├── ldm ├── data │ ├── __init__.py │ ├── base.py │ ├── lsun.py │ └── imagenet.py ├── models │ ├── diffusion │ │ ├── __init__.py │ │ ├── __pycache__ │ │ │ ├── ddim.cpython-38.pyc │ │ │ ├── ddpm.cpython-38.pyc │ │ │ ├── plms.cpython-38.pyc │ │ │ └── __init__.cpython-38.pyc │ │ ├── classifier.py │ │ ├── _ddim.py │ │ └── ddim.py │ ├── __pycache__ │ │ └── autoencoder.cpython-38.pyc │ └── autoencoder.py ├── modules │ ├── encoders │ │ ├── __init__.py │ │ ├── __pycache__ │ │ │ ├── __init__.cpython-38.pyc │ │ │ └── modules.cpython-38.pyc │ │ └── modules.py │ ├── distributions │ │ ├── __init__.py │ │ ├── __pycache__ │ │ │ ├── __init__.cpython-38.pyc │ │ │ └── distributions.cpython-38.pyc │ │ └── distributions.py │ ├── diffusionmodules │ │ ├── __init__.py │ │ ├── __pycache__ │ │ │ ├── model.cpython-38.pyc │ │ │ ├── util.cpython-38.pyc │ │ │ ├── __init__.cpython-38.pyc │ │ │ └── openaimodel.cpython-38.pyc │ │ └── util.py │ ├── losses │ │ ├── __init__.py │ │ ├── contperceptual.py │ │ └── vqperceptual.py │ ├── __pycache__ │ │ ├── ema.cpython-38.pyc │ │ ├── attention.cpython-38.pyc │ │ └── x_transformer.cpython-38.pyc │ ├── image_degradation │ │ ├── utils │ │ │ └── test.png │ │ └── __init__.py │ ├── ema.py │ ├── attention.py │ └── x_transformer.py ├── __pycache__ │ ├── casa.cpython-38.pyc │ └── util.cpython-38.pyc ├── lr_scheduler.py └── util.py ├── teaser.png ├── environment.yaml ├── README.md ├── configs └── stable-diffusion │ └── v1-inference.yaml └── utils.py /ldm/data/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ldm/models/diffusion/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ldm/modules/encoders/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ldm/modules/distributions/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ldm/modules/diffusionmodules/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /teaser.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/teaser.png -------------------------------------------------------------------------------- /ldm/modules/losses/__init__.py: -------------------------------------------------------------------------------- 1 | from ldm.modules.losses.contperceptual import LPIPSWithDiscriminator -------------------------------------------------------------------------------- /ldm/__pycache__/casa.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/__pycache__/casa.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/__pycache__/util.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/__pycache__/util.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/modules/__pycache__/ema.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/modules/__pycache__/ema.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/modules/image_degradation/utils/test.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/modules/image_degradation/utils/test.png -------------------------------------------------------------------------------- /ldm/models/__pycache__/autoencoder.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/models/__pycache__/autoencoder.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/modules/__pycache__/attention.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/modules/__pycache__/attention.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/models/diffusion/__pycache__/ddim.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/models/diffusion/__pycache__/ddim.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/models/diffusion/__pycache__/ddpm.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/models/diffusion/__pycache__/ddpm.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/models/diffusion/__pycache__/plms.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/models/diffusion/__pycache__/plms.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/modules/__pycache__/x_transformer.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/modules/__pycache__/x_transformer.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/models/diffusion/__pycache__/__init__.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/models/diffusion/__pycache__/__init__.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/modules/encoders/__pycache__/__init__.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/modules/encoders/__pycache__/__init__.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/modules/encoders/__pycache__/modules.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/modules/encoders/__pycache__/modules.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/modules/diffusionmodules/__pycache__/model.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/modules/diffusionmodules/__pycache__/model.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/modules/diffusionmodules/__pycache__/util.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/modules/diffusionmodules/__pycache__/util.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/modules/distributions/__pycache__/__init__.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/modules/distributions/__pycache__/__init__.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/modules/diffusionmodules/__pycache__/__init__.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/modules/diffusionmodules/__pycache__/__init__.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/modules/diffusionmodules/__pycache__/openaimodel.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/modules/diffusionmodules/__pycache__/openaimodel.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/modules/distributions/__pycache__/distributions.cpython-38.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UT-Mao/Initial-Noise-Construction/HEAD/ldm/modules/distributions/__pycache__/distributions.cpython-38.pyc -------------------------------------------------------------------------------- /ldm/modules/image_degradation/__init__.py: -------------------------------------------------------------------------------- 1 | from ldm.modules.image_degradation.bsrgan import degradation_bsrgan_variant as degradation_fn_bsr 2 | from ldm.modules.image_degradation.bsrgan_light import degradation_bsrgan_variant as degradation_fn_bsr_light 3 | -------------------------------------------------------------------------------- /environment.yaml: -------------------------------------------------------------------------------- 1 | name: ldm 2 | channels: 3 | - pytorch 4 | - defaults 5 | dependencies: 6 | - python=3.8.5 7 | - pip=20.3 8 | - cudatoolkit=11.0 9 | - pytorch=1.7.0 10 | - torchvision=0.8.1 11 | - numpy=1.19.2 12 | - pip: 13 | - albumentations==0.4.3 14 | - opencv-python==4.1.2.30 15 | - pudb==2019.2 16 | - imageio==2.9.0 17 | - imageio-ffmpeg==0.4.2 18 | - pytorch-lightning==1.6.1 19 | - omegaconf==2.1.1 20 | - test-tube>=0.7.5 21 | - streamlit>=0.73.1 22 | - einops==0.3.0 23 | - torch-fidelity==0.3.0 24 | - transformers==4.19.2 25 | - kornia==0.7.2 26 | - -e git+https://github.com/CompVis/taming-transformers.git@master#egg=taming-transformers 27 | - -e git+https://github.com/openai/CLIP.git@main#egg=clip 28 | - -e . -------------------------------------------------------------------------------- /ldm/data/base.py: -------------------------------------------------------------------------------- 1 | from abc import abstractmethod 2 | from torch.utils.data import Dataset, ConcatDataset, ChainDataset, IterableDataset 3 | 4 | 5 | class Txt2ImgIterableBaseDataset(IterableDataset): 6 | ''' 7 | Define an interface to make the IterableDatasets for text2img data chainable 8 | ''' 9 | def __init__(self, num_records=0, valid_ids=None, size=256): 10 | super().__init__() 11 | self.num_records = num_records 12 | self.valid_ids = valid_ids 13 | self.sample_ids = valid_ids 14 | self.size = size 15 | 16 | print(f'{self.__class__.__name__} dataset contains {self.__len__()} examples.') 17 | 18 | def __len__(self): 19 | return self.num_records 20 | 21 | @abstractmethod 22 | def __iter__(self): 23 | pass -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # The Lottery Ticket Hypothesis in Denoising: Towards Semantic-Driven Initialization (ECCV 2024) 2 | 3 | ## [Project Page] [Paper] 4 | 5 | Refer to our previous work for more discussion about initial noise in diffusion! 6 | - Guided Image Synthesis via Initial Image Editing in Diffusion Model (ACM MM 2023) 7 | 8 | ![teaser](./teaser.png) 9 | 10 | 11 | ## Setup 12 | 13 | Our codebase is built on [CompVis/stable-diffusion](https://github.com/CompVis/stable-diffusion) 14 | and has shared dependencies and model architecture. 15 | 16 | ### Creating a Conda Environment 17 | 18 | ``` 19 | conda env create -f environment.yaml 20 | conda activate ldm 21 | ``` 22 | 23 | ### Downloading StableDiffusion Weights 24 | 25 | Download the StableDiffusion weights from the [CompVis organization at Hugging Face](https://huggingface.co/CompVis/stable-diffusion-v-1-4-original) 26 | (download the `sd-v1-4.ckpt` file), and link them: 27 | ``` 28 | mkdir -p models/ldm/stable-diffusion-v1/ 29 | ln -s models/ldm/stable-diffusion-v1/model.ckpt 30 | ``` 31 | ## Hands on 32 | 33 | Play with [hands-on](./hands_on_ECCV.ipynb) to try our approach right away, refer to [utils.py](./utils.py) for the implementation. 34 | 35 | ## Citation 36 | ``` 37 | @article{mao2024theLottery, 38 | title={The Lottery Ticket Hypothesis in Denoising: Towards Semantic-Driven Initialization}, 39 | author={Mao, Jiafeng and Wang, Xueting and Aizawa, Kiyoharu}, 40 | journal={ECCV}, 41 | year={2024} 42 | } 43 | ``` 44 | -------------------------------------------------------------------------------- /configs/stable-diffusion/v1-inference.yaml: -------------------------------------------------------------------------------- 1 | model: 2 | base_learning_rate: 1.0e-04 3 | target: ldm.models.diffusion.ddpm.LatentDiffusion 4 | params: 5 | linear_start: 0.00085 6 | linear_end: 0.0120 7 | num_timesteps_cond: 1 8 | log_every_t: 200 9 | timesteps: 1000 10 | first_stage_key: "jpg" 11 | cond_stage_key: "txt" 12 | image_size: 64 13 | channels: 4 14 | cond_stage_trainable: false # Note: different from the one we trained before 15 | conditioning_key: crossattn 16 | monitor: val/loss_simple_ema 17 | scale_factor: 0.18215 18 | use_ema: False 19 | 20 | scheduler_config: # 10000 warmup steps 21 | target: ldm.lr_scheduler.LambdaLinearScheduler 22 | params: 23 | warm_up_steps: [ 10000 ] 24 | cycle_lengths: [ 10000000000000 ] # incredibly large number to prevent corner cases 25 | f_start: [ 1.e-6 ] 26 | f_max: [ 1. ] 27 | f_min: [ 1. ] 28 | 29 | unet_config: 30 | target: ldm.modules.diffusionmodules.openaimodel.UNetModel 31 | params: 32 | image_size: 32 # unused 33 | in_channels: 4 34 | out_channels: 4 35 | model_channels: 320 36 | attention_resolutions: [ 4, 2, 1 ] 37 | num_res_blocks: 2 38 | channel_mult: [ 1, 2, 4, 4 ] 39 | num_heads: 8 40 | use_spatial_transformer: True 41 | transformer_depth: 1 42 | context_dim: 768 43 | use_checkpoint: True 44 | legacy: False 45 | save_map: True 46 | 47 | first_stage_config: 48 | target: ldm.models.autoencoder.AutoencoderKL 49 | params: 50 | embed_dim: 4 51 | monitor: val/rec_loss 52 | ddconfig: 53 | double_z: true 54 | z_channels: 4 55 | resolution: 256 56 | in_channels: 3 57 | out_ch: 3 58 | ch: 128 59 | ch_mult: 60 | - 1 61 | - 2 62 | - 4 63 | - 4 64 | num_res_blocks: 2 65 | attn_resolutions: [] 66 | dropout: 0.0 67 | lossconfig: 68 | target: torch.nn.Identity 69 | 70 | cond_stage_config: 71 | target: ldm.modules.encoders.modules.FrozenCLIPEmbedder 72 | -------------------------------------------------------------------------------- /ldm/modules/ema.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch import nn 3 | 4 | 5 | class LitEma(nn.Module): 6 | def __init__(self, model, decay=0.9999, use_num_upates=True): 7 | super().__init__() 8 | if decay < 0.0 or decay > 1.0: 9 | raise ValueError('Decay must be between 0 and 1') 10 | 11 | self.m_name2s_name = {} 12 | self.register_buffer('decay', torch.tensor(decay, dtype=torch.float32)) 13 | self.register_buffer('num_updates', torch.tensor(0,dtype=torch.int) if use_num_upates 14 | else torch.tensor(-1,dtype=torch.int)) 15 | 16 | for name, p in model.named_parameters(): 17 | if p.requires_grad: 18 | #remove as '.'-character is not allowed in buffers 19 | s_name = name.replace('.','') 20 | self.m_name2s_name.update({name:s_name}) 21 | self.register_buffer(s_name,p.clone().detach().data) 22 | 23 | self.collected_params = [] 24 | 25 | def forward(self,model): 26 | decay = self.decay 27 | 28 | if self.num_updates >= 0: 29 | self.num_updates += 1 30 | decay = min(self.decay,(1 + self.num_updates) / (10 + self.num_updates)) 31 | 32 | one_minus_decay = 1.0 - decay 33 | 34 | with torch.no_grad(): 35 | m_param = dict(model.named_parameters()) 36 | shadow_params = dict(self.named_buffers()) 37 | 38 | for key in m_param: 39 | if m_param[key].requires_grad: 40 | sname = self.m_name2s_name[key] 41 | shadow_params[sname] = shadow_params[sname].type_as(m_param[key]) 42 | shadow_params[sname].sub_(one_minus_decay * (shadow_params[sname] - m_param[key])) 43 | else: 44 | assert not key in self.m_name2s_name 45 | 46 | def copy_to(self, model): 47 | m_param = dict(model.named_parameters()) 48 | shadow_params = dict(self.named_buffers()) 49 | for key in m_param: 50 | if m_param[key].requires_grad: 51 | m_param[key].data.copy_(shadow_params[self.m_name2s_name[key]].data) 52 | else: 53 | assert not key in self.m_name2s_name 54 | 55 | def store(self, parameters): 56 | """ 57 | Save the current parameters for restoring later. 58 | Args: 59 | parameters: Iterable of `torch.nn.Parameter`; the parameters to be 60 | temporarily stored. 61 | """ 62 | self.collected_params = [param.clone() for param in parameters] 63 | 64 | def restore(self, parameters): 65 | """ 66 | Restore the parameters stored with the `store` method. 67 | Useful to validate the model with EMA parameters without affecting the 68 | original optimization process. Store the parameters before the 69 | `copy_to` method. After validation (or model saving), use this to 70 | restore the former parameters. 71 | Args: 72 | parameters: Iterable of `torch.nn.Parameter`; the parameters to be 73 | updated with the stored parameters. 74 | """ 75 | for c_param, param in zip(self.collected_params, parameters): 76 | param.data.copy_(c_param.data) 77 | -------------------------------------------------------------------------------- /ldm/modules/distributions/distributions.py: -------------------------------------------------------------------------------- 1 | import torch 2 | import numpy as np 3 | 4 | 5 | class AbstractDistribution: 6 | def sample(self): 7 | raise NotImplementedError() 8 | 9 | def mode(self): 10 | raise NotImplementedError() 11 | 12 | 13 | class DiracDistribution(AbstractDistribution): 14 | def __init__(self, value): 15 | self.value = value 16 | 17 | def sample(self): 18 | return self.value 19 | 20 | def mode(self): 21 | return self.value 22 | 23 | 24 | class DiagonalGaussianDistribution(object): 25 | def __init__(self, parameters, deterministic=False): 26 | self.parameters = parameters 27 | self.mean, self.logvar = torch.chunk(parameters, 2, dim=1) 28 | self.logvar = torch.clamp(self.logvar, -30.0, 20.0) 29 | self.deterministic = deterministic 30 | self.std = torch.exp(0.5 * self.logvar) 31 | self.var = torch.exp(self.logvar) 32 | if self.deterministic: 33 | self.var = self.std = torch.zeros_like(self.mean).to(device=self.parameters.device) 34 | 35 | def sample(self): 36 | x = self.mean + self.std * torch.randn(self.mean.shape).to(device=self.parameters.device) 37 | return x 38 | 39 | def kl(self, other=None): 40 | if self.deterministic: 41 | return torch.Tensor([0.]) 42 | else: 43 | if other is None: 44 | return 0.5 * torch.sum(torch.pow(self.mean, 2) 45 | + self.var - 1.0 - self.logvar, 46 | dim=[1, 2, 3]) 47 | else: 48 | return 0.5 * torch.sum( 49 | torch.pow(self.mean - other.mean, 2) / other.var 50 | + self.var / other.var - 1.0 - self.logvar + other.logvar, 51 | dim=[1, 2, 3]) 52 | 53 | def nll(self, sample, dims=[1,2,3]): 54 | if self.deterministic: 55 | return torch.Tensor([0.]) 56 | logtwopi = np.log(2.0 * np.pi) 57 | return 0.5 * torch.sum( 58 | logtwopi + self.logvar + torch.pow(sample - self.mean, 2) / self.var, 59 | dim=dims) 60 | 61 | def mode(self): 62 | return self.mean 63 | 64 | 65 | def normal_kl(mean1, logvar1, mean2, logvar2): 66 | """ 67 | source: https://github.com/openai/guided-diffusion/blob/27c20a8fab9cb472df5d6bdd6c8d11c8f430b924/guided_diffusion/losses.py#L12 68 | Compute the KL divergence between two gaussians. 69 | Shapes are automatically broadcasted, so batches can be compared to 70 | scalars, among other use cases. 71 | """ 72 | tensor = None 73 | for obj in (mean1, logvar1, mean2, logvar2): 74 | if isinstance(obj, torch.Tensor): 75 | tensor = obj 76 | break 77 | assert tensor is not None, "at least one argument must be a Tensor" 78 | 79 | # Force variances to be Tensors. Broadcasting helps convert scalars to 80 | # Tensors, but it does not work for torch.exp(). 81 | logvar1, logvar2 = [ 82 | x if isinstance(x, torch.Tensor) else torch.tensor(x).to(tensor) 83 | for x in (logvar1, logvar2) 84 | ] 85 | 86 | return 0.5 * ( 87 | -1.0 88 | + logvar2 89 | - logvar1 90 | + torch.exp(logvar1 - logvar2) 91 | + ((mean1 - mean2) ** 2) * torch.exp(-logvar2) 92 | ) 93 | -------------------------------------------------------------------------------- /ldm/data/lsun.py: -------------------------------------------------------------------------------- 1 | import os 2 | import numpy as np 3 | import PIL 4 | from PIL import Image 5 | from torch.utils.data import Dataset 6 | from torchvision import transforms 7 | 8 | 9 | class LSUNBase(Dataset): 10 | def __init__(self, 11 | txt_file, 12 | data_root, 13 | size=None, 14 | interpolation="bicubic", 15 | flip_p=0.5 16 | ): 17 | self.data_paths = txt_file 18 | self.data_root = data_root 19 | with open(self.data_paths, "r") as f: 20 | self.image_paths = f.read().splitlines() 21 | self._length = len(self.image_paths) 22 | self.labels = { 23 | "relative_file_path_": [l for l in self.image_paths], 24 | "file_path_": [os.path.join(self.data_root, l) 25 | for l in self.image_paths], 26 | } 27 | 28 | self.size = size 29 | self.interpolation = {"linear": PIL.Image.LINEAR, 30 | "bilinear": PIL.Image.BILINEAR, 31 | "bicubic": PIL.Image.BICUBIC, 32 | "lanczos": PIL.Image.LANCZOS, 33 | }[interpolation] 34 | self.flip = transforms.RandomHorizontalFlip(p=flip_p) 35 | 36 | def __len__(self): 37 | return self._length 38 | 39 | def __getitem__(self, i): 40 | example = dict((k, self.labels[k][i]) for k in self.labels) 41 | image = Image.open(example["file_path_"]) 42 | if not image.mode == "RGB": 43 | image = image.convert("RGB") 44 | 45 | # default to score-sde preprocessing 46 | img = np.array(image).astype(np.uint8) 47 | crop = min(img.shape[0], img.shape[1]) 48 | h, w, = img.shape[0], img.shape[1] 49 | img = img[(h - crop) // 2:(h + crop) // 2, 50 | (w - crop) // 2:(w + crop) // 2] 51 | 52 | image = Image.fromarray(img) 53 | if self.size is not None: 54 | image = image.resize((self.size, self.size), resample=self.interpolation) 55 | 56 | image = self.flip(image) 57 | image = np.array(image).astype(np.uint8) 58 | example["image"] = (image / 127.5 - 1.0).astype(np.float32) 59 | return example 60 | 61 | 62 | class LSUNChurchesTrain(LSUNBase): 63 | def __init__(self, **kwargs): 64 | super().