autoregressive-diffusion-pytorch 0.0.1__tar.gz

autoregressive_diffusion_pytorch-0.0.1/.github/workflows/python-publish.yml

@@ -0,0 +1,36 @@
+# This workflow will upload a Python Package using Twine when a release is created
+# For more information see: https://help.github.com/en/actions/language-and-framework-guides/using-python-with-github-actions#publishing-to-package-registries
+
+# This workflow uses actions that are not certified by GitHub.
+# They are provided by a third-party and are governed by
+# separate terms of service, privacy policy, and support
+# documentation.
+
+name: Upload Python Package
+
+on:
+  release:
+    types: [published]
+
+jobs:
+  deploy:
+
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v2
+    - name: Set up Python
+      uses: actions/setup-python@v2
+      with:
+        python-version: '3.x'
+    - name: Install dependencies
+      run: |
+        python -m pip install --upgrade pip
+        pip install build
+    - name: Build package
+      run: python -m build
+    - name: Publish package
+      uses: pypa/gh-action-pypi-publish@27b31702a0e7fc50959f5ad993c78deac1bdfc29
+      with:
+        user: __token__
+        password: ${{ secrets.PYPI_API_TOKEN }}

autoregressive_diffusion_pytorch-0.0.1/.gitignore

@@ -0,0 +1,162 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#pdm.lock
+#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
+#   in version control.
+#   https://pdm.fming.dev/latest/usage/project/#working-with-version-control
+.pdm.toml
+.pdm-python
+.pdm-build/
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/

autoregressive_diffusion_pytorch-0.0.1/LICENSE

@@ -0,0 +1,21 @@
+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.

autoregressive_diffusion_pytorch-0.0.1/PKG-INFO

@@ -0,0 +1,94 @@
+Metadata-Version: 2.3
+Name: autoregressive-diffusion-pytorch
+Version: 0.0.1
+Summary: Autoregressive Diffusion - Pytorch
+Project-URL: Homepage, https://pypi.org/project/autoregressive-diffusion-pytorch/
+Project-URL: Repository, https://github.com/lucidrains/autoregressive-diffusion-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,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: torch>=2.0
+Requires-Dist: tqdm
+Requires-Dist: x-transformers>=1.31.14
+Provides-Extra: examples
+Requires-Dist: numpy; extra == 'examples'
+Requires-Dist: tqdm; extra == 'examples'
+Description-Content-Type: text/markdown
+
+<img src="./ar-diffusion.png" width="400px"></img>
+
+## Autoregressive Diffusion - Pytorch (wip)
+
+Implementation of the architecture behind <a href="https://arxiv.org/abs/2406.11838">Autoregressive Image Generation without Vector Quantization</a> in Pytorch
+
+You can discuss the paper temporarily [here](https://discord.com/invite/9myQVTbN)
+
+## Install
+
+```bash
+$ pip install autoregressive-diffusion-pytorch
+```
+
+## Usage
+
+```python
+import torch
+from autoregressive_diffusion_pytorch import AutoregressiveDiffusion
+
+model = AutoregressiveDiffusion(
+    dim = 512,
+    max_seq_len = 32
+)
+
+seq = torch.randn(3, 32, 512)
+
+loss = model(seq)
+loss.backward()
+
+sampled = model.sample(batch_size = 3)
+
+assert sampled.shape == seq.shape
+
+```
+
+## Citations
+
+```bibtex
+@article{Li2024AutoregressiveIG,
+    title   = {Autoregressive Image Generation without Vector Quantization},
+    author  = {Tianhong Li and Yonglong Tian and He Li and Mingyang Deng and Kaiming He},
+    journal = {ArXiv},
+    year    = {2024},
+    volume  = {abs/2406.11838},
+    url     = {https://api.semanticscholar.org/CorpusID:270560593}
+}
+```

