/publication/spline-based-transformers/}
}
```
Raw data
{
"_id": null,
"home_page": null,
"name": "spline-based-transformer",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.8",
"maintainer_email": null,
"keywords": "artificial intelligence, deep learning, transformers, attention mechanism, b-spline, latent trajectories",
"author": null,
"author_email": "Phil Wang <lucidrains@gmail.com>",
"download_url": "https://files.pythonhosted.org/packages/43/76/487dfa448da9ad2a987abf4e7823ca328cbca7cc2e20154029f6e5f48d9d/spline_based_transformer-0.0.14.tar.gz",
"platform": null,
"description": "<img src=\"./spline-based-transformer.png\" width=\"400px\"></img>\n\n## Spline-Based Transformer\n\nImplementation of the proposed <a href=\"https://www.youtube.com/watch?v=AzolLlIbKhg\">Spline-Based Transformer</a> ([paper](https://la.disneyresearch.com/wp-content/uploads/SBT.pdf)) from Disney Research\n\nThis is basically a transformer based autoencoder, but they cleverly use a set of latent tokens, where that set of tokens are the (high dimensional) control points for a spline.\n\n## Install\n\n```bash\n$ pip install spline-based-transformer\n```\n\n## Usage\n\n```python\nimport torch\nfrom spline_based_transformer import SplineBasedTransformer\n\nmodel = SplineBasedTransformer(\n dim = 512,\n enc_depth = 6,\n dec_depth = 6\n)\n\ndata = torch.randn(1, 1024, 512)\n\nloss = model(data, return_loss = True)\nloss.backward()\n\n# after much training\n\nrecon, control_points = model(data, return_latents = True)\nassert data.shape == recon.shape\n\n# mess with the control points, which should preserve continuity better\n\ncontrol_points += 1\n\ncontrolled_recon = model.decode_from_latents(control_points, num_times = 1024)\nassert controlled_recon.shape == data.shape\n```\n\nFor an example of an image autoencoder\n\n```python\nimport torch\n\nfrom spline_based_transformer import (\n SplineBasedTransformer,\n ImageAutoencoderWrapper\n)\n\nmodel = ImageAutoencoderWrapper(\n image_size = 256,\n patch_size = 32,\n spline_transformer = SplineBasedTransformer(\n dim = 512,\n enc_depth = 6,\n dec_depth = 6\n )\n)\n\nimages = torch.randn(2, 3, 256, 256)\n\nloss = model(images, return_loss = True)\nloss.backward()\n\n# after much training\n\nrecon_images, control_points = model(images, return_latents = True)\nassert images.shape == recon_images.shape\n\n# changing the control points\n\ncontrol_points += 1\n\ncontrolled_recon_images = model.decode_from_latents(control_points)\n\nassert controlled_recon_images.shape == images.shape\n```\n\n## Citations\n\n```bibtex\n@misc{Chandran2024,\n author = {Prashanth Chandran, Agon Serifi, Markus Gross, Moritz B\u00e4cher},\n url = {https://la.disneyresearch.com/publication/spline-based-transformers/}\n}\n```\n",
"bugtrack_url": null,
"license": "MIT License Copyright (c) 2024 Phil Wang Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the \"Software\"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. ",
"summary": "Spline Based Transformer",
"version": "0.0.14",
"project_urls": {
"Homepage": "https://pypi.org/project/spline-based-transformer",
"Repository": "https://github.com/lucidrains/spline-based-transformer"
},
"split_keywords": [
"artificial intelligence",
" deep learning",
" transformers",
" attention mechanism",
" b-spline",
" latent trajectories"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "ba1aedd7f34fa04836c0c27a8594092d26a7a0088f67add25a3e39d09d1aac07",
"md5": "bf9f1fb86a83b860f874bfa91eeb7193",
"sha256": "be0f28c8254c07cd1b47d250b973c8b868c8584485eb8fbeebb9095f2182bf69"
},
"downloads": -1,
"filename": "spline_based_transformer-0.0.14-py3-none-any.whl",
"has_sig": false,
"md5_digest": "bf9f1fb86a83b860f874bfa91eeb7193",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.8",
"size": 6816,
"upload_time": "2024-11-09T00:35:48",
"upload_time_iso_8601": "2024-11-09T00:35:48.746091Z",
"url": "https://files.pythonhosted.org/packages/ba/1a/edd7f34fa04836c0c27a8594092d26a7a0088f67add25a3e39d09d1aac07/spline_based_transformer-0.0.14-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "4376487dfa448da9ad2a987abf4e7823ca328cbca7cc2e20154029f6e5f48d9d",
"md5": "4b054206355453a4adb4868860d09733",
"sha256": "f023c41004b47c54639ca61288ba7fef979e19f9b076f06e618b22a6aaca86a2"
},
"downloads": -1,
"filename": "spline_based_transformer-0.0.14.tar.gz",
"has_sig": false,
"md5_digest": "4b054206355453a4adb4868860d09733",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.8",
"size": 5620,
"upload_time": "2024-11-09T00:35:50",
"upload_time_iso_8601": "2024-11-09T00:35:50.257430Z",
"url": "https://files.pythonhosted.org/packages/43/76/487dfa448da9ad2a987abf4e7823ca328cbca7cc2e20154029f6e5f48d9d/spline_based_transformer-0.0.14.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-11-09 00:35:50",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "lucidrains",
"github_project": "spline-based-transformer",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "spline-based-transformer"
}