__init__(txt_file="data/lsun/church_outdoor_train.txt", data_root="data/lsun/churches", **kwargs) 65 | 66 | 67 | class LSUNChurchesValidation(LSUNBase): 68 | def __init__(self, flip_p=0., **kwargs): 69 | super().__init__(txt_file="data/lsun/church_outdoor_val.txt", data_root="data/lsun/churches", 70 | flip_p=flip_p, **kwargs) 71 | 72 | 73 | class LSUNBedroomsTrain(LSUNBase): 74 | def __init__(self, **kwargs): 75 | super().__init__(txt_file="data/lsun/bedrooms_train.txt", data_root="data/lsun/bedrooms", **kwargs) 76 | 77 | 78 | class LSUNBedroomsValidation(LSUNBase): 79 | def __init__(self, flip_p=0.0, **kwargs): 80 | super().__init__(txt_file="data/lsun/bedrooms_val.txt", data_root="data/lsun/bedrooms", 81 | flip_p=flip_p, **kwargs) 82 | 83 | 84 | class LSUNCatsTrain(LSUNBase): 85 | def __init__(self, **kwargs): 86 | super().__init__(txt_file="data/lsun/cat_train.txt", data_root="data/lsun/cats", **kwargs) 87 | 88 | 89 | class LSUNCatsValidation(LSUNBase): 90 | def __init__(self, flip_p=0., **kwargs): 91 | super().__init__(txt_file="data/lsun/cat_val.txt", data_root="data/lsun/cats", 92 | flip_p=flip_p, **kwargs) 93 | -------------------------------------------------------------------------------- /ldm/lr_scheduler.py: -------------------------------------------------------------------------------- 1 | import numpy as np 2 | 3 | 4 | class LambdaWarmUpCosineScheduler: 5 | """ 6 | note: use with a base_lr of 1.0 7 | """ 8 | def __init__(self, warm_up_steps, lr_min, lr_max, lr_start, max_decay_steps, verbosity_interval=0): 9 | self.lr_warm_up_steps = warm_up_steps 10 | self.lr_start = lr_start 11 | self.lr_min = lr_min 12 | self.lr_max = lr_max 13 | self.lr_max_decay_steps = max_decay_steps 14 | self.last_lr = 0. 15 | self.verbosity_interval = verbosity_interval 16 | 17 | def schedule(self, n, **kwargs): 18 | if self.verbosity_interval > 0: 19 | if n % self.verbosity_interval == 0: print(f"current step: {n}, recent lr-multiplier: {self.last_lr}") 20 | if n < self.lr_warm_up_steps: 21 | lr = (self.lr_max - self.lr_start) / self.lr_warm_up_steps * n + self.lr_start 22 | self.last_lr = lr 23 | return lr 24 | else: 25 | t = (n - self.lr_warm_up_steps) / (self.lr_max_decay_steps - self.lr_warm_up_steps) 26 | t = min(t, 1.0) 27 | lr = self.lr_min + 0.5 * (self.lr_max - self.lr_min) * ( 28 | 1 + np.cos(t * np.pi)) 29 | self.last_lr = lr 30 | return lr 31 | 32 | def __call__(self, n, **kwargs): 33 | return self.schedule(n,**kwargs) 34 | 35 | 36 | class LambdaWarmUpCosineScheduler2: 37 | """ 38 | supports repeated iterations, configurable via lists 39 | note: use with a base_lr of 1.0. 40 | """ 41 | def __init__(self, warm_up_steps, f_min, f_max, f_start, cycle_lengths, verbosity_interval=0): 42 | assert len(warm_up_steps) == len(f_min) == len(f_max) == len(f_start) == len(cycle_lengths) 43 | self.lr_warm_up_steps = warm_up_steps 44 | self.f_start = f_start 45 | self.f_min = f_min 46 | self.f_max = f_max 47 | self.cycle_lengths = cycle_lengths 48 | self.cum_cycles = np.cumsum([0] + list(self.cycle_lengths)) 49 | self.last_f = 0. 50 | self.verbosity_interval = verbosity_interval 51 | 52 | def find_in_interval(self, n): 53 | interval = 0 54 | for cl in self.cum_cycles[1:]: 55 | if n <= cl: 56 | return interval 57 | interval += 1 58 | 59 | def schedule(self, n, **kwargs): 60 | cycle = self.find_in_interval(n) 61 | n = n - self.cum_cycles[cycle] 62 | if self.verbosity_interval > 0: 63 | if n % self.verbosity_interval == 0: print(f"current step: {n}, recent lr-multiplier: {self.last_f}, " 64 | f"current cycle {cycle}") 65 | if n < self.lr_warm_up_steps[cycle]: 66 | f = (self.f_max[cycle] - self.f_start[cycle]) / self.lr_warm_up_steps[cycle] * n + self.f_start[cycle] 67 | self.last_f = f 68 | return f 69 | else: 70 | t = (n - self.lr_warm_up_steps[cycle]) / (self.cycle_lengths[cycle] - self.lr_warm_up_steps[cycle]) 71 | t = min(t, 1.0) 72 | f = self.f_min[cycle] + 0.5 * (self.f_max[cycle] - self.f_min[cycle]) * ( 73 | 1 + np.cos(t * np.pi)) 74 | self.last_f = f 75 | return f 76 | 77 | def __call__(self, n, **kwargs): 78 | return self.schedule(n, **kwargs) 79 | 80 | 81 | class LambdaLinearScheduler(LambdaWarmUpCosineScheduler2): 82 | 83 | def schedule(self, n, **kwargs): 84 | cycle = self.find_in_interval(n) 85 | n = n - self.cum_cycles[cycle] 86 | if self.verbosity_interval > 0: 87 | if n % self.verbosity_interval == 0: print(f"current step: {n}, recent lr-multiplier: {self.last_f}, " 88 | f"current cycle {cycle}") 89 | 90 | if n < self.lr_warm_up_steps[cycle]: 91 | f = (self.f_max[cycle] - self.f_start[cycle]) / self.lr_warm_up_steps[cycle] * n + self.f_start[cycle] 92 | self.last_f = f 93 | return f 94 | else: 95 | f = self.f_min[cycle] + (self.f_max[cycle] - self.f_min[cycle]) * (self.cycle_lengths[cycle] - n) / (self.cycle_lengths[cycle]) 96 | self.last_f = f 97 | return f 98 | 99 | -------------------------------------------------------------------------------- /ldm/modules/losses/contperceptual.py: -------------------------------------------------------------------------------- 1 | import torch 2 | import torch.nn as nn 3 | 4 | from taming.modules.losses.vqperceptual import * # TODO: taming dependency yes/no? 5 | 6 | 7 | class LPIPSWithDiscriminator(nn.Module): 8 | def __init__(self, disc_start, logvar_init=0.0, kl_weight=1.0, pixelloss_weight=1.0, 9 | disc_num_layers=3, disc_in_channels=3, disc_factor=1.0, disc_weight=1.0, 10 | perceptual_weight=1.0, use_actnorm=False, disc_conditional=False, 11 | disc_loss="hinge"): 12 | 13 | super().__init__() 14 | assert disc_loss in ["hinge", "vanilla"] 15 | self.kl_weight = kl_weight 16 | self.pixel_weight = pixelloss_weight 17 | self.perceptual_loss = LPIPS().eval() 18 | self.perceptual_weight = perceptual_weight 19 | # output log variance 20 | self.logvar = nn.Parameter(torch.ones(size=()) * logvar_init) 21 | 22 | self.discriminator = NLayerDiscriminator(input_nc=disc_in_channels, 23 | n_layers=disc_num_layers, 24 | use_actnorm=use_actnorm 25 | ).apply(weights_init) 26 | self.discriminator_iter_start = disc_start 27 | self.disc_loss = hinge_d_loss if disc_loss == "hinge" else vanilla_d_loss 28 | self.disc_factor = disc_factor 29 | self.discriminator_weight = disc_weight 30 | self.disc_conditional = disc_conditional 31 | 32 | def calculate_adaptive_weight(self, nll_loss, g_loss, last_layer=None): 33 | if last_layer is not None: 34 | nll_grads = torch.autograd.grad(nll_loss, last_layer, retain_graph=True)[0] 35 | g_grads = torch.autograd.grad(g_loss, last_layer, retain_graph=True)[0] 36 | else: 37 | nll_grads = torch.autograd.grad(nll_loss, self.last_layer[0], retain_graph=True)[0] 38 | g_grads = torch.autograd.grad(g_loss, self.last_layer[0], retain_graph=True)[0] 39 | 40 | d_weight = torch.norm(nll_grads) / (torch.norm(g_grads) + 1e-4) 41 | d_weight = torch.clamp(d_weight, 0.0, 1e4).detach() 42 | d_weight = d_weight * self.discriminator_weight 43 | return d_weight 44 | 45 | def forward(self, inputs, reconstructions, posteriors, optimizer_idx, 46 | global_step, last_layer=None, cond=None, split="train", 47 | weights=None): 48 | rec_loss = torch.abs(inputs.contiguous() - reconstructions.contiguous()) 49 | if self.perceptual_weight > 0: 50 | p_loss = self.perceptual_loss(inputs.contiguous(), reconstructions.contiguous()) 51 | rec_loss = rec_loss + self.perceptual_weight * p_loss 52 | 53 | nll_loss = rec_loss / torch.exp(self.logvar) + self.logvar 54 | weighted_nll_loss = nll_loss 55 | if weights is not None: 56 | weighted_nll_loss = weights*nll_loss 57 | weighted_nll_loss = torch.sum(weighted_nll_loss) / weighted_nll_loss.shape[0] 58 | nll_loss = torch.sum(nll_loss) / nll_loss.shape[0] 59 | kl_loss = posteriors.kl() 60 | kl_loss = torch.sum(kl_loss) / kl_loss.shape[0] 61 | 62 | # now the GAN part 63 | if optimizer_idx == 0: 64 | # generator update 65 | if cond is None: 66 | assert not self.disc_conditional 67 | logits_fake = self.discriminator(reconstructions.contiguous()) 68 | else: 69 | assert self.disc_conditional 70 | logits_fake = self.discriminator(torch.cat((reconstructions.contiguous(), cond), dim=1)) 71 | g_loss = -torch.mean(logits_fake) 72 | 73 | if self.disc_factor > 0.0: 74 | try: 75 | d_weight = self.calculate_adaptive_weight(nll_loss, g_loss, last_layer=last_layer) 76 | except RuntimeError: 77 | assert not self.training 78 | d_weight = torch.tensor(0.0) 79 | else: 80 | d_weight = torch.tensor(0.0) 81 | 82 | disc_factor = adopt_weight(self.disc_factor, global_step, threshold=self.discriminator_iter_start) 83 | loss = weighted_nll_loss + self.kl_weight * kl_loss + d_weight * disc_factor * g_loss 84 | 85 | log = {"{}/total_loss".format(split): loss.clone().detach().mean(), "{}/logvar".format(split): self.logvar.detach(), 86 | "{}/kl_loss".format(split): kl_loss.detach().mean(), "{}/nll_loss".format(split): nll_loss.detach().mean(), 87 | "{}/rec_loss".format(split): rec_loss.detach().mean(), 88 | "{}/d_weight".format(split): d_weight.detach(), 89 | "{}/disc_factor".format(split): torch.tensor(disc_factor), 90 | "{}/g_loss".format(split): g_loss.detach().mean(), 91 | } 92 | return loss, log 93 | 94 | if optimizer_idx == 1: 95 | # second pass for discriminator update 96 | if cond is None: 97 | logits_real = self.discriminator(inputs.contiguous().detach()) 98 | logits_fake = self.discriminator(reconstructions.contiguous().detach()) 99 | else: 100 | logits_real = self.discriminator(torch.cat((inputs.contiguous().detach(), cond), dim=1)) 101 | logits_fake = self.discriminator(torch.cat((reconstructions.contiguous().detach(), cond), dim=1)) 102 | 103 | disc_factor = adopt_weight(self.disc_factor, global_step, threshold=self.discriminator_iter_start) 104 | d_loss = disc_factor * self.disc_loss(logits_real, logits_fake) 105 | 106 | log = {"{}/disc_loss".format(split): d_loss.clone().detach().mean(), 107 | "{}/logits_real".format(split): logits_real.detach().mean(), 108 | "{}/logits_fake".format(split): logits_fake.detach().mean() 109 | } 110 | return d_loss, log 111 | 112 | -------------------------------------------------------------------------------- /ldm/util.py: -------------------------------------------------------------------------------- 1 | import importlib 2 | 3 | import torch 4 | import numpy as np 5 | from collections import abc 6 | from einops import rearrange 7 | from functools import partial 8 | 9 | import multiprocessing as mp 10 | from threading import Thread 11 | from queue import Queue 12 | 13 | from inspect import isfunction 14 | from PIL import Image, ImageDraw, ImageFont 15 | import torch.nn.functional as F 16 | 17 | def log_txt_as_img(wh, xc, size=10): 18 | # wh a tuple of (width, height) 19 | # xc a list of captions to plot 20 | b = len(xc) 21 | txts = list() 22 | for bi in range(b): 23 | txt = Image.new("RGB", wh, color="white") 24 | draw = ImageDraw.Draw(txt) 25 | font = ImageFont.truetype('data/DejaVuSans.ttf', size=size) 26 | nc = int(40 * (wh[0] / 256)) 27 | lines = "\n".join(xc[bi][start:start + nc] for start in range(0, len(xc[bi]), nc)) 28 | 29 | try: 30 | draw.text((0, 0), lines, fill="black", font=font) 31 | except UnicodeEncodeError: 32 | print("Cant encode string for logging. Skipping.") 33 | 34 | txt = np.array(txt).transpose(2, 0, 1) / 127.5 - 1.0 35 | txts.append(txt) 36 | txts = np.stack(txts) 37 | txts = torch.tensor(txts) 38 | return txts 39 | 40 | 41 | def ismap(x): 42 | if not isinstance(x, torch.Tensor): 43 | return False 44 | return (len(x.shape) == 4) and (x.shape[1] > 3) 45 | 46 | 47 | def isimage(x): 48 | if not isinstance(x, torch.Tensor): 49 | return False 50 | return (len(x.shape) == 4) and (x.shape[1] == 3 or x.shape[1] == 1) 51 | 52 | 53 | def exists(x): 54 | return x is not None 55 | 56 | 57 | def default(val, d): 58 | if exists(val): 59 | return val 60 | return d() if isfunction(d) else d 61 | 62 | 63 | def mean_flat(tensor): 64 | """ 65 | https://github.com/openai/guided-diffusion/blob/27c20a8fab9cb472df5d6bdd6c8d11c8f430b924/guided_diffusion/nn.py#L86 66 | Take the mean over all non-batch dimensions. 67 | """ 68 | return tensor.mean(dim=list(range(1, len(tensor.shape)))) 69 | 70 | 71 | def count_params(model, verbose=False): 72 | total_params = sum(p.numel() for p in model.parameters()) 73 | if verbose: 74 | print(f"{model.__class__.__name__} has {total_params * 1.e-6:.2f} M params.") 75 | return total_params 76 | 77 | 78 | def instantiate_from_config(config): 79 | if not "target" in config: 80 | if config == '__is_first_stage__': 81 | return None 82 | elif config == "__is_unconditional__": 83 | return None 84 | raise KeyError("Expected key `target` to instantiate.") 85 | return get_obj_from_str(config["target"])(**config.get("params", dict())) 86 | 87 | 88 | def get_obj_from_str(string, reload=False): 89 | module, cls = string.rsplit(".", 1) 90 | if reload: 91 | module_imp = importlib.import_module(module) 92 | importlib.reload(module_imp) 93 | return getattr(importlib.import_module(module, package=None), cls) 94 | 95 | 96 | def _do_parallel_data_prefetch(func, Q, data, idx, idx_to_fn=False): 97 | # create dummy dataset instance 98 | 99 | # run prefetching 100 | if idx_to_fn: 101 | res = func(data, worker_id=idx) 102 | else: 103 | res = func(data) 104 | Q.put([idx, res]) 105 | Q.put("Done") 106 | 107 | 108 | def parallel_data_prefetch( 109 | func: callable, data, n_proc, target_data_type="ndarray", cpu_intensive=True, use_worker_id=False 110 | ): 111 | # if target_data_type not in ["ndarray", "list"]: 112 | # raise ValueError( 113 | # "Data, which is passed to parallel_data_prefetch has to be either of type list or ndarray." 114 | # ) 115 | if isinstance(data, np.ndarray) and target_data_type == "list": 116 | raise ValueError("list expected but function got ndarray.") 117 | elif isinstance(data, abc.Iterable): 118 | if isinstance(data, dict): 119 | print( 120 | f'WARNING:"data" argument passed to parallel_data_prefetch is a dict: Using only its values and disregarding keys.' 121 | ) 122 | data = list(data.values()) 123 | if target_data_type == "ndarray": 124 | data = np.asarray(data) 125 | else: 126 | data = list(data) 127 | else: 128 | raise TypeError( 129 | f"The data, that shall be processed parallel has to be either an np.ndarray or an Iterable, but is actually {type(data)}." 130 | ) 131 | 132 | if cpu_intensive: 133 | Q = mp.Queue(1000) 134 | proc = mp.Process 135 | else: 136 | Q = Queue(1000) 137 | proc = Thread 138 | # spawn processes 139 | if target_data_type == "ndarray": 140 | arguments = [ 141 | [func, Q, part, i, use_worker_id] 142 | for i, part in enumerate(np.array_split(data, n_proc)) 143 | ] 144 | else: 145 | step = ( 146 | int(len(data) / n_proc + 1) 147 | if len(data) % n_proc != 0 148 | else int(len(data) / n_proc) 149 | ) 150 | arguments = [ 151 | [func, Q, part, i, use_worker_id] 152 | for i, part in enumerate( 153 | [data[i: i + step] for i in range(0, len(data), step)] 154 | ) 155 | ] 156 | processes = [] 157 | for i in range(n_proc): 158 | p = proc(target=_do_parallel_data_prefetch, args=arguments[i]) 159 | processes += [p] 160 | 161 | # start processes 162 | print(f"Start prefetching...") 163 | import time 164 | 165 | start = time.time() 166 | gather_res = [[] for _ in range(n_proc)] 167 | try: 168 | for p in processes: 169 | p.start() 170 | 171 | k = 0 172 | while k < n_proc: 173 | # get result 174 | res = Q.get() 175 | if res == "Done": 176 | k += 1 177 | else: 178 | gather_res[res[0]] = res[1] 179 | 180 | except Exception as e: 181 | print("Exception: ", e) 182 | for p in processes: 183 | p.terminate() 184 | 185 | raise e 186 | finally: 187 | for p in processes: 188 | p.join() 189 | print(f"Prefetching complete. [{time.time() - start} sec.]") 190 | 191 | if target_data_type == 'ndarray': 192 | if not isinstance(gather_res[0], np.ndarray): 193 | return np.concatenate([np.asarray(r) for r in gather_res], axis=0) 194 | 195 | # order outputs 196 | return np.concatenate(gather_res, axis=0) 197 | elif target_data_type == 'list': 198 | out = [] 199 | for r in gather_res: 200 | out.extend(r) 201 | return out 202 | else: 203 | return gather_res 204 | 205 | def preprocess_prompts(prompts): 206 | if isinstance(prompts, (list, tuple)): 207 | return [p.lower().strip().strip(".").strip() for p in prompts] 208 | elif isinstance(prompts, str): 209 | return prompts.lower().strip().strip(".").strip() 210 | else: 211 | raise NotImplementedError 212 | 213 | def block_single_pixel(value, scale_factor=4): 214 | e = torch.zeros(256) 215 | e[value] = 1 216 | e = rearrange(e, '(w h)-> w h', w=16) 217 | e_resized = F.interpolate(e.reshape(1,1,16,16), scale_factor=scale_factor)[0][0] 218 | e_resized = rearrange(e_resized, 'w h -> (w h)') 219 | return torch.where(e_resized==1)[0] 220 | 221 | def priority_for_2_class(pixel_avai, cls_mask): 222 | assert len(pixel_avai) == 2 223 | rank_1 = cls_mask[0][pixel_avai[0].sort(descending=True)[1]] 224 | rank_2 = cls_mask[1][pixel_avai[1].sort(descending=True)[1]] 225 | 226 | priority_for_1 = torch.cat((rank_1, rank_2.flip(dims=(0,))),0) 227 | priority_for_2 = torch.cat((rank_2, rank_1.flip(dims=(0,))),0) 228 | return [priority_for_1, priority_for_2] 229 | 230 | def priority_for_1_class(pixel_avai, cls_mask): 231 | assert len(pixel_avai) == 1 232 | rank_1 = cls_mask[0][pixel_avai[0].sort(descending=True)[1]] 233 | return [rank_1] 234 | 235 | def flatten(maps): 236 | return rearrange(maps, 'w h -> (w h)') 237 | 238 | def diff(t1, t2): 239 | combined = torch.cat((t1, t2)) 240 | uniques, counts = combined.unique(return_counts=True) 241 | difference = uniques[counts == 1] 242 | return difference -------------------------------------------------------------------------------- /ldm/modules/losses/vqperceptual.