autoregressive_diffusion_pytorch-0.0.1/README.md

@@ -0,0 +1,48 @@
+<img src="./ar-diffusion.png" width="400px"></img>
+
+## Autoregressive Diffusion - Pytorch (wip)
+
+Implementation of the architecture behind <a href="https://arxiv.org/abs/2406.11838">Autoregressive Image Generation without Vector Quantization</a> in Pytorch
+
+You can discuss the paper temporarily [here](https://discord.com/invite/9myQVTbN)
+
+## Install
+
+```bash
+$ pip install autoregressive-diffusion-pytorch
+```
+
+## Usage
+
+```python
+import torch
+from autoregressive_diffusion_pytorch import AutoregressiveDiffusion
+
+model = AutoregressiveDiffusion(
+    dim = 512,
+    max_seq_len = 32
+)
+
+seq = torch.randn(3, 32, 512)
+
+loss = model(seq)
+loss.backward()
+
+sampled = model.sample(batch_size = 3)
+
+assert sampled.shape == seq.shape
+
+```
+
+## Citations
+
+```bibtex
+@article{Li2024AutoregressiveIG,
+    title   = {Autoregressive Image Generation without Vector Quantization},
+    author  = {Tianhong Li and Yonglong Tian and He Li and Mingyang Deng and Kaiming He},
+    journal = {ArXiv},
+    year    = {2024},
+    volume  = {abs/2406.11838},
+    url     = {https://api.semanticscholar.org/CorpusID:270560593}
+}
+```

autoregressive_diffusion_pytorch-0.0.1/autoregressive_diffusion_pytorch/__init__.py

@@ -0,0 +1,4 @@
+from autoregressive_diffusion_pytorch.autoregressive_diffusion import (
+    AutoregressiveDiffusion,
+    MLP
+)

autoregressive_diffusion_pytorch-0.0.1/autoregressive_diffusion_pytorch/autoregressive_diffusion.py

