Metadata-Version: 2.3
Name: transfusion-pytorch
Version: 0.0.30
Summary: Transfusion in Pytorch
Project-URL: Homepage, https://pypi.org/project/transfusion-pytorch/
Project-URL: Repository, https://github.com/lucidrains/transfusion-pytorch
Author-email: Phil Wang <lucidrains@gmail.com>
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.
License-File: LICENSE
Keywords: artificial intelligence,attention mechanism,deep learning,denoising diffusion,transformers
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3.8
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.8
Requires-Dist: einops>=0.8.0
Requires-Dist: einx>=0.3.0
Requires-Dist: jaxtyping
Requires-Dist: rotary-embedding-torch>=0.8.3
Requires-Dist: torch>=2.0
Requires-Dist: torchdiffeq
Requires-Dist: tqdm
Provides-Extra: examples
Provides-Extra: test
Requires-Dist: pytest; extra == 'test'
Description-Content-Type: text/markdown

<img src="./transfusion.png" width="400px"></img>

## Transfusion - Pytorch (wip)

Pytorch implementation of [Transfusion](https://www.arxiv.org/abs/2408.11039), "Predict the Next Token and Diffuse Images with One Multi-Modal Model", from MetaAI.

In this repo, we will substitute diffusion with flow matching given the success of Flux from Black Forest Labs (but will keep the original paper title given Transflow does not have the same ring). This repository will also attempt to extend to any number of modalities.

## Install

```bash
$ pip install transfusion-pytorch
```

## Usage

One modality, say images

```python
from torch import randint, randn
from transfusion_pytorch import Transfusion

model = Transfusion(
    num_text_tokens = 256,
    dim_latent = 384,
    transformer = dict(
        dim = 512,
        depth = 8
    )
)

text_and_images = [
    [randint(0, 256, (16,)), randn(4, 384), randint(0, 256, (8,)), randn(6, 384)],
    [randint(0, 256, (16,)), randn(7, 384), randint(0, 256, (5,)), randn(2, 384), randint(0, 256, (9,))]
]

loss = model(text_and_images)

loss.backward()

# after much training

one_multimodal_sample = model.sample()
```

Multiple different modalities

```python
from torch import randint, randn
from transfusion_pytorch import Transfusion

model = Transfusion(
    num_text_tokens = 256,
    dim_latent = (384, 192), # specify multiple latent dimensions
    transformer = dict(
        dim = 512,
        depth = 8
    )
)

# then for the Tensors of type float, you can pass a tuple[int, Tensor] and specify the modality index in the first position

text_images_and_audio = [
    [randint(0, 256, (16,)), (0, randn(4, 384)), randint(0, 256, (8,)), (1, randn(6, 192))],
    [randint(0, 256, (16,)), randn(7, 384), randint(0, 256, (5,)), (1, randn(2, 192)), randint(0, 256, (9,))]
]

loss = model(text_images_and_audio)

loss.backward()

# after much training

one_multimodal_sample = model.sample()
```

## Citations

```bibtex
@inproceedings{Zhou2024TransfusionPT,
    title  = {Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Model},
    author = {Chunting Zhou and Lili Yu and Arun Babu and Kushal Tirumala and Michihiro Yasunaga and Leonid Shamis and Jacob Kahn and Xuezhe Ma and Luke Zettlemoyer and Omer Levy},
    year   = {2024},
    url    = {https://api.semanticscholar.org/CorpusID:271909855}
}
```

```bibtex
@misc{Rubin2024,
    author  = {Ohad Rubin},
    url     = {https://medium.com/@ohadrubin/exploring-weight-decay-in-layer-normalization-challenges-and-a-reparameterization-solution-ad4d12c24950}
}
```

```bibtex
@article{Nguyen2024MinPS,
    title   = {Min P Sampling: Balancing Creativity and Coherence at High Temperature},
    author  = {Minh Nguyen and Andrew Baker and Andreas Kirsch and Clement Neo},
    journal = {ArXiv},
    year    = {2024},
    volume  = {abs/2407.01082},
    url     = {https://api.semanticscholar.org/CorpusID:270870613}
}
```