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch import nn 3 | import torch.nn.functional as F 4 | from einops import repeat 5 | 6 | from taming.modules.discriminator.model import NLayerDiscriminator, weights_init 7 | from taming.modules.losses.lpips import LPIPS 8 | from taming.modules.losses.vqperceptual import hinge_d_loss, vanilla_d_loss 9 | 10 | 11 | def hinge_d_loss_with_exemplar_weights(logits_real, logits_fake, weights): 12 | assert weights.shape[0] == logits_real.shape[0] == logits_fake.shape[0] 13 | loss_real = torch.mean(F.relu(1. - logits_real), dim=[1,2,3]) 14 | loss_fake = torch.mean(F.relu(1. + logits_fake), dim=[1,2,3]) 15 | loss_real = (weights * loss_real).sum() / weights.sum() 16 | loss_fake = (weights * loss_fake).sum() / weights.sum() 17 | d_loss = 0.5 * (loss_real + loss_fake) 18 | return d_loss 19 | 20 | def adopt_weight(weight, global_step, threshold=0, value=0.): 21 | if global_step < threshold: 22 | weight = value 23 | return weight 24 | 25 | 26 | def measure_perplexity(predicted_indices, n_embed): 27 | # src: https://github.com/karpathy/deep-vector-quantization/blob/main/model.py 28 | # eval cluster perplexity. when perplexity == num_embeddings then all clusters are used exactly equally 29 | encodings = F.one_hot(predicted_indices, n_embed).float().reshape(-1, n_embed) 30 | avg_probs = encodings.mean(0) 31 | perplexity = (-(avg_probs * torch.log(avg_probs + 1e-10)).sum()).exp() 32 | cluster_use = torch.sum(avg_probs > 0) 33 | return perplexity, cluster_use 34 | 35 | def l1(x, y): 36 | return torch.abs(x-y) 37 | 38 | 39 | def l2(x, y): 40 | return torch.pow((x-y), 2) 41 | 42 | 43 | class VQLPIPSWithDiscriminator(nn.Module): 44 | def __init__(self, disc_start, codebook_weight=1.0, pixelloss_weight=1.0, 45 | disc_num_layers=3, disc_in_channels=3, disc_factor=1.0, disc_weight=1.0, 46 | perceptual_weight=1.0, use_actnorm=False, disc_conditional=False, 47 | disc_ndf=64, disc_loss="hinge", n_classes=None, perceptual_loss="lpips", 48 | pixel_loss="l1"): 49 | super().__init__() 50 | assert disc_loss in ["hinge", "vanilla"] 51 | assert perceptual_loss in ["lpips", "clips", "dists"] 52 | assert pixel_loss in ["l1", "l2"] 53 | self.codebook_weight = codebook_weight 54 | self.pixel_weight = pixelloss_weight 55 | if perceptual_loss == "lpips": 56 | print(f"{self.__class__.__name__}: Running with LPIPS.") 57 | self.perceptual_loss = LPIPS().eval() 58 | else: 59 | raise ValueError(f"Unknown perceptual loss: >> {perceptual_loss} <<") 60 | self.perceptual_weight = perceptual_weight 61 | 62 | if pixel_loss == "l1": 63 | self.pixel_loss = l1 64 | else: 65 | self.pixel_loss = l2 66 | 67 | self.discriminator = NLayerDiscriminator(input_nc=disc_in_channels, 68 | n_layers=disc_num_layers, 69 | use_actnorm=use_actnorm, 70 | ndf=disc_ndf 71 | ).apply(weights_init) 72 | self.discriminator_iter_start = disc_start 73 | if disc_loss == "hinge": 74 | self.disc_loss = hinge_d_loss 75 | elif disc_loss == "vanilla": 76 | self.disc_loss = vanilla_d_loss 77 | else: 78 | raise ValueError(f"Unknown GAN loss '{disc_loss}'.") 79 | print(f"VQLPIPSWithDiscriminator running with {disc_loss} loss.") 80 | self.disc_factor = disc_factor 81 | self.discriminator_weight = disc_weight 82 | self.disc_conditional = disc_conditional 83 | self.n_classes = n_classes 84 | 85 | def calculate_adaptive_weight(self, nll_loss, g_loss, last_layer=None): 86 | if last_layer is not None: 87 | nll_grads = torch.autograd.grad(nll_loss, last_layer, retain_graph=True)[0] 88 | g_grads = torch.autograd.grad(g_loss, last_layer, retain_graph=True)[0] 89 | else: 90 | nll_grads = torch.autograd.grad(nll_loss, self.last_layer[0], retain_graph=True)[0] 91 | g_grads = torch.autograd.grad(g_loss, self.last_layer[0], retain_graph=True)[0] 92 | 93 | d_weight = torch.norm(nll_grads) / (torch.norm(g_grads) + 1e-4) 94 | d_weight = torch.clamp(d_weight, 0.0, 1e4).detach() 95 | d_weight = d_weight * self.discriminator_weight 96 | return d_weight 97 | 98 | def forward(self, codebook_loss, inputs, reconstructions, optimizer_idx, 99 | global_step, last_layer=None, cond=None, split="train", predicted_indices=None): 100 | if not exists(codebook_loss): 101 | codebook_loss = torch.tensor([0.]).to(inputs.device) 102 | #rec_loss = torch.abs(inputs.contiguous() - reconstructions.contiguous()) 103 | rec_loss = self.pixel_loss(inputs.contiguous(), reconstructions.contiguous()) 104 | if self.perceptual_weight > 0: 105 | p_loss = self.perceptual_loss(inputs.contiguous(), reconstructions.contiguous()) 106 | rec_loss = rec_loss + self.perceptual_weight * p_loss 107 | else: 108 | p_loss = torch.tensor([0.0]) 109 | 110 | nll_loss = rec_loss 111 | #nll_loss = torch.sum(nll_loss) / nll_loss.shape[0] 112 | nll_loss = torch.mean(nll_loss) 113 | 114 | # now the GAN part 115 | if optimizer_idx == 0: 116 | # generator update 117 | if cond is None: 118 | assert not self.disc_conditional 119 | logits_fake = self.discriminator(reconstructions.contiguous()) 120 | else: 121 | assert self.disc_conditional 122 | logits_fake = self.discriminator(torch.cat((reconstructions.contiguous(), cond), dim=1)) 123 | g_loss = -torch.mean(logits_fake) 124 | 125 | try: 126 | d_weight = self.calculate_adaptive_weight(nll_loss, g_loss, last_layer=last_layer) 127 | except RuntimeError: 128 | assert not self.training 129 | d_weight = torch.tensor(0.0) 130 | 131 | disc_factor = adopt_weight(self.disc_factor, global_step, threshold=self.discriminator_iter_start) 132 | loss = nll_loss + d_weight * disc_factor * g_loss + self.codebook_weight * codebook_loss.mean() 133 | 134 | log = {"{}/total_loss".format(split): loss.clone().detach().mean(), 135 | "{}/quant_loss".format(split): codebook_loss.detach().mean(), 136 | "{}/nll_loss".format(split): nll_loss.detach().mean(), 137 | "{}/rec_loss".format(split): rec_loss.detach().mean(), 138 | "{}/p_loss".format(split): p_loss.detach().mean(), 139 | "{}/d_weight".format(split): d_weight.detach(), 140 | "{}/disc_factor".format(split): torch.tensor(disc_factor), 141 | "{}/g_loss".format(split): g_loss.detach().mean(), 142 | } 143 | if predicted_indices is not None: 144 | assert self.n_classes is not None 145 | with torch.no_grad(): 146 | perplexity, cluster_usage = measure_perplexity(predicted_indices, self.n_classes) 147 | log[f"{split}/perplexity"] = perplexity 148 | log[f"{split}/cluster_usage"] = cluster_usage 149 | return loss, log 150 | 151 | if optimizer_idx == 1: 152 | # second pass for discriminator update 153 | if cond is None: 154 | logits_real = self.discriminator(inputs.contiguous().detach()) 155 | logits_fake = self.discriminator(reconstructions.contiguous().detach()) 156 | else: 157 | logits_real = self.discriminator(torch.cat((inputs.contiguous().detach(), cond), dim=1)) 158 | logits_fake = self.discriminator(torch.cat((reconstructions.contiguous().detach(), cond), dim=1)) 159 | 160 | disc_factor = adopt_weight(self.disc_factor, global_step, threshold=self.discriminator_iter_start) 161 | d_loss = disc_factor * self.disc_loss(logits_real, logits_fake) 162 | 163 | log = {"{}/disc_loss".format(split): d_loss.clone().detach().mean(), 164 | "{}/logits_real".format(split): logits_real.detach().mean(), 165 | "{}/logits_fake".format(split): logits_fake.detach().mean() 166 | } 167 | return d_loss, log 168 | -------------------------------------------------------------------------------- /ldm/modules/encoders/modules.py: -------------------------------------------------------------------------------- 1 | import torch 2 | import torch.nn as nn 3 | from functools import partial 4 | import clip 5 | from einops import rearrange, repeat 6 | from transformers import CLIPTokenizer, CLIPTextModel 7 | import kornia 8 | 9 | from ldm.modules.x_transformer import Encoder, TransformerWrapper # TODO: can we directly rely on lucidrains code and simply add this as a reuirement? --> test 10 | 11 | 12 | class AbstractEncoder(nn.Module): 13 | def __init__(self): 14 | super().__init__() 15 | 16 | def encode(self, *args, **kwargs): 17 | raise NotImplementedError 18 | 19 | 20 | 21 | class ClassEmbedder(nn.Module): 22 | def __init__(self, embed_dim, n_classes=1000, key='class'): 23 | super().__init__() 24 | self.key = key 25 | self.embedding = nn.Embedding(n_classes, embed_dim) 26 | 27 | def forward(self, batch, key=None): 28 | if key is None: 29 | key = self.key 30 | # this is for use in crossattn 31 | c = batch[key][:, None] 32 | c = self.embedding(c) 33 | return c 34 | 35 | 36 | class TransformerEmbedder(AbstractEncoder): 37 | """Some transformer encoder layers""" 38 | def __init__(self, n_embed, n_layer, vocab_size, max_seq_len=77, device="cuda"): 39 | super().__init__() 40 | self.device = device 41 | self.transformer = TransformerWrapper(num_tokens=vocab_size, max_seq_len=max_seq_len, 42 | attn_layers=Encoder(dim=n_embed, depth=n_layer)) 43 | 44 | def forward(self, tokens): 45 | tokens = tokens.to(self.device) # meh 46 | z = self.transformer(tokens, return_embeddings=True) 47 | return z 48 | 49 | def encode(self, x): 50 | return self(x) 51 | 52 | 53 | class BERTTokenizer(AbstractEncoder): 54 | """ Uses a pretrained BERT tokenizer by huggingface. Vocab size: 30522 (?)""" 55 | def __init__(self, device="cuda", vq_interface=True, max_length=77): 56 | super().__init__() 57 | from transformers import BertTokenizerFast # TODO: add to reuquirements 58 | self.tokenizer = BertTokenizerFast.from_pretrained("bert-base-uncased") 59 | self.device = device 60 | self.vq_interface = vq_interface 61 | self.max_length = max_length 62 | 63 | def forward(self, text): 64 | batch_encoding = self.tokenizer(text, truncation=True, max_length=self.max_length, return_length=True, 65 | return_overflowing_tokens=False, padding="max_length", return_tensors="pt") 66 | tokens = batch_encoding["input_ids"].to(self.device) 67 | return tokens 68 | 69 | @torch.no_grad() 70 | def encode(self, text): 71 | tokens = self(text) 72 | if not self.vq_interface: 73 | return tokens 74 | return None, None, [None, None, tokens] 75 | 76 | def decode(self, text): 77 | return text 78 | 79 | 80 | class BERTEmbedder(AbstractEncoder): 81 | """Uses the BERT tokenizr model and add some transformer encoder layers""" 82 | def __init__(self, n_embed, n_layer, vocab_size=30522, max_seq_len=77, 83 | device="cuda",use_tokenizer=True, embedding_dropout=0.0): 84 | super().__init__() 85 | self.use_tknz_fn = use_tokenizer 86 | if self.use_tknz_fn: 87 | self.tknz_fn = BERTTokenizer(vq_interface=False, max_length=max_seq_len) 88 | self.device = device 89 | self.transformer = TransformerWrapper(num_tokens=vocab_size, max_seq_len=max_seq_len, 90 | attn_layers=Encoder(dim=n_embed, depth=n_layer), 91 | emb_dropout=embedding_dropout) 92 | 93 | def forward(self, text): 94 | if self.use_tknz_fn: 95 | tokens = self.tknz_fn(text)#.to(self.device) 96 | else: 97 | tokens = text 98 | z = self.transformer(tokens, return_embeddings=True) 99 | return z 100 | 101 | def encode(self, text): 102 | # output of length 77 103 | return self(text) 104 | 105 | 106 | class SpatialRescaler(nn.Module): 107 | def __init__(self, 108 | n_stages=1, 109 | method='bilinear', 110 | multiplier=0.5, 111 | in_channels=3, 112 | out_channels=None, 113 | bias=False): 114 | super().__init__() 115 | self.n_stages = n_stages 116 | assert self.n_stages >= 0 117 | assert method in ['nearest','linear','bilinear','trilinear','bicubic','area'] 118 | self.multiplier = multiplier 119 | self.interpolator = partial(torch.nn.functional.interpolate, mode=method) 120 | self.remap_output = out_channels is not None 121 | if self.remap_output: 122 | print(f'Spatial Rescaler mapping from {in_channels} to {out_channels} channels after resizing.') 123 | self.channel_mapper = nn.Conv2d(in_channels,out_channels,1,bias=bias) 124 | 125 | def forward(self,x): 126 | for stage in range(self.n_stages): 127 | x = self.interpolator(x, scale_factor=self.multiplier) 128 | 129 | 130 | if self.remap_output: 131 | x = self.channel_mapper(x) 132 | return x 133 | 134 | def encode(self, x): 135 | return self(x) 136 | 137 | class FrozenCLIPEmbedder(AbstractEncoder): 138 | """Uses the CLIP transformer encoder for text (from Hugging Face)""" 139 | def __init__(self, version="openai/clip-vit-large-patch14", device="cuda", max_length=77): 140 | super().__init__() 141 | self.tokenizer = CLIPTokenizer.from_pretrained(version) 142 | self.transformer = CLIPTextModel.from_pretrained(version) 143 | self.device = device 144 | self.max_length = max_length 145 | self.freeze() 146 | 147 | def freeze(self): 148 | self.transformer = self.transformer.eval() 149 | for param in self.parameters(): 150 | param.requires_grad = False 151 | 152 | def forward(self, text): 153 | batch_encoding = self.tokenizer(text, truncation=True, max_length=self.max_length, return_length=True, 154 | return_overflowing_tokens=False, padding="max_length", return_tensors="pt") 155 | tokens = batch_encoding["input_ids"].to(self.device) 156 | outputs = self.transformer(input_ids=tokens) 157 | 158 | z = outputs.last_hidden_state 159 | return z 160 | 161 | def encode(self, text): 162 | return self(text) 163 | 164 | 165 | class FrozenCLIPTextEmbedder(nn.Module): 166 | """ 167 | Uses the CLIP transformer encoder for text. 168 | """ 169 | def __init__(self, version='ViT-L/14', device="cuda", max_length=77, n_repeat=1, normalize=True): 170 | super().__init__() 171 | self.model, _ = clip.load(version, jit=False, device="cpu") 172 | self.device = device 173 | self.max_length = max_length 174 | self.n_repeat = n_repeat 175 | self.normalize = normalize 176 | 177 | def freeze(self): 178 | self.model = self.model.eval() 179 | for param in self.parameters(): 180 | param.requires_grad = False 181 | 182 | def forward(self, text): 183 | tokens = clip.tokenize(text).to(self.device) 184 | z = self.model.encode_text(tokens) 185 | if self.normalize: 186 | z = z / torch.linalg.norm(z, dim=1, keepdim=True) 187 | return z 188 | 189 | def encode(self, text): 190 | z = self(text) 191 | if z.ndim==2: 192 | z = z[:, None, :] 193 | z = repeat(z, 'b 1 d -> b k d', k=self.n_repeat) 194 | return z 195 | 196 | 197 | class FrozenClipImageEmbedder(nn.Module): 198 | """ 199 | Uses the CLIP image encoder. 200 | """ 201 | def __init__( 202 | self, 203 | model, 204 | jit=False, 205 | device='cuda' if torch.cuda.is_available() else 'cpu', 206 | antialias=False, 207 | ): 208 | super().__init__() 209 | self.model, _ = clip.load(name=model, device=device, jit=jit) 210 | 211 | self.antialias = antialias 212 | 213 | self.register_buffer('mean', torch.Tensor([0.48145466, 0.4578275, 0.40821073]), persistent=False) 214 | self.register_buffer('std', torch.Tensor([0.26862954, 0.26130258, 0.27577711]), persistent=False) 215 | 216 | def preprocess(self, x): 217 | # normalize to [0,1] 218 | x = kornia.geometry.resize(x, (224, 224), 219 | interpolation='bicubic',align_corners=True, 220 | antialias=self.antialias) 221 | x = (x + 1.) / 2. 222 | # renormalize according to clip 223 | x = kornia.enhance.normalize(x, self.mean, self.std) 224 | return x 225 | 226 | def forward(self, x): 227 | # x is assumed to be in range [-1,1] 228 | return self.model.encode_image(self.preprocess(x)) 229 | 230 | 231 | if __name__ == "__main__": 232 | from ldm.util import count_params 233 | model = FrozenCLIPEmbedder() 234 | count_params(model, verbose=True) -------------------------------------------------------------------------------- /ldm/modules/diffusionmodules/util.py: -------------------------------------------------------------------------------- 1 | # adopted from 2 | # https://github.com/openai/improved-diffusion/blob/main/improved_diffusion/gaussian_diffusion.py 3 | # and 4 | # https://github.com/lucidrains/denoising-diffusion-pytorch/blob/7706bdfc6f527f58d33f84b7b522e61e6e3164b3/denoising_diffusion_pytorch/denoising_diffusion_pytorch.py 5 | # and 6 | # https://github.com/openai/guided-diffusion/blob/0ba878e517b276c45d1195eb29f6f5f72659a05b/guided_diffusion/nn.py 7 | # 8 | # thanks! 9 | 10 | 11 | import os 12 | import math 13 | import torch 14 | import torch.nn as nn 15 | import numpy as np 16 | from einops import repeat 17 | 18 | from ldm.util import instantiate_from_config 19 | 20 | 21 | def make_beta_schedule(schedule, n_timestep, linear_start=1e-4, linear_end=2e-2, cosine_s=8e-3): 22 | if schedule == "linear": 23 | betas = ( 24 | torch.linspace(linear_start ** 0.5, linear_end ** 0.5, n_timestep, dtype=torch.float64) ** 2 25 | ) 26 | 27 | elif schedule == "cosine": 28 | timesteps = ( 29 | torch.arange(n_timestep + 1, dtype=torch.float64) / n_timestep + cosine_s 30 | ) 31 | alphas = timesteps / (1 + cosine_s) * np.pi / 2 32 | alphas = torch.cos(alphas).pow(2) 33 | alphas = alphas / alphas[0] 34 | betas = 1 - alphas[1:] / alphas[:-1] 35 | betas = np.clip(betas, a_min=0, a_max=0.999) 36 | 37 | elif schedule == "sqrt_linear": 38 | betas = torch.linspace(linear_start, linear_end, n_timestep, dtype=torch.float64) 39 | elif schedule == "sqrt": 40 | betas = torch.linspace(linear_start, linear_end, n_timestep, dtype=torch.float64) ** 0.5 41 | else: 42 | raise ValueError(f"schedule '{schedule}' unknown.") 43 | return betas.numpy() 44 | 45 | 46 | def make_ddim_timesteps(ddim_discr_method, num_ddim_timesteps, num_ddpm_timesteps, verbose=True): 47 | if ddim_discr_method == 'uniform': 48 | c = num_ddpm_timesteps // num_ddim_timesteps 49 | ddim_timesteps = np.asarray(list(range(0, num_ddpm_timesteps, c))) 50 | elif ddim_discr_method == 'quad': 51 | ddim_timesteps = ((np.linspace(0, np.sqrt(num_ddpm_timesteps * .8), num_ddim_timesteps)) ** 2).astype(int) 52 | else: 53 | raise NotImplementedError(f'There is no ddim discretization method called "{ddim_discr_method}"') 54 | 55 | # assert ddim_timesteps.shape[0] == num_ddim_timesteps 56 | # add one to get the final alpha values right (the ones from first scale to data during sampling) 57 | steps_out = ddim_timesteps + 1 58 | if verbose: 59 | print(f'Selected timesteps for ddim sampler: {steps_out}') 60 | return steps_out 61 | 62 | 63 | def make_ddim_sampling_parameters(alphacums, ddim_timesteps, eta, verbose=True): 64 | # select alphas for computing the variance schedule 65 | alphas = alphacums[ddim_timesteps] 66 | alphas_prev = np.asarray([alphacums[0]] + alphacums[ddim_timesteps[:-1]].tolist()) 67 | 68 | # according the the formula provided in https://arxiv.org/abs/2010.02502 69 | sigmas = eta * np.sqrt((1 - alphas_prev) / (1 - alphas) * (1 - alphas / alphas_prev)) 70 | if verbose: 71 | print(f'Selected alphas for ddim sampler: a_t: {alphas}; a_(t-1): {alphas_prev}') 72 | print(f'For the chosen value of eta, which is {eta}, ' 73 | f'this results in the following sigma_t schedule for ddim sampler {sigmas}') 74 | return sigmas, alphas, alphas_prev 75 | 76 | 77 | def betas_for_alpha_bar(num_diffusion_timesteps, alpha_bar, max_beta=0.999): 78 | """ 79 | Create a beta schedule that discretizes the given alpha_t_bar function, 80 | which defines the cumulative product of (1-beta) over time from t = [0,1]. 81 | :param num_diffusion_timesteps: the number of betas to produce. 82 | :param alpha_bar: a lambda that takes an argument t from 0 to 1 and 83 | produces the cumulative product of (1-beta) up to that 84 | part of the diffusion process. 85 | :param max_beta: the maximum beta to use; use values lower than 1 to 86 | prevent singularities. 