@@ -0,0 +1,474 @@
+from typing import Literal
+from functools import partial
+
+import torch
+from torch import nn, pi
+from torch.special import expm1
+import torch.nn.functional as F
+from torch.nn import Module, ModuleList
+
+import einx
+from einops import rearrange, repeat, reduce, pack, unpack
+from einops.layers.torch import Rearrange
+
+from tqdm import tqdm
+
+from x_transformers import (
+    ContinuousTransformerWrapper,
+    Decoder
+)
+
+# helpers
+
+def exists(v):
+    return v is not None
+
+def default(v, d):
+    return v if exists(v) else d
+
+def divisible_by(num, den):
+    return (num % den) == 0
+
+# tensor helpers
+
+def log(t, eps = 1e-20):
+    return torch.log(t.clamp(min = eps))
+
+def safe_div(num, den, eps = 1e-5):
+    return num / den.clamp(min = eps)
+
+def right_pad_dims_to(x, t):
+    padding_dims = x.ndim - t.ndim
+
+    if padding_dims <= 0:
+        return t
+
+    return t.view(*t.shape, *((1,) * padding_dims))
+
+def pack_one(t, pattern):
+    packed, ps = pack([t], pattern)
+
+    def unpack_one(to_unpack, unpack_pattern = None):
+        unpacked, = unpack(to_unpack, ps, default(unpack_pattern, pattern))
+        return unpacked
+
+    return packed, unpack_one
+
+# sinusoidal embedding
+
+class AdaptiveLayerNorm(Module):
+    def __init__(
+        self,
+        dim,
+        dim_condition = None
+    ):
+        super().__init__()
+        dim_condition = default(dim_condition, dim)
+
+        self.ln = nn.LayerNorm(dim, elementwise_affine = False)
+        self.to_gamma = nn.Linear(dim_condition, dim, bias = False)
+        nn.init.zeros_(self.to_gamma.weight)
+
+    def forward(self, x, *, condition):
+        normed = self.ln(x)
+        gamma = self.to_gamma(condition)
+        return normed * (gamma + 1.)
+
+class LearnedSinusoidalPosEmb(Module):
+    def __init__(self, dim):
+        super().__init__()
+        assert divisible_by(dim, 2)
+        half_dim = dim // 2
+        self.weights = nn.Parameter(torch.randn(half_dim))
+
+    def forward(self, x):
+        x = rearrange(x, 'b -> b 1')
+        freqs = x * rearrange(self.weights, 'd -> 1 d') * 2 * pi
+        fouriered = torch.cat((freqs.sin(), freqs.cos()), dim = -1)
+        fouriered = torch.cat((x, fouriered), dim = -1)
+        return fouriered
+
+# simple mlp
+
+class MLP(Module):
+    def __init__(
+        self,
+        dim,
+        dim_cond,
+        depth = 3,
+        width = 1024,
+        dropout = 0.
+    ):
+        super().__init__()
+        layers = ModuleList([])
+
+        self.to_time_emb = nn.Sequential(
+            LearnedSinusoidalPosEmb(dim_cond),
+            nn.Linear(dim_cond + 1, dim_cond),
+            nn.SiLU(),
+            nn.Dropout(dropout),
+            nn.Linear(dim_cond, dim_cond)
+        )
+
+        for _ in range(depth):
+
+            adaptive_layernorm = AdaptiveLayerNorm(
+                dim,
+                dim_condition = dim_cond
+            )
+
+            block = nn.Sequential(
+                nn.Linear(dim, dim),
+                nn.SiLU(),
+                nn.Linear(dim, dim)
+            )
+
+            layers.append(ModuleList([
+                adaptive_layernorm,
+                block
+            ]))
+
+        self.layers = layers
+
+    def forward(
+        self,
+        noised,
+        *,
+        times,
+        cond
+    ):
+        assert noised.ndim == 2
+
+        time_emb = self.to_time_emb(times)
+        cond = time_emb + cond
+
+        denoised = noised
+
+        for adaln, block in self.layers:
+            residual = denoised
+            denoised = adaln(denoised, condition = cond)
+            denoised = block(denoised) + residual
+
+        return denoised
+
+# gaussian diffusion
+
+def simple_linear_schedule(t, clip_min = 1e-9):
+    return (1. - t).clamp(min = clip_min)
+
+def cosine_schedule(t, start = 0, end = 1, tau = 1, clip_min = 1e-9):
+    start, end, tau = map(torch.tensor, (start, end, tau))
+    power = 2 * tau
+    v_start = torch.cos(start * pi / 2) ** power
+    v_end = torch.cos(end * pi / 2) ** power
+    output = torch.cos((t * (end - start) + start) * pi / 2) ** power
+    output = (v_end - output) / (v_end - v_start)
+    return output.clamp(min = clip_min)
+
+def gamma_to_alpha_sigma(gamma, scale = 1):
+    return torch.sqrt(gamma) * scale, torch.sqrt(1 - gamma)
+
+def gamma_to_log_snr(gamma, scale = 1, eps = 1e-5):
+    return log(gamma * (scale ** 2) / (1. - gamma), eps = eps)
+
+class GaussianDiffusion(Module):
+    def __init__(
+        self,
+        model: MLP,
+        *,
+        timesteps = 1000,
+        sampling_timesteps = None,
+        use_ddim = True,
+        noise_schedule: Literal['linear', 'cosine'] = 'cosine',
+        objective: Literal['eps', 'v'] = 'v',
+        schedule_kwargs: dict = dict(),
+        min_snr_loss_weight = True,
+        min_snr_gamma = 5,
+    ):
+        super().__init__()
+        self.model = model
+        self.objective = objective
+
+        if noise_schedule == 'linear':
+            self.gamma_schedule = simple_linear_schedule
+        elif noise_schedule == 'cosine':
+            self.gamma_schedule = cosine_schedule
+        else:
+            raise ValueError(f'invalid noise schedule {noise_schedule}')
+
+        # gamma schedules
+
+        self.gamma_schedule = partial(self.gamma_schedule, **schedule_kwargs)
+
+        self.timesteps = timesteps
+        self.sampling_timesteps = default(sampling_timesteps, timesteps)
+
+        self.use_ddim = use_ddim
+
+        # min snr loss weight
+
+        self.min_snr_loss_weight = min_snr_loss_weight
+        self.min_snr_gamma = min_snr_gamma
+
+    @property
+    def device(self):
+        return next(self.model.parameters()).device
+
+    def get_sampling_timesteps(self, batch, *, device):
+        times = torch.linspace(1., 0., self.sampling_timesteps + 1, device = device)
+        times = repeat(times, 't -> b t', b = batch)
+        times = torch.stack((times[:, :-1], times[:, 1:]), dim = 0)
+        times = times.unbind(dim = -1)
+        return times
+
+    @torch.no_grad()
+    def ddpm_sample(self, cond):
+        batch, device = cond.shape[0], self.device
+
+        time_pairs = self.get_sampling_timesteps(batch, device = device)
+
+        seq = torch.randn(cond.shape, device = device)
+
+        for time, time_next in tqdm(time_pairs, desc = 'sampling loop time step', total = self.timesteps, leave = False):
+
+            # get predicted x0
+
+            model_output = self.model(seq, times = time, cond = cond)
+
+            # get log(snr)
+
+            gamma = self.gamma_schedule(time)