87 | """ 88 | betas = [] 89 | for i in range(num_diffusion_timesteps): 90 | t1 = i / num_diffusion_timesteps 91 | t2 = (i + 1) / num_diffusion_timesteps 92 | betas.append(min(1 - alpha_bar(t2) / alpha_bar(t1), max_beta)) 93 | return np.array(betas) 94 | 95 | 96 | def extract_into_tensor(a, t, x_shape): 97 | b, *_ = t.shape 98 | out = a.gather(-1, t) 99 | return out.reshape(b, *((1,) * (len(x_shape) - 1))) 100 | 101 | 102 | def checkpoint(func, inputs, params, flag): 103 | """ 104 | Evaluate a function without caching intermediate activations, allowing for 105 | reduced memory at the expense of extra compute in the backward pass. 106 | :param func: the function to evaluate. 107 | :param inputs: the argument sequence to pass to `func`. 108 | :param params: a sequence of parameters `func` depends on but does not 109 | explicitly take as arguments. 110 | :param flag: if False, disable gradient checkpointing. 111 | """ 112 | if flag: 113 | args = tuple(inputs) + tuple(params) 114 | return CheckpointFunction.apply(func, len(inputs), *args) 115 | else: 116 | return func(*inputs) 117 | 118 | 119 | class CheckpointFunction(torch.autograd.Function): 120 | @staticmethod 121 | def forward(ctx, run_function, length, *args): 122 | ctx.run_function = run_function 123 | ctx.input_tensors = list(args[:length]) 124 | ctx.input_params = list(args[length:]) 125 | 126 | with torch.no_grad(): 127 | output_tensors = ctx.run_function(*ctx.input_tensors) 128 | return output_tensors 129 | 130 | @staticmethod 131 | def backward(ctx, *output_grads): 132 | ctx.input_tensors = [x.detach().requires_grad_(True) for x in ctx.input_tensors] 133 | with torch.enable_grad(): 134 | # Fixes a bug where the first op in run_function modifies the 135 | # Tensor storage in place, which is not allowed for detach()'d 136 | # Tensors. 137 | shallow_copies = [x.view_as(x) for x in ctx.input_tensors] 138 | output_tensors = ctx.run_function(*shallow_copies) 139 | input_grads = torch.autograd.grad( 140 | output_tensors, 141 | ctx.input_tensors + ctx.input_params, 142 | output_grads, 143 | allow_unused=True, 144 | ) 145 | del ctx.input_tensors 146 | del ctx.input_params 147 | del output_tensors 148 | return (None, None) + input_grads 149 | 150 | 151 | def timestep_embedding(timesteps, dim, max_period=10000, repeat_only=False): 152 | """ 153 | Create sinusoidal timestep embeddings. 154 | :param timesteps: a 1-D Tensor of N indices, one per batch element. 155 | These may be fractional. 156 | :param dim: the dimension of the output. 157 | :param max_period: controls the minimum frequency of the embeddings. 158 | :return: an [N x dim] Tensor of positional embeddings. 159 | """ 160 | if not repeat_only: 161 | half = dim // 2 162 | freqs = torch.exp( 163 | -math.log(max_period) * torch.arange(start=0, end=half, dtype=torch.float32) / half 164 | ).to(device=timesteps.device) 165 | args = timesteps[:, None].float() * freqs[None] 166 | embedding = torch.cat([torch.cos(args), torch.sin(args)], dim=-1) 167 | if dim % 2: 168 | embedding = torch.cat([embedding, torch.zeros_like(embedding[:, :1])], dim=-1) 169 | else: 170 | embedding = repeat(timesteps, 'b -> b d', d=dim) 171 | return embedding 172 | 173 | 174 | def zero_module(module): 175 | """ 176 | Zero out the parameters of a module and return it. 177 | """ 178 | for p in module.parameters(): 179 | p.detach().zero_() 180 | return module 181 | 182 | 183 | def scale_module(module, scale): 184 | """ 185 | Scale the parameters of a module and return it. 186 | """ 187 | for p in module.parameters(): 188 | p.detach().mul_(scale) 189 | return module 190 | 191 | 192 | def mean_flat(tensor): 193 | """ 194 | Take the mean over all non-batch dimensions. 195 | """ 196 | return tensor.mean(dim=list(range(1, len(tensor.shape)))) 197 | 198 | 199 | def normalization(channels): 200 | """ 201 | Make a standard normalization layer. 202 | :param channels: number of input channels. 203 | :return: an nn.Module for normalization. 204 | """ 205 | return GroupNorm32(32, channels) 206 | 207 | 208 | # PyTorch 1.7 has SiLU, but we support PyTorch 1.5. 209 | class SiLU(nn.Module): 210 | def forward(self, x): 211 | return x * torch.sigmoid(x) 212 | 213 | 214 | class GroupNorm32(nn.GroupNorm): 215 | def forward(self, x): 216 | return super().forward(x.float()).type(x.dtype) 217 | 218 | def conv_nd(dims, *args, **kwargs): 219 | """ 220 | Create a 1D, 2D, or 3D convolution module. 221 | """ 222 | if dims == 1: 223 | return nn.Conv1d(*args, **kwargs) 224 | elif dims == 2: 225 | return nn.Conv2d(*args, **kwargs) 226 | elif dims == 3: 227 | return nn.Conv3d(*args, **kwargs) 228 | raise ValueError(f"unsupported dimensions: {dims}") 229 | 230 | 231 | def linear(*args, **kwargs): 232 | """ 233 | Create a linear module. 234 | """ 235 | return nn.Linear(*args, **kwargs) 236 | 237 | 238 | def avg_pool_nd(dims, *args, **kwargs): 239 | """ 240 | Create a 1D, 2D, or 3D average pooling module. 241 | """ 242 | if dims == 1: 243 | return nn.AvgPool1d(*args, **kwargs) 244 | elif dims == 2: 245 | return nn.AvgPool2d(*args, **kwargs) 246 | elif dims == 3: 247 | return nn.AvgPool3d(*args, **kwargs) 248 | raise ValueError(f"unsupported dimensions: {dims}") 249 | 250 | 251 | class HybridConditioner(nn.Module): 252 | 253 | def __init__(self, c_concat_config, c_crossattn_config): 254 | super().__init__() 255 | self.concat_conditioner = instantiate_from_config(c_concat_config) 256 | self.crossattn_conditioner = instantiate_from_config(c_crossattn_config) 257 | 258 | def forward(self, c_concat, c_crossattn): 259 | c_concat = self.concat_conditioner(c_concat) 260 | c_crossattn = self.crossattn_conditioner(c_crossattn) 261 | return {'c_concat': [c_concat], 'c_crossattn': [c_crossattn]} 262 | 263 | 264 | def noise_like(shape, device, repeat=False): 265 | repeat_noise = lambda: torch.randn((1, *shape[1:]), device=device).repeat(shape[0], *((1,) * (len(shape) - 1))) 266 | noise = lambda: torch.randn(shape, device=device) 267 | return repeat_noise() if repeat else noise() -------------------------------------------------------------------------------- /ldm/models/diffusion/classifier.py: -------------------------------------------------------------------------------- 1 | import os 2 | import torch 3 | import pytorch_lightning as pl 4 | from omegaconf import OmegaConf 5 | from torch.nn import functional as F 6 | from torch.optim import AdamW 7 | from torch.optim.lr_scheduler import LambdaLR 8 | from copy import deepcopy 9 | from einops import rearrange 10 | from glob import glob 11 | from natsort import natsorted 12 | 13 | from ldm.modules.diffusionmodules.openaimodel import EncoderUNetModel, UNetModel 14 | from ldm.util import log_txt_as_img, default, ismap, instantiate_from_config 15 | 16 | __models__ = { 17 | 'class_label': EncoderUNetModel, 18 | 'segmentation': UNetModel 19 | } 20 | 21 | 22 | def disabled_train(self, mode=True): 23 | """Overwrite model.train with this function to make sure train/eval mode 24 | does not change anymore.""" 25 | return self 26 | 27 | 28 | class NoisyLatentImageClassifier(pl.LightningModule): 29 | 30 | def __init__(self, 31 | diffusion_path, 32 | num_classes, 33 | ckpt_path=None, 34 | pool='attention', 35 | label_key=None, 36 | diffusion_ckpt_path=None, 37 | scheduler_config=None, 38 | weight_decay=1.e-2, 39 | log_steps=10, 40 | monitor='val/loss', 41 | *args, 42 | **kwargs): 43 | super().__init__(*args, **kwargs) 44 | self.num_classes = num_classes 45 | # get latest config of diffusion model 46 | diffusion_config = natsorted(glob(os.path.join(diffusion_path, 'configs', '*-project.yaml')))[-1] 47 | self.diffusion_config = OmegaConf.load(diffusion_config).model 48 | self.diffusion_config.params.ckpt_path = diffusion_ckpt_path 49 | self.load_diffusion() 50 | 51 | self.monitor = monitor 52 | self.numd = self.diffusion_model.first_stage_model.encoder.num_resolutions - 1 53 | self.log_time_interval = self.diffusion_model.num_timesteps // log_steps 54 | self.log_steps = log_steps 55 | 56 | self.label_key = label_key if not hasattr(self.diffusion_model, 'cond_stage_key') \ 57 | else self.diffusion_model.cond_stage_key 58 | 59 | assert self.label_key is not None, 'label_key neither in diffusion model nor in model.params' 60 | 61 | if self.label_key not in __models__: 62 | raise NotImplementedError() 63 | 64 | self.load_classifier(ckpt_path, pool) 65 | 66 | self.scheduler_config = scheduler_config 67 | self.use_scheduler = self.scheduler_config is not None 68 | self.weight_decay = weight_decay 69 | 70 | def init_from_ckpt(self, path, ignore_keys=list(), only_model=False): 71 | sd = torch.load(path, map_location="cpu") 72 | if "state_dict" in list(sd.keys()): 73 | sd = sd["state_dict"] 74 | keys = list(sd.keys()) 75 | for k in keys: 76 | for ik in ignore_keys: 77 | if k.startswith(ik): 78 | print("Deleting key {} from state_dict.".format(k)) 79 | del sd[k] 80 | missing, unexpected = self.load_state_dict(sd, strict=False) if not only_model else self.model.load_state_dict( 81 | sd, strict=False) 82 | print(f"Restored from {path} with {len(missing)} missing and {len(unexpected)} unexpected keys") 83 | if len(missing) > 0: 84 | print(f"Missing Keys: {missing}") 85 | if len(unexpected) > 0: 86 | print(f"Unexpected Keys: {unexpected}") 87 | 88 | def load_diffusion(self): 89 | model = instantiate_from_config(self.diffusion_config) 90 | self.diffusion_model = model.eval() 91 | self.diffusion_model.train = disabled_train 92 | for param in self.diffusion_model.parameters(): 93 | param.requires_grad = False 94 | 95 | def load_classifier(self, ckpt_path, pool): 96 | model_config = deepcopy(self.diffusion_config.params.unet_config.params) 97 | model_config.in_channels = self.diffusion_config.params.unet_config.params.out_channels 98 | model_config.out_channels = self.num_classes 99 | if self.label_key == 'class_label': 100 | model_config.pool = pool 101 | 102 | self.model = __models__[self.label_key](**model_config) 103 | if ckpt_path is not None: 104 | print('#####################################################################') 105 | print(f'load from ckpt "{ckpt_path}"') 106 | print('#####################################################################') 107 | self.init_from_ckpt(ckpt_path) 108 | 109 | @torch.no_grad() 110 | def get_x_noisy(self, x, t, noise=None): 111 | noise = default(noise, lambda: torch.randn_like(x)) 112 | continuous_sqrt_alpha_cumprod = None 113 | if self.diffusion_model.use_continuous_noise: 114 | continuous_sqrt_alpha_cumprod = self.diffusion_model.sample_continuous_noise_level(x.shape[0], t + 1) 115 | # todo: make sure t+1 is correct here 116 | 117 | return self.diffusion_model.q_sample(x_start=x, t=t, noise=noise, 118 | continuous_sqrt_alpha_cumprod=continuous_sqrt_alpha_cumprod) 119 | 120 | def forward(self, x_noisy, t, *args, **kwargs): 121 | return self.model(x_noisy, t) 122 | 123 | @torch.no_grad() 124 | def get_input(self, batch, k): 125 | x = batch[k] 126 | if len(x.shape) == 3: 127 | x = x[..., None] 128 | x = rearrange(x, 'b h w c -> b c h w') 129 | x = x.to(memory_format=torch.contiguous_format).float() 130 | return x 131 | 132 | @torch.no_grad() 133 | def get_conditioning(self, batch, k=None): 134 | if k is None: 135 | k = self.label_key 136 | assert k is not None, 'Needs to provide label key' 137 | 138 | targets = batch[k].to(self.device) 139 | 140 | if self.label_key == 'segmentation': 141 | targets = rearrange(targets, 'b h w c -> b c h w') 142 | for down in range(self.numd): 143 | h, w = targets.shape[-2:] 144 | targets = F.interpolate(targets, size=(h // 2, w // 2), mode='nearest') 145 | 146 | # targets = rearrange(targets,'b c h w -> b h w c') 147 | 148 | return targets 149 | 150 | def compute_top_k(self, logits, labels, k, reduction="mean"): 151 | _, top_ks = torch.topk(logits, k, dim=1) 152 | if reduction == "mean": 153 | return (top_ks == labels[:, None]).float().sum(dim=-1).mean().item() 154 | elif reduction == "none": 155 | return (top_ks == labels[:, None]).float().sum(dim=-1) 156 | 157 | def on_train_epoch_start(self): 158 | # save some memory 159 | self.diffusion_model.model.to('cpu') 160 | 161 | @torch.no_grad() 162 | def write_logs(self, loss, logits, targets): 163 | log_prefix = 'train' if self.training else 'val' 164 | log = {} 165 | log[f"{log_prefix}/loss"] = loss.mean() 166 | log[f"{log_prefix}/acc@1"] = self.compute_top_k( 167 | logits, targets, k=1, reduction="mean" 168 | ) 169 | log[f"{log_prefix}/acc@5"] = self.compute_top_k( 170 | logits, targets, k=5, reduction="mean" 171 | ) 172 | 173 | self.log_dict(log, prog_bar=False, logger=True, on_step=self.training, on_epoch=True) 174 | self.log('loss', log[f"{log_prefix}/loss"], prog_bar=True, logger=False) 175 | self.log('global_step', self.global_step, logger=False, on_epoch=False, prog_bar=True) 176 | lr = self.optimizers().param_groups[0]['lr'] 177 | self.log('lr_abs', lr, on_step=True, logger=True, on_epoch=False, prog_bar=True) 178 | 179 | def shared_step(self, batch, t=None): 180 | x, *_ = self.diffusion_model.get_input(batch, k=self.diffusion_model.first_stage_key) 181 | targets = self.get_conditioning(batch) 182 | if targets.dim() == 4: 183 | targets = targets.argmax(dim=1) 184 | if t is None: 185 | t = torch.randint(0, self.diffusion_model.num_timesteps, (x.shape[0],), device=self.device).long() 186 | else: 187 | t = torch.full(size=(x.shape[0],), fill_value=t, device=self.device).long() 188 | x_noisy = self.get_x_noisy(x, t) 189 | logits = self(x_noisy, t) 190 | 191 | loss = F.cross_entropy(logits, targets, reduction='none') 192 | 193 | self.write_logs(loss.detach(), logits.detach(), targets.detach()) 194 | 195 | loss = loss.mean() 196 | return loss, logits, x_noisy, targets 197 | 198 | def training_step(self, batch, batch_idx): 199 | loss, *_ = self.shared_step(batch) 200 | return loss 201 | 202 | def reset_noise_accs(self): 203 | self.noisy_acc = {t: {'acc@1': [], 'acc@5': []} for t in 204 | range(0, self.diffusion_model.num_timesteps, self.diffusion_model.log_every_t)} 205 | 206 | def on_validation_start(self): 207 | self.reset_noise_accs() 208 | 209 | @torch.no_grad() 210 | def validation_step(self, batch, batch_idx): 211 | loss, *_ = self.shared_step(batch) 212 | 213 | for t in self.noisy_acc: 214 | _, logits, _, targets = self.shared_step(batch, t) 215 | self.noisy_acc[t]['acc@1'].append(self.compute_top_k(logits, targets, k=1, reduction='mean')) 216 | self.noisy_acc[t]['acc@5'].append(self.compute_top_k(logits, targets, k=5, reduction='mean')) 217 | 218 | return loss 219 | 220 | def configure_optimizers(self): 221 | optimizer = AdamW(self.model.parameters(), lr=self.learning_rate, weight_decay=self.weight_decay) 222 | 223 | if self.use_scheduler: 224 | scheduler = instantiate_from_config(self.scheduler_config) 225 | 226 | print("Setting up LambdaLR scheduler...") 227 | scheduler = [ 228 | { 229 | 'scheduler': LambdaLR(optimizer, lr_lambda=scheduler.schedule), 230 | 'interval': 'step', 231 | 'frequency': 1 232 | }] 233 | return [optimizer], scheduler 234 | 235 | return optimizer 236 | 237 | @torch.no_grad() 238 | def log_images(self, batch, N=8, *args, **kwargs): 239 | log = dict() 240 | x = self.get_input(batch, self.diffusion_model.first_stage_key) 241 | log['inputs'] = x 242 | 243 | y = self.get_conditioning(batch) 244 | 245 | if self.label_key == 'class_label': 246 | y = log_txt_as_img((x.shape[2], x.shape[3]), batch["human_label"]) 247 | log['labels'] = y 248 | 249 | if ismap(y): 250 | log['labels'] = self.diffusion_model.to_rgb(y) 251 | 252 | for step in range(self.log_steps): 253 | current_time = step * self.log_time_interval 254 | 255 | _, logits, x_noisy, _ = self.shared_step(batch, t=current_time) 256 | 257 | log[f'inputs@t{current_time}'] = x_noisy 258 | 259 | pred = F.one_hot(logits.argmax(dim=1), num_classes=self.num_classes) 260 | pred = rearrange(pred, 'b h w c -> b c h w') 261 | 262 | log[f'pred@t{current_time}'] = self.diffusion_model.to_rgb(pred) 263 | 264 | for key in log: 265 | log[key] = log[key][:N] 266 | 267 | return log 268 | -------------------------------------------------------------------------------- /ldm/models/diffusion/_ddim.py: -------------------------------------------------------------------------------- 1 | """SAMPLING ONLY.""" 2 | 3 | import torch 4 | import numpy as np 5 | from tqdm import tqdm 6 | from functools import partial 7 | 8 | from ldm.modules.diffusionmodules.util import make_ddim_sampling_parameters, make_ddim_timesteps, noise_like, \ 9 | extract_into_tensor 10 | 11 | 12 | class DDIMSampler(object): 13 | def __init__(self, model, schedule="linear", **kwargs): 14 | super().__init__() 15 | self.model = model 16 | self.ddpm_num_timesteps = model.num_timesteps 17 | self.schedule = schedule 18 | 19 | def register_buffer(self, name, attr): 20 | if type(attr) == torch.Tensor: 21 | if attr.device != torch.device("cuda"): 22 | attr = attr.to(torch.device("cuda")) 23 | setattr(self, name, attr) 24 | 25 | def make_schedule(self, ddim_num_steps, ddim_discretize="uniform", ddim_eta=0., verbose=True): 26 | self.ddim_timesteps = make_ddim_timesteps(ddim_discr_method=ddim_discretize, num_ddim_timesteps=ddim_num_steps, 27 | num_ddpm_timesteps=self.ddpm_num_timesteps,verbose=verbose) 28 | alphas_cumprod = self.model.alphas_cumprod 29 | assert alphas_cumprod.shape[0] == self.ddpm_num_timesteps, 'alphas have to be defined for each timestep' 30 | to_torch = lambda x: x.clone().detach().to(torch.float32).to(self.model.device) 31 | 32 | self.register_buffer('betas', to_torch(self.model.betas)) 33 | self.register_buffer('alphas_cumprod', to_torch(alphas_cumprod)) 34 | self.register_buffer('alphas_cumprod_prev', to_torch(self.model.alphas_cumprod_prev)) 35 | 36 | # calculations for diffusion q(x_t | x_{t-1}) and others 37 | self.register_buffer('sqrt_alphas_cumprod', to_torch(np.sqrt(alphas_cumprod.cpu()))) 38 | self.register_buffer('sqrt_one_minus_alphas_cumprod', to_torch(np.sqrt(1. - alphas_cumprod.cpu()))) 39 | self.register_buffer('log_one_minus_alphas_cumprod', to_torch(np.log(1. - alphas_cumprod.cpu()))) 40 | self.register_buffer('sqrt_recip_alphas_cumprod', to_torch(np.sqrt(1. / alphas_cumprod.cpu()))) 41 | self.register_buffer('sqrt_recipm1_alphas_cumprod', to_torch(np.sqrt(1. / alphas_cumprod.cpu() - 1))) 42 | 43 | # ddim sampling parameters 44 | ddim_sigmas, ddim_alphas, ddim_alphas_prev = make_ddim_sampling_parameters(alphacums=alphas_cumprod.cpu(), 45 | ddim_timesteps=self.ddim_timesteps, 46 | eta=ddim_eta,verbose=verbose) 47 | self.register_buffer('ddim_sigmas', ddim_sigmas) 48 | self.register_buffer('ddim_alphas', ddim_alphas) 49 | self.register_buffer('ddim_alphas_prev', ddim_alphas_prev) 50 | self.register_buffer('ddim_sqrt_one_minus_alphas', np.sqrt(1. - ddim_alphas)) 51 | sigmas_for_original_sampling_steps = ddim_eta * torch.sqrt( 52 | (1 - self.alphas_cumprod_prev) / (1 - self.alphas_cumprod) * ( 53 | 1 - self.alphas_cumprod / self.alphas_cumprod_prev)) 54 | self.register_buffer('ddim_sigmas_for_original_num_steps', sigmas_for_original_sampling_steps) 55 | 56 | @torch.no_grad() 57 | def sample(self, 58 | S, 59 | batch_size, 60 | shape, 61 | conditioning=None, 62 | callback=None, 63 | normals_sequence=None, 64 | img_callback=None, 65 | quantize_x0=False, 66 | eta=0., 67 | mask=None, 68 | x0=None, 69 | temperature=1., 70 | noise_dropout=0., 71 | score_corrector=None, 72 | corrector_kwargs=None, 73 | verbose=True, 74 | x_T=None, 75 | log_every_t=100, 76 | unconditional_guidance_scale=1., 77 | unconditional_conditioning=None, 78 | # this has to come in the same format as the conditioning, # e.g. as encoded tokens, ... 79 | **kwargs 80 | ): 81 | if conditioning is not None: 82 | if isinstance(conditioning, dict): 83 | cbs = conditioning[list(conditioning.keys())[0]].shape[0] 84 | if cbs != batch_size: 85 | print(f"Warning: Got {cbs} conditionings but batch-size is {batch_size}") 86 | else: 87 | if conditioning.shape[0] != batch_size: 88 | print(f"Warning: Got {conditioning.shape[0]} conditionings but batch-size is {batch_size}") 89 | 90 | self.make_schedule(ddim_num_steps=S, ddim_eta=eta, verbose=verbose) 91 | # sampling 92 | C, H, W = shape 93 | size = (batch_size, C, H, W) 94 | print(f'Data shape for DDIM sampling is {size}, eta {eta}') 95 | 96 | samples, intermediates = self.ddim_sampling(conditioning, size, 97 | callback=callback, 98 | img_callback=img_callback, 99 | quantize_denoised=quantize_x0, 100 | mask=mask, x0=x0, 101 | ddim_use_original_steps=False, 102 | noise_dropout=noise_dropout, 103 | temperature=temperature, 104 | score_corrector=score_corrector, 105 | corrector_kwargs=corrector_kwargs, 106 | x_T=x_T, 107 | log_every_t=log_every_t, 108 | unconditional_guidance_scale=unconditional_guidance_scale, 109 | unconditional_conditioning=unconditional_conditioning, 110 | ) 111 | return samples, intermediates 112 | 113 | @torch.no_grad() 114 | def ddim_sampling(self, cond, shape, 115 | x_T=None, ddim_use_original_steps=False, 116 | callback=None, timesteps=None, quantize_denoised=False, 117 | mask=None, x0=None, img_callback=None, log_every_t=100, 118 | temperature=1., noise_dropout=0., score_corrector=None, corrector_kwargs=None, 119 | unconditional_guidance_scale=1., unconditional_conditioning=None,): 120 | device = self.model.betas.device 121 | b = shape[0] 122 | if x_T is None: 123 | img = torch.randn(shape, device=device) 124 | else: 125 | img = x_T 126 | 127 | if timesteps is None: 128 | timesteps = self.ddpm_num_timesteps if ddim_use_original_steps else self.ddim_timesteps 129 | elif timesteps is not None and not ddim_use_original_steps: 130 | subset_end = int(min(timesteps / self.ddim_timesteps.shape[0], 1) * self.ddim_timesteps.shape[0]) - 1 131 | timesteps = self.ddim_timesteps[:subset_end] 132 | 133 | intermediates = {'x_inter': [img], 'pred_x0': [img]} 134 | time_range = reversed(range(0,timesteps)) if ddim_use_original_steps else np.flip(timesteps) 135 | total_steps = timesteps if ddim_use_original_steps else timesteps.shape[0] 136 | print(f"Running DDIM Sampling with {total_steps} timesteps") 137 | 138 | iterator = tqdm(time_range, desc='DDIM Sampler', total=total_steps) 139 | 140 | for i, step in enumerate(iterator): 141 | index = total_steps - i - 1 142 | ts = torch.full((b,), step, device=device, dtype=torch.long) 143 | 144 | if mask is not None: 145 | assert x0 is not None 146 | img_orig = self.model.q_sample(x0, ts) # TODO: deterministic forward pass? 147 | img = img_orig * mask + (1. - mask) * img 148 | 149 | outs = self.p_sample_ddim(img, cond, ts, index=index, use_original_steps=ddim_use_original_steps, 150 | quantize_denoised=quantize_denoised, temperature=temperature, 151 | noise_dropout=noise_dropout, score_corrector=score_corrector, 152 | corrector_kwargs=corrector_kwargs, 153 | unconditional_guidance_scale=unconditional_guidance_scale, 154 | unconditional_conditioning=unconditional_conditioning) 155 | img, pred_x0 = outs 156 | if callback: callback(i) 157 | if img_callback: img_callback(pred_x0, i) 158 | 159 | if index % log_every_t == 0 or index == total_steps - 1: 160 | intermediates['x_inter'].append(img) 161 | intermediates['pred_x0'].append(pred_x0) 162 | 163 | return img, intermediates 164 | 165 | @torch.no_grad() 166 | def p_sample_ddim(self, x, c, t, index, repeat_noise=False, use_original_steps=False, quantize_denoised=False, 167 | temperature=1., noise_dropout=0., score_corrector=None, corrector_kwargs=None, 168 | unconditional_guidance_scale=1., unconditional_conditioning=None): 169 | b, *_, device = *x.shape, x.device 170 | 171 | if unconditional_conditioning is None or unconditional_guidance_scale == 1.: 172 | e_t = self.model.apply_model(x, t, c) 173 | else: 174 | x_in = torch.cat([x] * 2) 175 | t_in = torch.cat([t] * 2) 176 | c_in = torch.cat([unconditional_conditioning, c]) 177 | e_t_uncond, e_t = self.model.apply_model(x_in, t_in, c_in).chunk(2) 178 | e_t = e_t_uncond + unconditional_guidance_scale * (e_t - e_t_uncond) 179 | 180 | if score_corrector is not None: 181 | assert self.model.parameterization == "eps" 182 | e_t = score_corrector.modify_score(self.model, e_t, x, t, c, **corrector_kwargs) 183 | 184 | alphas = self.model.alphas_cumprod if use_original_steps else self.ddim_alphas 185 | alphas_prev = self.model.alphas_cumprod_prev if use_original_steps else self.ddim_alphas_prev 186 | sqrt_one_minus_alphas = self.model.sqrt_one_minus_alphas_cumprod if use_original_steps else self.ddim_sqrt_one_minus_alphas 187 | sigmas = self.model.ddim_sigmas_for_original_num_steps if use_original_steps else self.ddim_sigmas 188 | # select parameters corresponding to the currently considered timestep 189 | a_t = torch.full((b, 1, 1, 1), alphas[index], device=device) 190 | a_prev = torch.full((b, 1, 1, 1), alphas_prev[index], device=device) 191 | sigma_t = torch.full((b, 1, 1, 1), sigmas[index], device=device) 192 | sqrt_one_minus_at = torch.full((b, 1, 1, 1), sqrt_one_minus_alphas[index],device=device) 193 | 194 | # current prediction for x_0 195 | pred_x0 = (x - sqrt_one_minus_at * e_t) / a_t.sqrt() 196 | if quantize_denoised: 197 | pred_x0, _, *_ = self.model.first_stage_model.quantize(pred_x0) 198 | # direction pointing to x_t 199 | dir_xt = (1. - a_prev - sigma_t**2).sqrt() * e_t 200 | noise = sigma_t * noise_like(x.shape, device, repeat_noise) * temperature 201 | if noise_dropout > 0.: 202 | noise = torch.nn.functional.dropout(noise, p=noise_dropout) 203 | x_prev = a_prev.sqrt() * pred_x0 + dir_xt + noise 204 | return x_prev, pred_x0 205 | 206 | @torch.no_grad() 207 | def stochastic_encode(self, x0, t, use_original_steps=False, noise=None): 208 | # fast, but does not allow for exact reconstruction 209 | # t serves as an index to gather the correct alphas 210 | if use_original_steps: 211 | sqrt_alphas_cumprod = self.sqrt_alphas_cumprod 212 | sqrt_one_minus_alphas_cumprod = self.sqrt_one_minus_alphas_cumprod 213 | else: 214 | sqrt_alphas_cumprod = torch.sqrt(self.ddim_alphas) 215 | sqrt_one_minus_alphas_cumprod = self.ddim_sqrt_one_minus_alphas 216 | 217 | if noise is None: 218 | noise = torch.randn_like(x0) 219 | return (extract_into_tensor(sqrt_alphas_cumprod, t, x0.shape) * x0 + 220 | extract_into_tensor(sqrt_one_minus_alphas_cumprod, t, x0.shape) * noise) 221 | 222 | @torch.no_grad() 223 | def decode(self, x_latent, cond, t_start, unconditional_guidance_scale=1.0, unconditional_conditioning=None, 224 | use_original_steps=False): 225 | 226 | timesteps = np.arange(self.ddpm_num_timesteps) if use_original_steps else self.ddim_timesteps 227 | timesteps = timesteps[:t_start] 228 | 229 | time_range = np.flip(timesteps) 230 | total_steps = timesteps.shape[0] 231 | print(f"Running DDIM Sampling with {total_steps} timesteps") 232 | 233 | iterator = tqdm(time_range, desc='Decoding image', total=total_steps) 234 | x_dec = x_latent 235 | for i, step in enumerate(iterator): 236 | index = total_steps - i - 1 237 | ts = torch.full((x_latent.shape[0],), step, device=x_latent.device, dtype=torch.long) 238 | x_dec, _ = self.p_sample_ddim(x_dec, cond, ts, index=index, use_original_steps=use_original_steps, 239 | unconditional_guidance_scale=unconditional_guidance_scale, 240 | unconditional_conditioning=unconditional_conditioning) 241 | return x_dec -------------------------------------------------------------------------------- /ldm/data/imagenet.py: -------------------------------------------------------------------------------- 1 | import os, yaml, pickle, shutil, tarfile, glob 2 | import cv2 3 | import albumentations 4 | import PIL 5 | import numpy as np 6 | import torchvision.transforms.functional as TF 7 | from omegaconf import OmegaConf 8 | from functools import partial 9 | from PIL import Image 10 | from tqdm import tqdm 11 | from torch.utils.data import Dataset, Subset 12 | 13 | import taming.data.utils as tdu 14 | from taming.data.imagenet import str_to_indices, give_synsets_from_indices, download, retrieve 15 | from taming.data.imagenet import ImagePaths 16 | 17 | from ldm.modules.image_degradation import degradation_fn_bsr, degradation_fn_bsr_light 18 | 19 | 20 | def synset2idx(path_to_yaml="data/index_synset.yaml"): 21 | with open(path_to_yaml) as f: 22 | di2s = yaml.load(f) 23 | return dict((v,k) for k,v in di2s.items()) 24 | 25 | 26 | class ImageNetBase(Dataset): 27 | def __init__(self, config=None): 28 | self.config = config or OmegaConf.create() 29 | if not type(self.config)==dict: 30 | self.config = OmegaConf.to_container(self.config) 31 | self.keep_orig_class_label = self.config.get("keep_orig_class_label", False) 32 | self.process_images = True # if False we skip loading & processing images and self.data contains filepaths 33 | self._prepare() 34 | self._prepare_synset_to_human() 35 | self._prepare_idx_to_synset() 36 | self._prepare_human_to_integer_label() 37 | self._load() 38 | 39 | def __len__(self): 40 | return len(self.data) 41 | 42 | def __getitem__(self, i): 43 | return self.data[i] 44 | 45 | def _prepare(self): 46 | raise NotImplementedError() 47 | 48 | def _filter_relpaths(self, relpaths): 49 | ignore = set([ 50 | "n06596364_9591.JPEG", 51 | ]) 52 | relpaths = [rpath for rpath in relpaths if not rpath.split("/")[-1] in ignore] 53 | if "sub_indices" in self.config: 54 | indices = str_to_indices(self.config["sub_indices"]) 55 | synsets = give_synsets_from_indices(indices, path_to_yaml=self.idx2syn) # returns a list of strings 56 | self.synset2idx = synset2idx(path_to_yaml=self.idx2syn) 57 | files = [] 58 | for rpath in relpaths: 59 | syn = rpath.split("/")[0] 60 | if syn in synsets: 61 | files.append(rpath) 62 | return files 63 | else: 64 | return relpaths 65 | 66 | def _prepare_synset_to_human(self): 67 | SIZE = 2655750 68 | URL = "https://heibox.uni-heidelberg.de/f/9f28e956cd304264bb82/?dl=1" 69 | self.human_dict = os.path.join(self.root, "synset_human.txt") 70 | if (not os.path.exists(self.human_dict) or 71 | not os.path.getsize(self.human_dict)==SIZE): 72 | download(URL, self.human_dict) 73 | 74 | def _prepare_idx_to_synset(self): 75 | URL = "https://heibox.uni-heidelberg.de/f/d835d5b6ceda4d3aa910/?dl=1" 76 | self.idx2syn = os.path.join(self.root, "index_synset.yaml") 77 | if (not os.path.exists(self.idx2syn)): 78 | download(URL, self.idx2syn) 79 | 80 | def _prepare_human_to_integer_label(self): 81 | URL = "https://heibox.uni-heidelberg.de/f/2362b797d5be43b883f6/?dl=1" 82 | self.human2integer = os.path.join(self.root, "imagenet1000_clsidx_to_labels.txt") 83 | if (not os.path.exists(self.human2integer)): 84 | download(URL, self.human2integer) 85 | with open(self.human2integer, "r") as f: 86 | lines = f.read().splitlines() 87 | assert len(lines) == 1000 88 | self.human2integer_dict = dict() 89 | for line in lines: 90 | value, key = line.split(":") 91 | self.human2integer_dict[key] = int(value) 92 | 93 | def _load(self): 94 | with open(self.txt_filelist, "r") as f: 95 | self.relpaths = f.read().splitlines() 96 | l1 = len(self.relpaths) 97 | self.relpaths = self._filter_relpaths(self.relpaths) 98 | print("Removed {} files from filelist during filtering.".format(l1 - len(self.relpaths))) 99 | 100 | self.synsets = [p.split("/")[0] for p in self.relpaths] 101 | self.abspaths = [os.path.join(self.datadir, p) for p in self.relpaths] 102 | 103 | unique_synsets = np.unique(self.synsets) 104 | class_dict = dict((synset, i) for i, synset in enumerate(unique_synsets)) 105 | if not self.keep_orig_class_label: 106 | self.class_labels = [class_dict[s] for s in self.synsets] 107 | else: 108 | self.class_labels = [self.synset2idx[s] for s in self.synsets] 109 | 110 | with open(self.human_dict, "r") as f: 111 | human_dict = f.read().splitlines() 112 | human_dict = dict(line.split(maxsplit=1) for line in human_dict) 113 | 114 | self.human_labels = [human_dict[s] for s in self.synsets] 115 | 116 | labels = { 117 | "relpath": np.array(self.relpaths), 118 | "synsets": np.array(self.synsets), 119 | "class_label": np.array(self.class_labels), 120 | "human_label": np.array(self.human_labels), 121 | } 122 | 123 | if self.process_images: 124 | self.size = retrieve(self.config, "size", default=256) 125 | self.data = ImagePaths(self.abspaths, 126 | labels=labels, 127 | size=self.size, 128 | random_crop=self.random_crop, 129 | ) 130 | else: 131 | self.data = self.abspaths 132 | 133 | 134 | class ImageNetTrain(ImageNetBase): 135 | NAME = "ILSVRC2012_train" 136 | URL = "http://www.image-net.org/challenges/LSVRC/2012/" 137 | AT_HASH = "a306397ccf9c2ead27155983c254227c0fd938e2" 138 | FILES = [ 139 | "ILSVRC2012_img_train.tar", 140 | ] 141 | SIZES = [ 142 | 147897477120, 143 | ] 144 | 145 | def __init__(self, process_images=True, data_root=None, **kwargs): 146 | self.process_images = process_images 147 | self.data_root = data_root 148 | super().__init__(**kwargs) 149 | 150 | def _prepare(self): 151 | if self.data_root: 152 | self.root = os.path.join(self.data_root, self.NAME) 153 | else: 154 | cachedir = os.environ.get("XDG_CACHE_HOME", os.path.expanduser("~/.cache")) 155 | self.root = os.path.join(cachedir, "autoencoders/data", self.NAME) 156 | 157 | self.datadir = os.path.join(self.root, "data") 158 | self.txt_filelist = os.path.join(self.root, "filelist.txt") 159 | self.expected_length = 1281167 160 | self.random_crop = retrieve(self.config, "ImageNetTrain/random_crop", 161 | default=True) 162 | if not tdu.is_prepared(self.root): 163 | # prep 164 | print("Preparing dataset {} in {}".format(self.NAME, self.root)) 165 | 166 | datadir = self.datadir 167 | if not os.path.exists(datadir): 168 | path = os.path.join(self.root, self.FILES[0]) 169 | if not os.path.exists(path) or not os.path.getsize(path)==self.SIZES[0]: 170 | import academictorrents as at 171 | atpath = at.get(self.AT_HASH, datastore=self.root) 172 | assert atpath == path 173 | 174 | print("Extracting {} to {}".format(path, datadir)) 175 | os.makedirs(datadir, exist_ok=True) 176 | with tarfile.open(path, "r:") as tar: 177 | tar.extractall(path=datadir) 178 | 179 | print("Extracting sub-tars.") 180 | subpaths = sorted(glob.glob(os.path.join(datadir, "*.tar"))) 181 | for subpath in tqdm(subpaths): 182 | subdir = subpath[:-len(".tar")] 183 | os.makedirs(subdir, exist_ok=True) 184 | with tarfile.open(subpath, "r:") as tar: 185 | tar.extractall(path=subdir) 186 | 187 | filelist = glob.glob(os.path.join(datadir, "**", "*.JPEG")) 188 | filelist = [os.path.relpath(p, start=datadir) for p in filelist] 189 | filelist = sorted(filelist) 190 | filelist = "\n".join(filelist)+"\n" 191 | with open(self.txt_filelist, "w") as f: 192 | f.write(filelist) 193 | 194 | tdu.mark_prepared(self.root) 195 | 196 | 197 | class ImageNetValidation(ImageNetBase): 198 | NAME = "ILSVRC2012_validation" 199 | URL = "http://www.image-net.org/challenges/LSVRC/2012/" 200 | AT_HASH = "5d6d0df7ed81efd49ca99ea4737e0ae5e3a5f2e5" 201 | VS_URL = "https://heibox.uni-heidelberg.de/f/3e0f6e9c624e45f2bd73/?dl=1" 202 | FILES = [ 203 | "ILSVRC2012_img_val.tar", 204 | "validation_synset.txt", 205 | ] 206 | SIZES = [ 207 | 6744924160, 208 | 1950000, 209 | ] 210 | 211 | def __init__(self, process_images=True, data_root=None, **kwargs): 212 | self.data_root = data_root 213 | self.process_images = process_images 214 | super().__init__(**kwargs) 215 | 216 | def _prepare(self): 217 | if self.data_root: 218 | self.root = os.path.join(self.data_root, self.NAME) 219 | else: 220 | cachedir = os.environ.get("XDG_CACHE_HOME", os.path.expanduser("~/.cache")) 221 | self.root = os.path.join(cachedir, "autoencoders/data", self.NAME) 222 | self.datadir = os.path.join(self.root, "data") 223 | self.txt_filelist = os.path.join(self.root, "filelist.txt") 224 | self.expected_length = 50000 225 | self.random_crop = retrieve(self.config, "ImageNetValidation/random_crop", 226 | default=False) 227 | if not tdu.is_prepared(self.root): 228 | # prep 229 | print("Preparing dataset {} in {}".format(self.NAME, self.root)) 230 | 231 | datadir = self.datadir 232 | if not os.path.exists(datadir): 233 | path = os.path.join(self.root, self.FILES[0]) 234 | if not os.path.exists(path) or not os.path.getsize(path)==self.SIZES[0]: 235 | import academictorrents as at 236 | atpath = at.get(self.AT_HASH, datastore=self.root) 237 | assert atpath == path 238 | 239 | print("Extracting {} to {}".format(path, datadir)) 240 | os.makedirs(datadir, exist_ok=True) 241 | with tarfile.open(path, "r:") as tar: 242 | tar.extractall(path=datadir) 243 | 244 | vspath = os.path.join(self.root, self.FILES[1]) 245 | if not os.path.exists(vspath) or not os.path.getsize(vspath)==self.SIZES[1]: 246 | download(self.VS_URL, vspath) 247 | 248 | with open(vspath, "r") as f: 249 | synset_dict = f.read().splitlines() 250 | synset_dict = dict(line.split() for line in synset_dict) 251 | 252 | print("Reorganizing into synset folders") 253 | synsets = np.unique(list(synset_dict.values())) 254 | for s in synsets: 255 | os.makedirs(os.path.join(datadir, s), exist_ok=True) 256 | for k, v in synset_dict.items(): 257 | src = os.path.join(datadir, k) 258 | dst = os.path.join(datadir, v) 259 | shutil.move(src, dst) 260 | 261 | filelist = glob.glob(os.path.join(datadir, "**", "*.JPEG")) 262 | filelist = [os.path.relpath(p, start=datadir) for p in filelist] 263 | filelist = sorted(filelist) 264 | filelist = "\n".join(filelist)+"\n" 265 | with open(self.txt_filelist, "w") as f: 266 | f.write(filelist) 267 | 268 | tdu.mark_prepared(self.root) 269 | 270 | 271 | 272 | class ImageNetSR(Dataset): 273 | def __init__(self, size=None, 274 | degradation=None, downscale_f=4, min_crop_f=0.5, max_crop_f=1., 275 | random_crop=True): 276 | """ 277 | Imagenet Superresolution Dataloader 278 | Performs following ops in order: 279 | 1. crops a crop of size s from image either as random or center crop 280 | 2. resizes crop to size with cv2.area_interpolation 281 | 3. degrades resized crop with degradation_fn 282 | 283 | :param size: resizing to size after cropping 284 | :param degradation: degradation_fn, e.g. cv_bicubic or bsrgan_light 285 | :param downscale_f: Low Resolution Downsample factor 286 | :param min_crop_f: determines crop size s, 287 | where s = c * min_img_side_len with c sampled from interval (min_crop_f, max_crop_f) 288 | :param max_crop_f: "" 289 | :param data_root: 290 | :param random_crop: 291 | """ 292 | self.base = self.get_base() 293 | assert size 294 | assert (size / downscale_f).is_integer() 295 | self.size = size 296 | self.LR_size = int(size / downscale_f) 297 | self.min_crop_f = min_crop_f 298 | self.max_crop_f = max_crop_f 299 | assert(max_crop_f <= 1.) 