+            gamma_next = self.gamma_schedule(time_next)
+
+            gamma, gamma_next = map(partial(right_pad_dims_to, seq), (gamma, gamma_next))
+
+            # get alpha sigma of time and next time
+
+            alpha, sigma = gamma_to_alpha_sigma(gamma)
+            alpha_next, sigma_next = gamma_to_alpha_sigma(gamma_next)
+
+            # calculate x0 and noise
+
+            if self.objective == 'eps':
+                x_start = safe_div(seq - sigma * model_output, alpha)
+
+            elif self.objective == 'v':
+                x_start = alpha * seq - sigma * model_output
+
+            # clip x0
+
+            x_start.clamp_(-1., 1.)
+
+            # derive posterior mean and variance
+
+            log_snr, log_snr_next = map(gamma_to_log_snr, (gamma, gamma_next))
+
+            c = -expm1(log_snr - log_snr_next)
+
+            mean = alpha_next * (seq * (1 - c) / alpha + c * x_start)
+            variance = (sigma_next ** 2) * c
+            log_variance = log(variance)
+
+            # get noise
+
+            noise = einx.where(
+                'b, b d, -> b d',
+                time_next > 0,
+                torch.randn_like(seq),
+                0.
+            )
+
+            seq = mean + (0.5 * log_variance).exp() * noise
+
+        print(seq.shape)
+        return seq
+
+    @torch.no_grad()
+    def ddim_sample(self, cond):
+        batch, device = cond.shape[0], self.device
+
+        time_pairs = self.get_sampling_timesteps(batch, device = device)
+
+        seq = torch.randn(cond.shape, device = device)
+
+        for times, times_next in tqdm(time_pairs, desc = 'sampling loop time step', leave = False):
+
+            # get times and noise levels
+
+            gamma = self.gamma_schedule(times)
+            gamma_next = self.gamma_schedule(times_next)
+
+            padded_gamma, padded_gamma_next = map(partial(right_pad_dims_to, seq), (gamma, gamma_next))
+
+            alpha, sigma = gamma_to_alpha_sigma(padded_gamma)
+            alpha_next, sigma_next = gamma_to_alpha_sigma(padded_gamma_next)
+
+            # predict x0
+
+            model_output = self.model(seq, times = times, cond = cond)
+
+            # calculate x0 and noise
+
+            if self.objective == 'eps':
+                x_start = safe_div(seq - sigma * model_output, alpha)
+
+            elif self.objective == 'v':
+                x_start = alpha * seq - sigma * model_output
+
+            # clip x0
+
+            x_start.clamp_(-1., 1.)
+
+            # get predicted noise
+
+            pred_noise = safe_div(seq - alpha * x_start, sigma)
+
+            # calculate x next
+
+            seq = x_start * alpha_next + pred_noise * sigma_next
+
+        return seq
+
+    @torch.no_grad()
+    def sample(self, *, cond, batch_size = 16):
+        sample_fn = self.ddpm_sample if not self.use_ddim else self.ddim_sample
+        return sample_fn(cond)
+
+    def forward(self, seq, *args, cond, **kwargs):
+
+        batch, device = seq.shape[0], seq.device
+
+        # sample random times
+
+        times = torch.rand((batch,), device = device)
+
+        # noise sample
+
+        noise = torch.randn_like(seq)
+
+        gamma = self.gamma_schedule(times)
+        padded_gamma = right_pad_dims_to(seq, gamma)
+        alpha, sigma =  gamma_to_alpha_sigma(padded_gamma)
+
+        noised_seq = alpha * seq + sigma * noise
+
+        # predict and take gradient step
+
+        pred = self.model(noised_seq, times = times, cond = cond)
+
+        if self.objective == 'eps':
+            target = noise
+
+        elif self.objective == 'v':
+            target = alpha * noise - sigma * seq
+
+        loss = F.mse_loss(pred, target, reduction = 'none')
+        loss = reduce(loss, 'b ... -> b', 'mean')
+
+        # min snr loss weight
+
+        snr = (alpha * alpha) / (sigma * sigma)
+        maybe_clipped_snr = snr.clone()
+
+        if self.min_snr_loss_weight:
+            maybe_clipped_snr.clamp_(max = self.min_snr_gamma)
+
+        if self.objective == 'eps':
+            loss_weight = maybe_clipped_snr / snr
+
+        elif self.objective == 'v':
+            loss_weight = maybe_clipped_snr / (snr + 1)
+
+        return (loss * loss_weight).mean()
+
+# main model, a decoder with continuous wrapper + small denoising mlp
+
+class AutoregressiveDiffusion(Module):
+    def __init__(
+        self,
+        dim,
+        *,
+        max_seq_len,
+        depth = 8,
+        dim_head = 64,
+        heads = 8,
+        mlp_depth = 3,
+        mlp_width = 1024,
+        decoder_kwargs: dict = dict(),
+        mlp_kwargs: dict = dict(),
+        diffusion_kwargs: dict = dict(
+            timesteps = 1000,
+            sampling_timesteps = 100,
+            use_ddim = False
+        )
+    ):
+        super().__init__()
+
+        self.start_token = nn.Parameter(torch.zeros(dim))
+        self.max_seq_len = max_seq_len
+        self.abs_pos_emb = nn.Embedding(max_seq_len, dim)
+
+        self.transformer = Decoder(
+            dim = dim,
+            depth = depth,
+            heads = heads,
+            attn_dim_head = dim_head,
+            **decoder_kwargs
+        )
+
+        self.denoiser = MLP(
+            dim = dim,
+            dim_cond = dim,
+            depth = mlp_depth,
+            width = mlp_width,
+            **mlp_kwargs
+        )
+
+        self.diffusion = GaussianDiffusion(
+            self.denoiser,
+            **diffusion_kwargs
+        )
+
+    def sample(
+        self,
+        batch_size = 1
+    ):
+        out = repeat(self.start_token, 'd -> b 1 d', b = batch_size)
+
+        for _ in tqdm(range(self.max_seq_len), desc = 'tokens'):
+
+            cond = self.transformer(out)
+            last_cond = cond[:, -1]
+
+            denoised_pred = self.diffusion.sample(cond = last_cond)
+
+            denoised_pred = rearrange(denoised_pred, 'b d -> b 1 d')
+            out = torch.cat((out, denoised_pred), dim = 1)
+
+        return out[:, 1:]
+
+    def forward(
+        self,
+        seq
+    ):
+        b = seq.shape[0]
+
+        # append start tokens
+
+        start_token = repeat(self.start_token, 'd -> b 1 d', b = b)
+        seq = torch.cat((start_token, seq), dim = 1)
+
+        # break into seq and the continuous targets to be predicted
+
+        seq, target = seq[:, :-1], seq[:, 1:]
+
+        cond = self.transformer(seq)
+
+        # pack batch and sequence dimensions, so to train each token with different noise levels
+
+        target, _ = pack_one(target, '* d')
+        cond, _ = pack_one(cond, '* d')
+
+        diffusion_loss = self.diffusion(target, cond = cond)
+
+        return diffusion_loss