300 | self.center_crop = not random_crop 301 | 302 | self.image_rescaler = albumentations.SmallestMaxSize(max_size=size, interpolation=cv2.INTER_AREA) 303 | 304 | self.pil_interpolation = False # gets reset later if incase interp_op is from pillow 305 | 306 | if degradation == "bsrgan": 307 | self.degradation_process = partial(degradation_fn_bsr, sf=downscale_f) 308 | 309 | elif degradation == "bsrgan_light": 310 | self.degradation_process = partial(degradation_fn_bsr_light, sf=downscale_f) 311 | 312 | else: 313 | interpolation_fn = { 314 | "cv_nearest": cv2.INTER_NEAREST, 315 | "cv_bilinear": cv2.INTER_LINEAR, 316 | "cv_bicubic": cv2.INTER_CUBIC, 317 | "cv_area": cv2.INTER_AREA, 318 | "cv_lanczos": cv2.INTER_LANCZOS4, 319 | "pil_nearest": PIL.Image.NEAREST, 320 | "pil_bilinear": PIL.Image.BILINEAR, 321 | "pil_bicubic": PIL.Image.BICUBIC, 322 | "pil_box": PIL.Image.BOX, 323 | "pil_hamming": PIL.Image.HAMMING, 324 | "pil_lanczos": PIL.Image.LANCZOS, 325 | }[degradation] 326 | 327 | self.pil_interpolation = degradation.startswith("pil_") 328 | 329 | if self.pil_interpolation: 330 | self.degradation_process = partial(TF.resize, size=self.LR_size, interpolation=interpolation_fn) 331 | 332 | else: 333 | self.degradation_process = albumentations.SmallestMaxSize(max_size=self.LR_size, 334 | interpolation=interpolation_fn) 335 | 336 | def __len__(self): 337 | return len(self.base) 338 | 339 | def __getitem__(self, i): 340 | example = self.base[i] 341 | image = Image.open(example["file_path_"]) 342 | 343 | if not image.mode == "RGB": 344 | image = image.convert("RGB") 345 | 346 | image = np.array(image).astype(np.uint8) 347 | 348 | min_side_len = min(image.shape[:2]) 349 | crop_side_len = min_side_len * np.random.uniform(self.min_crop_f, self.max_crop_f, size=None) 350 | crop_side_len = int(crop_side_len) 351 | 352 | if self.center_crop: 353 | self.cropper = albumentations.CenterCrop(height=crop_side_len, width=crop_side_len) 354 | 355 | else: 356 | self.cropper = albumentations.RandomCrop(height=crop_side_len, width=crop_side_len) 357 | 358 | image = self.cropper(image=image)["image"] 359 | image = self.image_rescaler(image=image)["image"] 360 | 361 | if self.pil_interpolation: 362 | image_pil = PIL.Image.fromarray(image) 363 | LR_image = self.degradation_process(image_pil) 364 | LR_image = np.array(LR_image).astype(np.uint8) 365 | 366 | else: 367 | LR_image = self.degradation_process(image=image)["image"] 368 | 369 | example["image"] = (image/127.5 - 1.0).astype(np.float32) 370 | example["LR_image"] = (LR_image/127.5 - 1.0).astype(np.float32) 371 | 372 | return example 373 | 374 | 375 | class ImageNetSRTrain(ImageNetSR): 376 | def __init__(self, **kwargs): 377 | super().__init__(**kwargs) 378 | 379 | def get_base(self): 380 | with open("data/imagenet_train_hr_indices.p", "rb") as f: 381 | indices = pickle.load(f) 382 | dset = ImageNetTrain(process_images=False,) 383 | return Subset(dset, indices) 384 | 385 | 386 | class ImageNetSRValidation(ImageNetSR): 387 | def __init__(self, **kwargs): 388 | super().__init__(**kwargs) 389 | 390 | def get_base(self): 391 | with open("data/imagenet_val_hr_indices.p", "rb") as f: 392 | indices = pickle.load(f) 393 | dset = ImageNetValidation(process_images=False,) 394 | return Subset(dset, indices) 395 | -------------------------------------------------------------------------------- /ldm/modules/attention.py: -------------------------------------------------------------------------------- 1 | from inspect import isfunction 2 | import math 3 | import pdb 4 | import torch 5 | import torch.nn.functional as F 6 | from torch import nn, einsum 7 | from einops import rearrange, repeat 8 | 9 | from ldm.modules.diffusionmodules.util import checkpoint 10 | import numpy as np 11 | import scipy.stats as st 12 | 13 | 14 | def exists(val): 15 | return val is not None 16 | 17 | 18 | def uniq(arr): 19 | return{el: True for el in arr}.keys() 20 | 21 | 22 | def default(val, d): 23 | if exists(val): 24 | return val 25 | return d() if isfunction(d) else d 26 | 27 | 28 | def max_neg_value(t): 29 | return -torch.finfo(t.dtype).max 30 | 31 | 32 | def init_(tensor): 33 | dim = tensor.shape[-1] 34 | std = 1 / math.sqrt(dim) 35 | tensor.uniform_(-std, std) 36 | return tensor 37 | 38 | 39 | # feedforward 40 | class GEGLU(nn.Module): 41 | def __init__(self, dim_in, dim_out): 42 | super().__init__() 43 | self.proj = nn.Linear(dim_in, dim_out * 2) 44 | 45 | def forward(self, x): 46 | x, gate = self.proj(x).chunk(2, dim=-1) 47 | return x * F.gelu(gate) 48 | 49 | 50 | class FeedForward(nn.Module): 51 | def __init__(self, dim, dim_out=None, mult=4, glu=False, dropout=0.): 52 | super().__init__() 53 | inner_dim = int(dim * mult) 54 | dim_out = default(dim_out, dim) 55 | project_in = nn.Sequential( 56 | nn.Linear(dim, inner_dim), 57 | nn.GELU() 58 | ) if not glu else GEGLU(dim, inner_dim) 59 | 60 | self.net = nn.Sequential( 61 | project_in, 62 | nn.Dropout(dropout), 63 | nn.Linear(inner_dim, dim_out) 64 | ) 65 | 66 | def forward(self, x): 67 | return self.net(x) 68 | 69 | 70 | def zero_module(module): 71 | """ 72 | Zero out the parameters of a module and return it. 73 | """ 74 | for p in module.parameters(): 75 | p.detach().zero_() 76 | return module 77 | 78 | 79 | def Normalize(in_channels): 80 | return torch.nn.GroupNorm(num_groups=32, num_channels=in_channels, eps=1e-6, affine=True) 81 | 82 | 83 | class LinearAttention(nn.Module): 84 | def __init__(self, dim, heads=4, dim_head=32): 85 | super().__init__() 86 | self.heads = heads 87 | hidden_dim = dim_head * heads 88 | self.to_qkv = nn.Conv2d(dim, hidden_dim * 3, 1, bias = False) 89 | self.to_out = nn.Conv2d(hidden_dim, dim, 1) 90 | 91 | def forward(self, x): 92 | b, c, h, w = x.shape 93 | qkv = self.to_qkv(x) 94 | q, k, v = rearrange(qkv, 'b (qkv heads c) h w -> qkv b heads c (h w)', heads = self.heads, qkv=3) 95 | k = k.softmax(dim=-1) 96 | context = torch.einsum('bhdn,bhen->bhde', k, v) 97 | out = torch.einsum('bhde,bhdn->bhen', context, q) 98 | out = rearrange(out, 'b heads c (h w) -> b (heads c) h w', heads=self.heads, h=h, w=w) 99 | return self.to_out(out) 100 | 101 | 102 | class SpatialSelfAttention(nn.Module): 103 | def __init__(self, in_channels): 104 | super().__init__() 105 | self.in_channels = in_channels 106 | 107 | self.norm = Normalize(in_channels) 108 | self.q = torch.nn.Conv2d(in_channels, 109 | in_channels, 110 | kernel_size=1, 111 | stride=1, 112 | padding=0) 113 | self.k = torch.nn.Conv2d(in_channels, 114 | in_channels, 115 | kernel_size=1, 116 | stride=1, 117 | padding=0) 118 | self.v = torch.nn.Conv2d(in_channels, 119 | in_channels, 120 | kernel_size=1, 121 | stride=1, 122 | padding=0) 123 | self.proj_out = torch.nn.Conv2d(in_channels, 124 | in_channels, 125 | kernel_size=1, 126 | stride=1, 127 | padding=0) 128 | 129 | def forward(self, x): 130 | h_ = x 131 | h_ = self.norm(h_) 132 | q = self.q(h_) 133 | k = self.k(h_) 134 | v = self.v(h_) 135 | 136 | # compute attention 137 | b,c,h,w = q.shape 138 | q = rearrange(q, 'b c h w -> b (h w) c') 139 | k = rearrange(k, 'b c h w -> b c (h w)') 140 | w_ = torch.einsum('bij,bjk->bik', q, k) 141 | 142 | w_ = w_ * (int(c)**(-0.5)) 143 | w_ = torch.nn.functional.softmax(w_, dim=2) 144 | 145 | # attend to values 146 | v = rearrange(v, 'b c h w -> b c (h w)') 147 | w_ = rearrange(w_, 'b i j -> b j i') 148 | h_ = torch.einsum('bij,bjk->bik', v, w_) 149 | h_ = rearrange(h_, 'b c (h w) -> b c h w', h=h) 150 | h_ = self.proj_out(h_) 151 | 152 | return x+h_ 153 | 154 | 155 | class CrossAttention(nn.Module): 156 | def __init__(self, query_dim, context_dim=None, heads=8, dim_head=64, dropout=0., struct_attn=False, save_map=False): 157 | super().__init__() 158 | inner_dim = dim_head * heads 159 | context_dim = default(context_dim, query_dim) 160 | 161 | self.scale = dim_head ** -0.5 162 | self.heads = heads 163 | 164 | self.to_q = nn.Linear(query_dim, inner_dim, bias=False) 165 | self.to_k = nn.Linear(context_dim, inner_dim, bias=False) 166 | self.to_v = nn.Linear(context_dim, inner_dim, bias=False) 167 | 168 | self.to_out = nn.Sequential( 169 | nn.Linear(inner_dim, query_dim), 170 | nn.Dropout(dropout) 171 | ) 172 | 173 | self.struct_attn = struct_attn 174 | self.save_map = save_map 175 | 176 | def schedule(self, total_step=50): 177 | # from 4.6 to 0 178 | seq = {} 179 | seq['linear'] = (torch.arange(total_step).flip(0) / total_step) * 4.6 180 | seq['log_smooth'] = torch.log(torch.arange(total_step).flip(0) * 100 + 1) 181 | seq['log_rapid'] = - torch.log(torch.arange(total_step) + 0.001)/1.5 182 | return seq 183 | 184 | def extra_mask_generation(self, down_scale, mask_cond): 185 | hw = int(512/down_scale) 186 | attn_extra = torch.ones([8, int(hw**2), 77]).cuda() * 0 187 | attn_extra = rearrange(attn_extra, 'b (h w) l -> b h w l', h = hw) 188 | obj_infs = mask_cond['object_infs'] 189 | for category in obj_infs: 190 | for bbox in category['loc']: 191 | x1, y1, x2, y2 = [xy/down_scale for xy in bbox] 192 | x1 = int(max(x1 - 1, 0)) 193 | y1 = int(max(y1 - 1, 0)) 194 | x2 = int(min(x2 + 1, hw)) 195 | y2 = int(min(y2 + 1, hw)) 196 | if mask_cond['para']['soft']: 197 | w = x2 - x1 198 | h = y2 - y1 199 | if w > h: 200 | offset = int(w/2) 201 | w_step = 1 202 | h_step = w/h 203 | else: 204 | offset = int(h/2) 205 | h_step = 1 206 | w_step = h/w 207 | if w > 1 and h > 1: 208 | y, x = np.mgrid[-offset:(offset+0.2):h_step, -offset:(offset+0.2):w_step] 209 | pos = np.empty(y.shape + (2,)) 210 | pos[:, :, 0] = (x/offset) * mask_cond['para']['L_soft'] 211 | pos[:, :, 1] = (y/offset) * mask_cond['para']['L_soft'] 212 | rv = st.multivariate_normal([0, 0], [[1, 0], [0, 1]]) 213 | _, tw, th = attn_extra[:, y1:y2, x1:x2, 0].shape 214 | value = torch.from_numpy(rv.pdf(pos)/(rv.pdf(pos).max())).cuda()[:tw, :th] 215 | for p in category['prompt']: 216 | if mask_cond['para']['neg_out']: 217 | attn_extra[:, :, :, p] = -9999999999 218 | attn_extra[:, y1:y2, x1:x2, p] = value 219 | else: 220 | for p in category['prompt']: 221 | if mask_cond['para']['neg_out']: 222 | attn_extra[:, :, :, p] = -9999999999 223 | attn_extra[:, y1:y2, x1:x2, p] = 1 224 | else: 225 | for p in category['prompt']: 226 | if mask_cond['para']['neg_out']: 227 | attn_extra[:, :, :, p] = -9999999999 228 | attn_extra[:, y1:y2, x1:x2, p] = 1 229 | attn_extra = rearrange(attn_extra, 'b h w l -> b (h w) l') 230 | return attn_extra 231 | 232 | def forward(self, x, context=None, t=None, mask_cond=None, mask=None): 233 | h = self.heads 234 | q = self.to_q(x) 235 | down_scale = 512 / math.sqrt(x.shape[1]) 236 | 237 | if t is not None: 238 | t = 50 - t - 1 239 | weights = self.schedule(total_step=50) 240 | weight = weights['linear'][t] 241 | if isinstance(context, list): 242 | if self.struct_attn: 243 | out = self.masked_qkv(q, context, weight, down_scale, mask_cond, mask) 244 | else: 245 | context = torch.cat([context[0], context[1]['k'][0]], dim=0) # use key tensor for context 246 | out = self.normal_qkv(q, context, mask) 247 | else: 248 | context = default(context, x) 249 | out = self.normal_qkv(q, context, mask) 250 | 251 | return self.to_out(out) 252 | 253 | def masked_qkv(self, q, context, weight, down_scale, mask_cond, mask): 254 | """ 255 | context: list of [uc, list of conditional context] 256 | """ 257 | uc_context = context[0] 258 | context_k, context_v = context[1]['k'], context[1]['v'] 259 | 260 | if isinstance(context_k, list) and isinstance(context_v, list): 261 | out = self._masked_qkv(q, uc_context, context_k, context_v, mask_cond, weight, down_scale, mask) 262 | else: 263 | raise NotImplementedError 264 | 265 | return out 266 | 267 | def _masked_qkv(self, q, uc_context, context_k, context_v, mask_cond, weight, down_scale, mask): 268 | h = self.heads 269 | 270 | assert uc_context.size(0) == context_k[0].size(0) == context_v[0].size(0) 271 | true_bs = uc_context.size(0) * h 272 | 273 | k_uc, v_uc = self.get_kv(uc_context) 274 | k_c = [self.to_k(c_k) for c_k in context_k] 275 | v_c = [self.to_v(c_v) for c_v in context_v] 276 | 277 | 278 | q = rearrange(q, 'b n (h d) -> (b h) n d', h=h) 279 | k_uc = rearrange(k_uc, 'b n (h d) -> (b h) n d', h=h) 280 | v_uc = rearrange(v_uc, 'b n (h d) -> (b h) n d', h=h) 281 | 282 | k_c = [rearrange(k, 'b n (h d) -> (b h) n d', h=h) for k in k_c] 283 | v_c = [rearrange(v, 'b n (h d) -> (b h) n d', h=h) for v in v_c] 284 | 285 | sim_uc = einsum('b i d, b j d -> b i j', q[:true_bs], k_uc) * self.scale 286 | sim_c = [einsum('b i d, b j d -> b i j', q[true_bs:], k) * self.scale for k in k_c] 287 | 288 | if mask_cond['is_use']: 289 | attn_extra = self.extra_mask_generation(down_scale, mask_cond) 290 | w_dot = mask_cond['para']['w_dot'] 291 | w = w_dot * weight * sim_c[0].max() 292 | sim_c[0] = sim_c[0] + w * attn_extra 293 | 294 | attn_uc = sim_uc.softmax(dim=-1) 295 | 296 | attn_c = [sim.softmax(dim=-1) for sim in sim_c] 297 | if self.save_map and sim_uc.size(1) != sim_uc.size(2): 298 | self.save_attn_maps(attn_c) 299 | if mask_cond['is_use']: 300 | self.save_extra_attn_maps([attn_extra]) 301 | 302 | out_uc = einsum('b i j, b j d -> b i d', attn_uc, v_uc) 303 | n_keys, n_values = len(k_c), len(v_c) 304 | if n_keys == n_values: 305 | out_c = sum([einsum('b i j, b j d -> b i d', attn, v) for attn, v in zip(attn_c, v_c)]) / len(v_c) 306 | else: 307 | assert n_keys == 1 or n_values == 1 308 | out_c = sum([einsum('b i j, b j d -> b i d', attn, v) for attn in attn_c for v in v_c]) / (n_keys * n_values) 309 | 310 | out = torch.cat([out_uc, out_c], dim=0) 311 | out = rearrange(out, '(b h) n d -> b n (h d)', h=h) 312 | 313 | return out 314 | 315 | 316 | def normal_qkv(self, q, context, mask): 317 | h = self.heads 318 | 319 | k = self.to_k(context) 320 | v = self.to_v(context) 321 | 322 | q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> (b h) n d', h=h), (q, k, v)) 323 | 324 | sim = einsum('b i d, b j d -> b i j', q, k) * self.scale 325 | 326 | if exists(mask): 327 | mask = rearrange(mask, 'b ... -> b (...)') 328 | max_neg_value = -torch.finfo(sim.dtype).max 329 | mask = repeat(mask, 'b j -> (b h) () j', h=h) 330 | sim.masked_fill_(~mask, max_neg_value) 331 | 332 | attn = sim.softmax(dim=-1) 333 | 334 | if self.save_map and sim.size(1) != sim.size(2): 335 | self.save_attn_maps(attn.chunk(2)[1]) 336 | 337 | out = einsum('b i j, b j d -> b i d', attn, v) 338 | out = rearrange(out, '(b h) n d -> b n (h d)', h=h) 339 | 340 | return out 341 | 342 | def get_kv(self, context): 343 | return self.to_k(context), self.to_v(context) 344 | 345 | def save_attn_maps(self, attn): 346 | h = self.heads 347 | if isinstance(attn, list): 348 | height = width = int(math.sqrt(attn[0].size(1))) 349 | self.attn_maps = [rearrange(m.detach(), '(b x) (h w) l -> b x h w l', x=h, h=height, w=width)[...,:40].cpu() for m in attn] 350 | else: 351 | height = width = int(math.sqrt(attn.size(1))) 352 | self.attn_maps = rearrange(attn.detach(), '(b x) (h w) l -> b x h w l', x=h, h=height, w=width)[...,:40].cpu() 353 | def save_extra_attn_maps(self, attn): 354 | h = self.heads 355 | if isinstance(attn, list): 356 | height = width = int(math.sqrt(attn[0].size(1))) 357 | self.attn_extra = [rearrange(m.detach(), '(b x) (h w) l -> b x h w l', x=h, h=height, w=width)[...,:40].cpu() for m in attn] 358 | else: 359 | height = width = int(math.sqrt(attn.size(1))) 360 | self.attn_extra = rearrange(attn.detach(), '(b x) (h w) l -> b x h w l', x=h, h=height, w=width)[...,:40].cpu() 361 | 362 | 363 | class BasicTransformerBlock(nn.Module): 364 | def __init__(self, dim, n_heads, d_head, dropout=0., context_dim=None, gated_ff=True, checkpoint=True, struct_attn=False, save_map=False): 365 | super().__init__() 366 | self.attn1 = CrossAttention(query_dim=dim, heads=n_heads, dim_head=d_head, dropout=dropout) # is a self-attention 367 | self.ff = FeedForward(dim, dropout=dropout, glu=gated_ff) 368 | self.attn2 = CrossAttention(query_dim=dim, context_dim=context_dim, 369 | heads=n_heads, dim_head=d_head, dropout=dropout, 370 | struct_attn=struct_attn, save_map=save_map) # is self-attn if context is none 371 | self.norm1 = nn.LayerNorm(dim) 372 | self.norm2 = nn.LayerNorm(dim) 373 | self.norm3 = nn.LayerNorm(dim) 374 | self.checkpoint = checkpoint 375 | 376 | def forward(self, x, context=None, t=None, mask_cond=None): 377 | return checkpoint(self._forward, (x, context, t, mask_cond), self.parameters(), self.checkpoint) 378 | 379 | def _forward(self, x, context=None, t=None, mask_cond=None): 380 | x = self.attn1(self.norm1(x)) + x 381 | x = self.attn2(self.norm2(x), context=context, t=t, mask_cond=mask_cond) + x 382 | x = self.ff(self.norm3(x)) + x 383 | return x 384 | 385 | 386 | class SpatialTransformer(nn.Module): 387 | """ 388 | Transformer block for image-like data. 389 | First, project the input (aka embedding) 390 | and reshape to b, t, d. 391 | Then apply standard transformer action. 392 | Finally, reshape to image 393 | """ 394 | def __init__(self, in_channels, n_heads, d_head, 395 | depth=1, dropout=0., context_dim=None, struct_attn=False, save_map=False): 396 | super().__init__() 397 | self.in_channels = in_channels 398 | inner_dim = n_heads * d_head 399 | self.norm = Normalize(in_channels) 400 | 401 | self.proj_in = nn.Conv2d(in_channels, 402 | inner_dim, 403 | kernel_size=1, 404 | stride=1, 405 | padding=0) 406 | 407 | self.transformer_blocks = nn.ModuleList( 408 | [BasicTransformerBlock(inner_dim, n_heads, d_head, dropout=dropout, context_dim=context_dim, struct_attn=struct_attn, save_map=save_map) 409 | for d in range(depth)] 410 | ) 411 | 412 | self.proj_out = zero_module(nn.Conv2d(inner_dim, 413 | in_channels, 414 | kernel_size=1, 415 | stride=1, 416 | padding=0)) 417 | self.struct_attn = struct_attn 418 | 419 | def forward(self, x, context=None, t=None, mask_cond=None): 420 | # note: if no context is given, cross-attention defaults to self-attention 421 | b, c, h, w = x.shape 422 | x_in = x 423 | x = self.norm(x) 424 | x = self.proj_in(x) 425 | x = rearrange(x, 'b c h w -> b (h w) c') 426 | for block in self.transformer_blocks: 427 | x = block(x, context=context, t=t, mask_cond=mask_cond) 428 | x = rearrange(x, 'b (h w) c -> b c h w', h=h, w=w) 429 | x = self.proj_out(x) 430 | return x + x_in -------------------------------------------------------------------------------- /ldm/models/diffusion/ddim.py: -------------------------------------------------------------------------------- 1 | """SAMPLING ONLY.""" 2 | 3 | from collections import defaultdict 4 | import torch 5 | import numpy as np 6 | from tqdm import tqdm 7 | from functools import partial 8 | 9 | from ldm.modules.diffusionmodules.util import make_ddim_sampling_parameters, make_ddim_timesteps, noise_like 10 | 11 | 12 | class DDIMSampler(object): 13 | def __init__(self, model, schedule="linear", **kwargs): 14 | super().