autoregressive_diffusion_pytorch-0.0.1/pyproject.toml

@@ -0,0 +1,48 @@
+[project]
+name = "autoregressive-diffusion-pytorch"
+version = "0.0.1"
+description = "Autoregressive Diffusion - Pytorch"
+authors = [
+    { name = "Phil Wang", email = "lucidrains@gmail.com" }
+]
+readme = "README.md"
+requires-python = ">= 3.8"
+license = { file = "LICENSE" }
+keywords = [
+    'artificial intelligence',
+    'deep learning',
+    'transformers',
+    'denoising diffusion',
+  ]
+classifiers=[
+    'Development Status :: 4 - Beta',
+    'Intended Audience :: Developers',
+    'Topic :: Scientific/Engineering :: Artificial Intelligence',
+    'License :: OSI Approved :: MIT License',
+    'Programming Language :: Python :: 3.8',
+  ]
+
+dependencies = [
+    'einx>=0.3.0',
+    'einops>=0.8.0',
+    'x-transformers>=1.31.14',
+    'torch>=2.0',
+    'tqdm'
+]
+
+[project.urls]
+Homepage = "https://pypi.org/project/autoregressive-diffusion-pytorch/"
+Repository = "https://github.com/lucidrains/autoregressive-diffusion-pytorch"
+
+[project.optional-dependencies]
+examples = ["tqdm", "numpy"]
+
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+[tool.hatch.metadata]
+allow-direct-references = true
+
+[tool.hatch.build.targets.wheel]
+packages = ["autoregressive-diffusion-pytorch"]