__init__() 15 | self.model = model 16 | self.ddpm_num_timesteps = model.num_timesteps 17 | self.schedule = schedule 18 | 19 | def register_buffer(self, name, attr): 20 | if type(attr) == torch.Tensor: 21 | if attr.device != torch.device("cuda"): 22 | attr = attr.to(torch.device("cuda")) 23 | setattr(self, name, attr) 24 | 25 | def make_schedule(self, ddim_num_steps, ddim_discretize="uniform", ddim_eta=0., verbose=True): 26 | self.ddim_timesteps = make_ddim_timesteps(ddim_discr_method=ddim_discretize, num_ddim_timesteps=ddim_num_steps, 27 | num_ddpm_timesteps=self.ddpm_num_timesteps,verbose=verbose) 28 | alphas_cumprod = self.model.alphas_cumprod 29 | assert alphas_cumprod.shape[0] == self.ddpm_num_timesteps, 'alphas have to be defined for each timestep' 30 | to_torch = lambda x: x.clone().detach().to(torch.float32).to(self.model.device) 31 | 32 | self.register_buffer('betas', to_torch(self.model.betas)) 33 | self.register_buffer('alphas_cumprod', to_torch(alphas_cumprod)) 34 | self.register_buffer('alphas_cumprod_prev', to_torch(self.model.alphas_cumprod_prev)) 35 | 36 | # calculations for diffusion q(x_t | x_{t-1}) and others 37 | self.register_buffer('sqrt_alphas_cumprod', to_torch(np.sqrt(alphas_cumprod.cpu()))) 38 | self.register_buffer('sqrt_one_minus_alphas_cumprod', to_torch(np.sqrt(1. - alphas_cumprod.cpu()))) 39 | self.register_buffer('log_one_minus_alphas_cumprod', to_torch(np.log(1. - alphas_cumprod.cpu()))) 40 | self.register_buffer('sqrt_recip_alphas_cumprod', to_torch(np.sqrt(1. / alphas_cumprod.cpu()))) 41 | self.register_buffer('sqrt_recipm1_alphas_cumprod', to_torch(np.sqrt(1. / alphas_cumprod.cpu() - 1))) 42 | 43 | # ddim sampling parameters 44 | ddim_sigmas, ddim_alphas, ddim_alphas_prev = make_ddim_sampling_parameters(alphacums=alphas_cumprod.cpu(), 45 | ddim_timesteps=self.ddim_timesteps, 46 | eta=ddim_eta,verbose=verbose) 47 | self.register_buffer('ddim_sigmas', ddim_sigmas) 48 | self.register_buffer('ddim_alphas', ddim_alphas) 49 | self.register_buffer('ddim_alphas_prev', ddim_alphas_prev) 50 | self.register_buffer('ddim_sqrt_one_minus_alphas', np.sqrt(1. - ddim_alphas)) 51 | sigmas_for_original_sampling_steps = ddim_eta * torch.sqrt( 52 | (1 - self.alphas_cumprod_prev) / (1 - self.alphas_cumprod) * ( 53 | 1 - self.alphas_cumprod / self.alphas_cumprod_prev)) 54 | self.register_buffer('ddim_sigmas_for_original_num_steps', sigmas_for_original_sampling_steps) 55 | 56 | @torch.no_grad() 57 | def sample(self, 58 | S, 59 | batch_size, 60 | shape, 61 | conditioning=None, 62 | callback=None, 63 | normals_sequence=None, 64 | img_callback=None, 65 | quantize_x0=False, 66 | eta=0., 67 | mask=None, 68 | x0=None, 69 | temperature=1., 70 | noise_dropout=0., 71 | score_corrector=None, 72 | corrector_kwargs=None, 73 | verbose=True, 74 | x_T=None, 75 | log_every_t=100, 76 | unconditional_guidance_scale=1., 77 | unconditional_conditioning=None, 78 | # this has to come in the same format as the conditioning, # e.g. as encoded tokens, ... 79 | skip=False, 80 | quiet=False, 81 | mask_cond = None, 82 | **kwargs 83 | ): 84 | if conditioning is not None: 85 | assert isinstance(conditioning, dict) 86 | 87 | self.make_schedule(ddim_num_steps=S, ddim_eta=eta, verbose=verbose) 88 | # sampling 89 | C, H, W = shape 90 | size = (batch_size, C, H, W) 91 | 92 | samples, intermediates = self.ddim_sampling(conditioning, size, 93 | callback=callback, 94 | img_callback=img_callback, 95 | quantize_denoised=quantize_x0, 96 | mask=mask, x0=x0, 97 | ddim_use_original_steps=False, 98 | noise_dropout=noise_dropout, 99 | temperature=temperature, 100 | score_corrector=score_corrector, 101 | corrector_kwargs=corrector_kwargs, 102 | x_T=x_T, 103 | log_every_t=log_every_t, 104 | unconditional_guidance_scale=unconditional_guidance_scale, 105 | unconditional_conditioning=unconditional_conditioning, 106 | skip=skip, 107 | mask_cond = mask_cond, 108 | quiet=quiet, 109 | **kwargs 110 | ) 111 | return samples, intermediates 112 | 113 | @torch.no_grad() 114 | def ddim_sampling(self, cond, shape, 115 | x_T=None, ddim_use_original_steps=False, 116 | callback=None, timesteps=None, quantize_denoised=False, 117 | mask=None, x0=None, img_callback=None, log_every_t=100, 118 | temperature=1., noise_dropout=0., score_corrector=None, corrector_kwargs=None, mask_cond=None, 119 | unconditional_guidance_scale=1., unconditional_conditioning=None, skip=False, quiet=False, **kwargs): 120 | device = self.model.betas.device 121 | b = shape[0] 122 | if x_T is None: 123 | img = torch.randn(shape, device=device) 124 | else: 125 | img = x_T 126 | 127 | if timesteps is None: 128 | timesteps = self.ddpm_num_timesteps if ddim_use_original_steps else self.ddim_timesteps 129 | elif timesteps is not None and not ddim_use_original_steps: 130 | subset_end = int(min(timesteps / self.ddim_timesteps.shape[0], 1) * self.ddim_timesteps.shape[0]) - 1 131 | timesteps = self.ddim_timesteps[:subset_end] 132 | 133 | intermediates = {'x_inter': [img], 'pred_x0': [img]} 134 | if skip: 135 | return img, intermediates 136 | 137 | time_range = list(reversed(range(0,timesteps))) if ddim_use_original_steps else np.flip(timesteps) 138 | total_steps = timesteps if ddim_use_original_steps else timesteps.shape[0] 139 | 140 | if not quiet: 141 | iterator = tqdm(time_range, desc='DDIM Sampler', total=total_steps) 142 | else: 143 | iterator = time_range 144 | old_eps = [] 145 | self.attn_maps = defaultdict(list) 146 | 147 | for i, step in enumerate(iterator): 148 | index = total_steps - i - 1 149 | ts = torch.full((b,), step, device=device, dtype=torch.long) 150 | ts_next = torch.full((b,), time_range[min(i + 1, len(time_range) - 1)], device=device, dtype=torch.long) 151 | 152 | if mask is not None: 153 | assert x0 is not None 154 | img_orig = self.model.q_sample(x0, ts) # TODO: deterministic forward pass? 155 | img = img_orig * mask + (1. - mask) * img 156 | 157 | outs = self.p_sample_ddim(img, cond, ts, index=index, use_original_steps=ddim_use_original_steps, 158 | quantize_denoised=quantize_denoised, temperature=temperature, 159 | noise_dropout=noise_dropout, score_corrector=score_corrector, 160 | corrector_kwargs=corrector_kwargs, 161 | unconditional_guidance_scale=unconditional_guidance_scale, 162 | unconditional_conditioning=unconditional_conditioning, 163 | mask_cond=mask_cond, 164 | old_eps=old_eps, t_next=ts_next) 165 | img, pred_x0 = outs 166 | if callback: callback(i) 167 | if img_callback: img_callback(pred_x0, i) 168 | 169 | if index % log_every_t == 0 or index == total_steps - 1: 170 | intermediates['x_inter'].append(img) 171 | intermediates['pred_x0'].append(pred_x0) 172 | 173 | if kwargs.get('save_attn_maps', False): 174 | for name, module in self.model.model.diffusion_model.named_modules(): 175 | module_name = type(module).__name__ 176 | if module_name == 'CrossAttention' and 'attn2' in name: 177 | self.attn_maps[name].append(module.attn_maps) 178 | if mask_cond['is_use']: 179 | self.attn_maps[name+'_extra'].append(module.attn_extra) 180 | 181 | return img, intermediates 182 | 183 | @torch.no_grad() 184 | def get_attention(self, 185 | S, 186 | batch_size, 187 | shape, 188 | conditioning=None, 189 | callback=None, 190 | normals_sequence=None, 191 | img_callback=None, 192 | quantize_x0=False, 193 | eta=0., 194 | mask=None, 195 | x0=None, 196 | temperature=1., 197 | noise_dropout=0., 198 | score_corrector=None, 199 | corrector_kwargs=None, 200 | verbose=True, 201 | x_T=None, 202 | log_every_t=100, 203 | unconditional_guidance_scale=1., 204 | unconditional_conditioning=None, 205 | # this has to come in the same format as the conditioning, # e.g. as encoded tokens, ... 206 | skip=False, 207 | quiet=False, 208 | mask_cond = None, 209 | **kwargs 210 | ): 211 | if conditioning is not None: 212 | assert isinstance(conditioning, dict) 213 | 214 | self.make_schedule(ddim_num_steps=S, ddim_eta=eta, verbose=verbose) 215 | # sampling 216 | C, H, W = shape 217 | size = (batch_size, C, H, W) 218 | 219 | self.get_attention_(conditioning, size, 220 | callback=callback, 221 | img_callback=img_callback, 222 | quantize_denoised=quantize_x0, 223 | mask=mask, x0=x0, 224 | ddim_use_original_steps=False, 225 | noise_dropout=noise_dropout, 226 | temperature=temperature, 227 | score_corrector=score_corrector, 228 | corrector_kwargs=corrector_kwargs, 229 | x_T=x_T, 230 | log_every_t=log_every_t, 231 | unconditional_guidance_scale=unconditional_guidance_scale, 232 | unconditional_conditioning=unconditional_conditioning, 233 | skip=skip, 234 | mask_cond = mask_cond, 235 | quiet=quiet, 236 | **kwargs 237 | ) 238 | 239 | @torch.no_grad() 240 | def get_attention_(self, cond, shape, 241 | x_T=None, ddim_use_original_steps=False, 242 | callback=None, timesteps=None, quantize_denoised=False, 243 | mask=None, x0=None, img_callback=None, log_every_t=100, 244 | temperature=1., noise_dropout=0., score_corrector=None, corrector_kwargs=None, mask_cond=None, 245 | unconditional_guidance_scale=1., unconditional_conditioning=None, skip=False, quiet=False, **kwargs): 246 | device = self.model.betas.device 247 | b = shape[0] 248 | if x_T is None: 249 | img = torch.randn(shape, device=device) 250 | else: 251 | img = x_T 252 | 253 | timesteps = self.ddpm_num_timesteps if ddim_use_original_steps else self.ddim_timesteps 254 | 255 | time_range = list(reversed(range(0,timesteps))) if ddim_use_original_steps else np.flip(timesteps) 256 | total_steps = timesteps if ddim_use_original_steps else timesteps.shape[0] 257 | 258 | iterator = time_range 259 | old_eps = [] 260 | self.attn_maps = defaultdict(list) 261 | 262 | for i, step in enumerate(iterator): 263 | index = total_steps - i - 1 264 | ts = torch.full((b,), step, device=device, dtype=torch.long) 265 | ts_next = torch.full((b,), time_range[min(i + 1, len(time_range) - 1)], device=device, dtype=torch.long) 266 | 267 | outs = self.p_sample_ddim(img, cond, ts, index=index, use_original_steps=ddim_use_original_steps, 268 | quantize_denoised=quantize_denoised, temperature=temperature, 269 | noise_dropout=noise_dropout, score_corrector=score_corrector, 270 | corrector_kwargs=corrector_kwargs, 271 | unconditional_guidance_scale=unconditional_guidance_scale, 272 | unconditional_conditioning=unconditional_conditioning, 273 | mask_cond=mask_cond, 274 | old_eps=old_eps, t_next=ts_next) 275 | 276 | if kwargs.get('save_attn_maps', False): 277 | for name, module in self.model.model.diffusion_model.named_modules(): 278 | module_name = type(module).__name__ 279 | if module_name == 'CrossAttention' and 'attn2' in name: 280 | self.attn_maps[name].append(module.attn_maps) 281 | break 282 | 283 | @torch.no_grad() 284 | def p_sample_ddim(self, x, c, t, index, repeat_noise=False, use_original_steps=False, quantize_denoised=False, 285 | temperature=1., noise_dropout=0., score_corrector=None, corrector_kwargs=None,mask_cond=None, 286 | unconditional_guidance_scale=1., unconditional_conditioning=None, old_eps=None, t_next=None): 287 | b, *_, device = *x.shape, x.device 288 | 289 | #def get_model_output(x, t): 290 | if unconditional_conditioning is None or unconditional_guidance_scale == 1.: 291 | e_t = self.model.apply_model(x, t, c) 292 | else: 293 | x_in = torch.cat([x] * 2) 294 | t_in = torch.cat([t] * 2) 295 | if isinstance(c, (list, dict)): 296 | c_in = [unconditional_conditioning, c] 297 | else: 298 | c_in = torch.cat([unconditional_conditioning, c]) 299 | 300 | e_t_uncond, e_t = self.model.apply_model(x_in, t_in, c_in, mask_cond=mask_cond).chunk(2) 301 | e_t = e_t_uncond + unconditional_guidance_scale * (e_t - e_t_uncond) 302 | 303 | if score_corrector is not None: 304 | assert self.model.parameterization == "eps" 305 | e_t = score_corrector.modify_score(self.model, e_t, x, t, c, **corrector_kwargs) 306 | 307 | # return e_t 308 | 309 | alphas = self.model.alphas_cumprod if use_original_steps else self.ddim_alphas 310 | alphas_prev = self.model.alphas_cumprod_prev if use_original_steps else self.ddim_alphas_prev 311 | sqrt_one_minus_alphas = self.model.sqrt_one_minus_alphas_cumprod if use_original_steps else self.ddim_sqrt_one_minus_alphas 312 | sigmas = self.model.ddim_sigmas_for_original_num_steps if use_original_steps else self.ddim_sigmas 313 | 314 | #def get_x_prev_and_pred_x0(e_t, index): 315 | # select parameters corresponding to the currently considered timestep 316 | a_t = torch.full((b, 1, 1, 1), alphas[index], device=device) 317 | a_prev = torch.full((b, 1, 1, 1), alphas_prev[index], device=device) 318 | sigma_t = torch.full((b, 1, 1, 1), sigmas[index], device=device) 319 | sqrt_one_minus_at = torch.full((b, 1, 1, 1), sqrt_one_minus_alphas[index],device=device) 320 | 321 | # current prediction for x_0 322 | pred_x0 = (x - sqrt_one_minus_at * e_t) / a_t.sqrt() 323 | if quantize_denoised: 324 | pred_x0, _, *_ = self.model.first_stage_model.quantize(pred_x0) 325 | # direction pointing to x_t 326 | dir_xt = (1. - a_prev - sigma_t**2).sqrt() * e_t 327 | noise = sigma_t * noise_like(x.shape, device, repeat_noise) * temperature 328 | if noise_dropout > 0.: 329 | noise = torch.nn.functional.dropout(noise, p=noise_dropout) 330 | x_prev = a_prev.sqrt() * pred_x0 + dir_xt + noise 331 | return x_prev, pred_x0 332 | 333 | #e_t = get_model_output(x, t) 334 | #if len(old_eps) == 0: 335 | # # Pseudo Improved Euler (2nd order) 336 | # x_prev, pred_x0 = get_x_prev_and_pred_x0(e_t, index) 337 | # e_t_next = get_model_output(x_prev, t_next) 338 | # e_t_prime = (e_t + e_t_next) / 2 339 | #elif len(old_eps) == 1: 340 | # # 2nd order Pseudo Linear Multistep (Adams-Bashforth) 341 | # e_t_prime = (3 * e_t - old_eps[-1]) / 2 342 | #elif len(old_eps) == 2: 343 | # # 3nd order Pseudo Linear Multistep (Adams-Bashforth) 344 | # e_t_prime = (23 * e_t - 16 * old_eps[-1] + 5 * old_eps[-2]) / 12 345 | #elif len(old_eps) >= 3: 346 | # # 4nd order Pseudo Linear Multistep (Adams-Bashforth) 347 | # e_t_prime = (55 * e_t - 59 * old_eps[-1] + 37 * old_eps[-2] - 9 * old_eps[-3]) / 24 348 | 349 | #x_prev, pred_x0 = get_x_prev_and_pred_x0(e_t_prime, index) 350 | 351 | #return x_prev, pred_x0, e_t 352 | -------------------------------------------------------------------------------- /ldm/models/autoencoder.py: -------------------------------------------------------------------------------- 1 | import torch 2 | import pytorch_lightning as pl 3 | import torch.nn.functional as F 4 | from contextlib import contextmanager 5 | 6 | from taming.modules.vqvae.quantize import VectorQuantizer2 as VectorQuantizer 7 | 8 | from ldm.modules.diffusionmodules.model import Encoder, Decoder 9 | from ldm.modules.distributions.distributions import DiagonalGaussianDistribution 10 | 11 | from ldm.util import instantiate_from_config 12 | 13 | 14 | class VQModel(pl.LightningModule): 15 | def __init__(self, 16 | ddconfig, 17 | lossconfig, 18 | n_embed, 19 | embed_dim, 20 | ckpt_path=None, 21 | ignore_keys=[], 22 | image_key="image", 23 | colorize_nlabels=None, 24 | monitor=None, 25 | batch_resize_range=None, 26 | scheduler_config=None, 27 | lr_g_factor=1.0, 28 | remap=None, 29 | sane_index_shape=False, # tell vector quantizer to return indices as bhw 30 | use_ema=False 31 | ): 32 | super().__init__() 33 | self.embed_dim = embed_dim 34 | self.n_embed = n_embed 35 | self.image_key = image_key 36 | self.encoder = Encoder(**ddconfig) 37 | self.decoder = Decoder(**ddconfig) 38 | self.loss = instantiate_from_config(lossconfig) 39 | self.quantize = VectorQuantizer(n_embed, embed_dim, beta=0.25, 40 | remap=remap, 41 | sane_index_shape=sane_index_shape) 42 | self.quant_conv = torch.nn.Conv2d(ddconfig["z_channels"], embed_dim, 1) 43 | self.post_quant_conv = torch.nn.Conv2d(embed_dim, ddconfig["z_channels"], 1) 44 | if colorize_nlabels is not None: 45 | assert type(colorize_nlabels)==int 46 | self.register_buffer("colorize", torch.randn(3, colorize_nlabels, 1, 1)) 47 | if monitor is not None: 48 | self.monitor = monitor 49 | self.batch_resize_range = batch_resize_range 50 | if self.batch_resize_range is not None: 51 | print(f"{self.__class__.__name__}: Using per-batch resizing in range {batch_resize_range}.") 52 | 53 | self.use_ema = use_ema 54 | if self.use_ema: 55 | self.model_ema = LitEma(self) 56 | print(f"Keeping EMAs of {len(list(self.model_ema.buffers()))}.") 57 | 58 | if ckpt_path is not None: 59 | self.init_from_ckpt(ckpt_path, ignore_keys=ignore_keys) 60 | self.scheduler_config = scheduler_config 61 | self.lr_g_factor = lr_g_factor 62 | 63 | @contextmanager 64 | def ema_scope(self, context=None): 65 | if self.use_ema: 66 | self.model_ema.store(self.parameters()) 67 | self.model_ema.copy_to(self) 68 | if context is not None: 69 | print(f"{context}: Switched to EMA weights") 70 | try: 71 | yield None 72 | finally: 73 | if self.use_ema: 74 | self.model_ema.restore(self.parameters()) 75 | if context is not None: 76 | print(f"{context}: Restored training weights") 77 | 78 | def init_from_ckpt(self, path, ignore_keys=list()): 79 | sd = torch.load(path, map_location="cpu")["state_dict"] 80 | keys = list(sd.keys()) 81 | for k in keys: 82 | for ik in ignore_keys: 83 | if k.startswith(ik): 84 | print("Deleting key {} from state_dict.".format(k)) 85 | del sd[k] 86 | missing, unexpected = self.load_state_dict(sd, strict=False) 87 | print(f"Restored from {path} with {len(missing)} missing and {len(unexpected)} unexpected keys") 88 | if len(missing) > 0: 89 | print(f"Missing Keys: {missing}") 90 | print(f"Unexpected Keys: {unexpected}") 91 | 92 | def on_train_batch_end(self, *args, **kwargs): 93 | if self.use_ema: 94 | self.model_ema(self) 95 | 96 | def encode(self, x): 97 | h = self.encoder(x) 98 | h = self.quant_conv(h) 99 | quant, emb_loss, info = self.quantize(h) 100 | return quant, emb_loss, info 101 | 102 | def encode_to_prequant(self, x): 103 | h = self.encoder(x) 104 | h = self.quant_conv(h) 105 | return h 106 | 107 | def decode(self, quant): 108 | quant = self.post_quant_conv(quant) 109 | dec = self.decoder(quant) 110 | return dec 111 | 112 | def decode_code(self, code_b): 113 | quant_b = self.quantize.embed_code(code_b) 114 | dec = self.decode(quant_b) 115 | return dec 116 | 117 | def forward(self, input, return_pred_indices=False): 118 | quant, diff, (_,_,ind) = self.encode(input) 119 | dec = self.decode(quant) 120 | if return_pred_indices: 121 | return dec, diff, ind 122 | return dec, diff 123 | 124 | def get_input(self, batch, k): 125 | x = batch[k] 126 | if len(x.shape) == 3: 127 | x = x[..., None] 128 | x = x.permute(0, 3, 1, 2).to(memory_format=torch.contiguous_format).float() 129 | if self.batch_resize_range is not None: 130 | lower_size = self.batch_resize_range[0] 131 | upper_size = self.batch_resize_range[1] 132 | if self.global_step <= 4: 133 | # do the first few batches with max size to avoid later oom 134 | new_resize = upper_size 135 | else: 136 | new_resize = np.random.choice(np.arange(lower_size, upper_size+16, 16)) 137 | if new_resize != x.shape[2]: 138 | x = F.interpolate(x, size=new_resize, mode="bicubic") 139 | x = x.detach() 140 | return x 141 | 142 | def training_step(self, batch, batch_idx, optimizer_idx): 143 | # https://github.com/pytorch/pytorch/issues/37142 144 | # try not to fool the heuristics 145 | x = self.get_input(batch, self.image_key) 146 | xrec, qloss, ind = self(x, return_pred_indices=True) 147 | 148 | if optimizer_idx == 0: 149 | # autoencode 150 | aeloss, log_dict_ae = self.loss(qloss, x, xrec, optimizer_idx, self.global_step, 151 | last_layer=self.get_last_layer(), split="train", 152 | predicted_indices=ind) 153 | 154 | self.log_dict(log_dict_ae, prog_bar=False, logger=True, on_step=True, on_epoch=True) 155 | return aeloss 156 | 157 | if optimizer_idx == 1: 158 | # discriminator 159 | discloss, log_dict_disc = self.loss(qloss, x, xrec, optimizer_idx, self.global_step, 160 | last_layer=self.get_last_layer(), split="train") 161 | self.log_dict(log_dict_disc, prog_bar=False, logger=True, on_step=True, on_epoch=True) 162 | return discloss 163 | 164 | def validation_step(self, batch, batch_idx): 165 | log_dict = self._validation_step(batch, batch_idx) 166 | with self.ema_scope(): 167 | log_dict_ema = self._validation_step(batch, batch_idx, suffix="_ema") 168 | return log_dict 169 | 170 | def _validation_step(self, batch, batch_idx, suffix=""): 171 | x = self.get_input(batch, self.image_key) 172 | xrec, qloss, ind = self(x, return_pred_indices=True) 173 | aeloss, log_dict_ae = self.loss(qloss, x, xrec, 0, 174 | self.global_step, 175 | last_layer=self.get_last_layer(), 176 | split="val"+suffix, 177 | predicted_indices=ind 178 | ) 179 | 180 | discloss, log_dict_disc = self.loss(qloss, x, xrec, 1, 181 | self.global_step, 182 | last_layer=self.get_last_layer(), 183 | split="val"+suffix, 184 | predicted_indices=ind 185 | ) 186 | rec_loss = log_dict_ae[f"val{suffix}/rec_loss"] 187 | self.log(f"val{suffix}/rec_loss", rec_loss, 188 | prog_bar=True, logger=True, on_step=False, on_epoch=True, sync_dist=True) 189 | self.log(f"val{suffix}/aeloss", aeloss, 190 | prog_bar=True, logger=True, on_step=False, on_epoch=True, sync_dist=True) 191 | if version.parse(pl.__version__) >= version.parse('1.4.0'): 192 | del log_dict_ae[f"val{suffix}/rec_loss"] 193 | self.log_dict(log_dict_ae) 194 | self.log_dict(log_dict_disc) 195 | return self.log_dict 196 | 197 | def configure_optimizers(self): 198 | lr_d = self.learning_rate 199 | lr_g = self.lr_g_factor*self.learning_rate 200 | print("lr_d", lr_d) 201 | print("lr_g", lr_g) 202 | opt_ae = torch.optim.Adam(list(self.encoder.parameters())+ 203 | list(self.decoder.parameters())+ 204 | list(self.quantize.parameters())+ 205 | list(self.quant_conv.parameters())+ 206 | list(self.post_quant_conv.parameters()), 207 | lr=lr_g, betas=(0.5, 0.9)) 208 | opt_disc = torch.optim.Adam(self.loss.discriminator.parameters(), 209 | lr=lr_d, betas=(0.5, 0.9)) 210 | 211 | if self.scheduler_config is not None: 212 | scheduler = instantiate_from_config(self.scheduler_config) 213 | 214 | print("Setting up LambdaLR scheduler...") 215 | scheduler = [ 216 | { 217 | 'scheduler': LambdaLR(opt_ae, lr_lambda=scheduler.schedule), 218 | 'interval': 'step', 219 | 'frequency': 1 220 | }, 221 | { 222 | 'scheduler': LambdaLR(opt_disc, lr_lambda=scheduler.schedule), 223 | 'interval': 'step', 224 | 'frequency': 1 225 | }, 226 | ] 227 | return [opt_ae, opt_disc], scheduler 228 | return [opt_ae, opt_disc], [] 229 | 230 | def get_last_layer(self): 231 | return self.decoder.conv_out.weight 232 | 233 | def log_images(self, batch, only_inputs=False, plot_ema=False, **kwargs): 234 | log = dict() 235 | x = self.get_input(batch, self.image_key) 236 | x = x.to(self.device) 237 | if only_inputs: 238 | log["inputs"] = x 239 | return log 240 | xrec, _ = self(x) 241 | if x.shape[1] > 3: 242 | # colorize with random projection 243 | assert xrec.shape[1] > 3 244 | x = self.to_rgb(x) 245 | xrec = self.to_rgb(xrec) 246 | log["inputs"] = x 247 | log["reconstructions"] = xrec 248 | if plot_ema: 249 | with self.ema_scope(): 250 | xrec_ema, _ = self(x) 251 | if x.shape[1] > 3: xrec_ema = self.to_rgb(xrec_ema) 252 | log["reconstructions_ema"] = xrec_ema 253 | return log 254 | 255 | def to_rgb(self, x): 256 | assert self.image_key == "segmentation" 257 | if not hasattr(self, "colorize"): 258 | self.register_buffer("colorize", torch.randn(3, x.shape[1], 1, 1).to(x)) 259 | x = F.conv2d(x, weight=self.colorize) 260 | x = 2.*(x-x.min())/(x.max()-x.min()) - 1. 261 | return x 262 | 263 | 264 | class VQModelInterface(VQModel): 265 | def __init__(self, embed_dim, *args, **kwargs): 266 | super().__init__(embed_dim=embed_dim, *args, **kwargs) 267 | self.embed_dim = embed_dim 268 | 269 | def encode(self, x): 270 | h = self.encoder(x) 271 | h = self.quant_conv(h) 272 | return h 273 | 274 | def decode(self, h, force_not_quantize=False): 275 | # also go through quantization layer 276 | if not force_not_quantize: 277 | quant, emb_loss, info = self.quantize(h) 278 | else: 279 | quant = h 280 | quant = self.post_quant_conv(quant) 281 | dec = self.decoder(quant) 282 | return dec 283 | 284 | 285 | class AutoencoderKL(pl.LightningModule): 286 | def __init__(self, 287 | ddconfig, 288 | lossconfig, 289 | embed_dim, 290 | ckpt_path=None, 291 | ignore_keys=[], 292 | image_key="image", 293 | colorize_nlabels=None, 294 | monitor=None, 295 | ): 296 | super().__init__() 297 | self.image_key = image_key 298 | self.encoder = Encoder(**ddconfig) 299 | self.decoder = Decoder(**ddconfig) 300 | self.loss = instantiate_from_config(lossconfig) 301 | assert ddconfig["double_z"] 302 | self.quant_conv = torch.nn.Conv2d(2*ddconfig["z_channels"], 2*embed_dim, 1) 303 | self.post_quant_conv = torch.nn.Conv2d(embed_dim, ddconfig["z_channels"], 1) 304 | self.embed_dim = embed_dim 305 | if colorize_nlabels is not None: 306 | assert type(colorize_nlabels)==int 307 | self.register_buffer("colorize", torch.randn(3, colorize_nlabels, 1, 1)) 308 | if monitor is not None: 309 | self.monitor = monitor 310 | if ckpt_path is not None: 311 | self.init_from_ckpt(ckpt_path, ignore_keys=ignore_keys) 312 | 313 | def init_from_ckpt(self, path, ignore_keys=list()): 314 | sd = torch.load(path, map_location="cpu")["state_dict"] 315 | keys = list(sd.keys()) 316 | for k in keys: 317 | for ik in ignore_keys: 318 | if k.startswith(ik): 319 | print("Deleting key {} from state_dict.".format(k)) 320 | del sd[k] 321 | self.load_state_dict(sd, strict=False) 322 | print(f"Restored from {path}") 323 | 324 | def encode(self, x): 325 | h = self.encoder(x) 326 | moments = self.quant_conv(h) 327 | posterior = DiagonalGaussianDistribution(moments) 328 | return posterior 329 | 330 | def decode(self, z): 331 | z = self.post_quant_conv(z) 332 | dec = self.decoder(z) 333 | return dec 334 | 335 | def forward(self, input, sample_posterior=True): 336 | posterior = self.encode(input) 337 | if sample_posterior: 338 | z = posterior.sample() 339 | else: 340 | z = posterior.mode() 341 | dec = self.decode(z) 342 | return dec, posterior 343 | 344 | def get_input(self, batch, k): 345 | x = batch[k] 346 | if len(x.shape) == 3: 347 | x = x[..., None] 348 | x = x.permute(0, 3, 1, 2).to(memory_format=torch.contiguous_format).float() 349 | return x 350 | 351 | def training_step(self, batch, batch_idx, optimizer_idx): 352 | inputs = self.get_input(batch, self.image_key) 353 | reconstructions, posterior = self(inputs) 354 | 355 | if optimizer_idx == 0: 356 | # train encoder+decoder+logvar 357 | aeloss, log_dict_ae = self.loss(inputs, reconstructions, posterior, optimizer_idx, self.global_step, 358 | last_layer=self.get_last_layer(), split="train") 359 | self.log("aeloss", aeloss, prog_bar=True, logger=True, on_step=True, on_epoch=True) 360 | self.log_dict(log_dict_ae, prog_bar=False, logger=True, on_step=True, on_epoch=False) 361 | return aeloss 362 | 363 | if optimizer_idx == 1: 364 | # train the discriminator 365 | discloss, log_dict_disc = self.loss(inputs, reconstructions, posterior, optimizer_idx, self.global_step, 366 | last_layer=self.get_last_layer(), split="train") 367 | 368 | self.log("discloss", discloss, prog_bar=True, logger=True, on_step=True, on_epoch=True) 369 | self.log_dict(log_dict_disc, prog_bar=False, logger=True, on_step=True, on_epoch=False) 370 | return discloss 371 | 372 | def validation_step(self, batch, batch_idx): 373 | inputs = self.get_input(batch, self.image_key) 374 | reconstructions, posterior = self(inputs) 375 | aeloss, log_dict_ae = self.loss(inputs, reconstructions, posterior, 0, self.global_step, 376 | last_layer=self.get_last_layer(), split="val") 377 | 378 | discloss, log_dict_disc = self.loss(inputs, reconstructions, posterior, 1, self.global_step, 379 | last_layer=self.get_last_layer(), split="val") 380 | 381 | self.log("val/rec_loss", log_dict_ae["val/rec_loss"]) 382 | self.log_dict(log_dict_ae) 383 | self.log_dict(log_dict_disc) 384 | return self.log_dict 385 | 386 | def configure_optimizers(self): 387 | lr = self.learning_rate 388 | opt_ae = torch.optim.Adam(list(self.encoder.parameters())+ 389 | list(self.decoder.parameters())+ 390 | list(self.quant_conv.parameters())+ 391 | list(self.post_quant_conv.parameters()), 392 | lr=lr, betas=(0.5, 0.9)) 393 | opt_disc = torch.optim.Adam(self.loss.discriminator.parameters(), 394 | lr=lr, betas=(0.5, 0.9)) 395 | return [opt_ae, opt_disc], [] 396 | 397 | def get_last_layer(self): 398 | return self.decoder.conv_out.weight 399 | 400 | @torch.no_grad() 401 | def log_images(self, batch, only_inputs=False, **kwargs): 402 | log = dict() 403 | x = self.get_input(batch, self.image_key) 404 | x = x.to(self.device) 405 | if not only_inputs: 406 | xrec, posterior = self(x) 407 | if x.shape[1] > 3: 408 | # colorize with random projection 409 | assert xrec.shape[1] > 3 410 | x = self.to_rgb(x) 411 | xrec = self.to_rgb(xrec) 412 | log["samples"] = self.decode(torch.randn_like(posterior.sample())) 413 | log["reconstructions"] = xrec 414 | log["inputs"] = x 415 | return log 416 | 417 | def to_rgb(self, x): 418 | assert self.image_key == "segmentation" 419 | if not hasattr(self, "colorize"): 420 | self.register_buffer("colorize", torch.randn(3, x.shape[1], 1, 1).to(x)) 421 | x = F.conv2d(x, weight=self.colorize) 422 | x = 2.*(x-x.min())/(x.max()-x.min()) - 1. 423 | return x 424 | 425 | 426 | class IdentityFirstStage(torch.nn.Module): 427 | def __init__(self, *args, vq_interface=False, **kwargs): 428 | self.vq_interface = vq_interface # TODO: Should be true by default but check to not break older stuff 429 | super().__init__() 430 | 431 | def encode(self, x, *args, **kwargs): 432 | return x 433 | 434 | def decode(self, x, *args, **kwargs): 435 | return x 436 | 437 | def quantize(self, x, *args, **kwargs): 438 | if self.vq_interface: 439 | return x, None, [None, None, None] 440 | return x 441 | 442 | def forward(self, x, *args, **kwargs): 443 | return x 444 | -------------------------------------------------------------------------------- /utils.py: -------------------------------------------------------------------------------- 1 | import torch 2 | import random 3 | 4 | import matplotlib.pyplot as plt 5 | import torch.nn.functional as F 6 | import numpy as np 7 | 8 | from PIL import Image 9 | from tqdm import tqdm, trange 10 | from einops import rearrange 11 | from torchvision.utils import make_grid 12 | from torch import autocast 13 | from ldm.util import instantiate_from_config 14 | from itertools import combinations 15 | 16 | 17 | def load_model_from_config(config, ckpt, verbose=False): 18 | print(f"Loading model from {ckpt}") 19 | pl_sd = torch.load(ckpt, map_location="cpu") 20 | if "global_step" in pl_sd: 21 | print(f"Global Step: {pl_sd['global_step']}") 22 | sd = pl_sd["state_dict"] 23 | model = instantiate_from_config(config.model) 24 | m, u = model.load_state_dict(sd, strict=False) 25 | if len(m) > 0 and verbose: 26 | print("missing keys:") 27 | print(m) 28 | if len(u) > 0 and verbose: 29 | print("unexpected keys:") 30 | print(u) 31 | 32 | model.cuda() 33 | model.eval() 34 | return model 35 | 36 | def sampling(model, sampler, prompt, n_samples, scale=7.5, steps=50, conjunction=False, mask_cond=None, img=None): 37 | H = W = 512 38 | C = 4 39 | f = 8 40 | precision_scope = autocast 41 | with torch.no_grad(): 42 | with precision_scope("cuda"): 43 | with model.ema_scope(): 44 | all_samples = list() 45 | for n in range(n_samples): 46 | for bid, p in enumerate(prompt): 47 | 48 | uc = model.get_learned_conditioning([""]) 49 | _c = model.get_learned_conditioning(p) 50 | c = {'k': [_c], 'v': [_c]} 51 | shape = [C, H // f, W // f] 52 | 53 | samples_ddim, _ = sampler.sample(S=steps, 54 | conditioning=c, 55 | batch_size=1, 56 | shape=shape, 57 | verbose=False, 58 | unconditional_guidance_scale=scale, 59 | unconditional_conditioning=uc, 60 | eta=0.0, 61 | x_T=img, 62 | quiet=True, 63 | mask_cond = mask_cond, 64 | save_attn_maps=True) 65 | x_samples_ddim = model.decode_first_stage(samples_ddim) 66 | x_samples_ddim = torch.clamp((x_samples_ddim + 1.0) / 2.0, min=0.0, max=1.0) 67 | x_samples_ddim = x_samples_ddim.cpu().permute(0, 2, 3, 1).numpy() 68 | 69 | x_checked_image = x_samples_ddim 70 | x_checked_image_torch = torch.from_numpy(x_checked_image).permute(0, 3, 1, 2) 71 | all_samples.append(x_checked_image_torch) 72 | return all_samples 73 | 74 | def diff(t1, t2): 75 | combined = torch.cat((t1, t2)) 76 | uniques, counts = combined.unique(return_counts=True) 77 | difference = uniques[counts == 1] 78 | return difference 79 | 80 | def intersection(t1, t2): 81 | i = np.intersect1d(t1, t2) 82 | return torch.from_numpy(i) 83 | 84 | def block(value, scale_factor=4): 85 | vs = [] 86 | for v in value: 87 | e = torch.zeros(256) 88 | e[v] = 1 89 | e = rearrange(e, '(w h)-> w h', w=16) 90 | e_resized = F.interpolate(e.reshape(1,1,16,16), scale_factor=scale_factor)[0][0] 91 | e_resized = rearrange(e_resized, 'w h -> (w h)') 92 | vs.append(torch.where(e_resized==1)[0]) 93 | return vs 94 | 95 | def image_to_blocks(img): 96 | # input: [1,4,64,64] image 97 | # output: list of blocks, lenth is 256 98 | # block : [1, 4, 4, 4] 99 | blocks = [] 100 | for i in range(16): 101 | for j in range(16): 102 | block = img[:, :, i * 4: (i + 1) * 4, j * 4: (j + 1) * 4] 103 | blocks.append(block) 104 | return blocks 105 | 106 | def generate(model, sampler, img_, prompt, ind=None): 107 | mask_cond = { 108 | 'is_use': False, 109 | } 110 | ddim_steps = 50 111 | n_samples = 1 112 | scale = 7.5 113 | all_samples = sampling(model, sampler, prompt, 114 | n_samples, scale, 115 | ddim_steps, mask_cond=mask_cond, conjunction=False, img=img_) 116 | grid = torch.stack(all_samples, 0) 117 | grid = rearrange(grid, 'n b c h w -> (n b) c h w') 118 | grid = make_grid(grid, nrow=int(np.sqrt(n_samples))) 119 | grid = 255. * rearrange(grid, 'c h w -> h w c').cpu().numpy() 120 | img = Image.fromarray(grid.astype(np.uint8)) 121 | attn_maps = [item[0][0] for item in sampler.attn_maps['input_blocks.8.1.transformer_blocks.0.attn2']] 122 | maps = [torch.mean(item, axis=0) for item in attn_maps] 123 | maps = [rearrange(item, 'w h d -> d w h')[None,:] for item in maps] 124 | maps = rearrange(torch.cat(maps,dim=0), 't word w h -> word t w h') 125 | if ind is not None: 126 | plt.subplot(1, 5, 1) 127 | plt.imshow(img) 128 | plt.axis("off") 129 | plt.subplot(1, 5, 2) 130 | plt.imshow(maps[ind[0]][0],cmap = 'gray') 131 | plt.axis("off") 132 | plt.subplot(1, 5, 3) 133 | plt.imshow(maps[ind[1]][0],cmap = 'gray') 134 | plt.axis("off") 135 | plt.subplot(1, 5, 4) 136 | plt.imshow(maps[ind[0]][-1],cmap = 'gray') 137 | plt.axis("off") 138 | plt.subplot(1, 5, 5) 139 | plt.imshow(maps[ind[1]][-1],cmap = 'gray') 140 | plt.axis("off") 141 | plt.show() 142 | else: 143 | plt.imshow(img) 144 | plt.axis("off") 145 | plt.show() 146 | 147 | def preprocess_prompts(prompts): 148 | if isinstance(prompts, (list, tuple)): 149 | return [p.lower().strip().strip(".").strip() for p in prompts] 150 | elif isinstance(prompts, str): 151 | return prompts.lower().strip().strip(".").strip() 152 | else: 153 | raise NotImplementedError 154 | 155 | def intersection(t1, t2): 156 | i = np.intersect1d(t1, t2) 157 | return torch.from_numpy(i) 158 | 159 | def attention_to_score(attns): 160 | # input: [16, 16] attention maps 161 | # out put: list of score 162 | scores = [] 163 | for attn in attns: 164 | scores.append(rearrange(attn, 'w h -> (w h)').tolist()) 165 | return scores 166 | 167 | def score_normalize(scores): 168 | std = torch.std(scores, unbiased=False) 169 | mean = torch.mean(scores) 170 | scores = (scores - mean)/std 171 | scores = (scores - torch.min(scores)) / (torch.max(scores) - torch.min(scores)) 172 | return scores 173 | 174 | 175 | class pixel_block_base: 176 | def __init__(self, model, sampler, labels): 177 | self.model = model 178 | self.sampler = sampler 179 | self.H = 512 180 | self.W = 512 181 | self.C = 4 182 | self.f = 8 183 | self.normalize = False 184 | self.shape = [1, self.C, self.H // self.f, self.W // self.f] 185 | self.cond = {'is_use': False} 186 | 187 | self.base = {} 188 | self.base['blocks'] = [] 189 | for w in labels: 190 | self.base[w] = {} 191 | 192 | self.labels = labels 193 | self.combinations = list(combinations(labels, 2)) 194 | for pair in self.combinations: 195 | self.base[pair[0]][pair[1]] = torch.tensor([]) 196 | self.base[pair[1]][pair[0]] = torch.tensor([]) 197 | 198 | self.prompt = [] 199 | for pair in self.combinations: 200 | self.prompt.append('a ' + pair[0] + ' and a ' + pair[1] + '.') 201 | 202 | def _get_attention(self, prompt, img, scale=7.5, steps=50): 203 | precision_scope = autocast 204 | with torch.no_grad(): 205 | with precision_scope("cuda"): 206 | with self.model.ema_scope(): 207 | for bid, p in enumerate(prompt): 208 | p = preprocess_prompts(p) 209 | uc = self.model.get_learned_conditioning([""]) 210 | kv = self.model.get_learned_conditioning(p[0]) 211 | c = {'k':[kv], 'v': [kv]} 212 | shape = [self.C, self.H // self.f, self.W // self.f] 213 | self.sampler.get_attention(S=steps, 214 | conditioning=c, 215 | batch_size=1, 216 | shape=shape, 217 | verbose=False, 218 | unconditional_guidance_scale=scale, 219 | unconditional_conditioning=uc, 220 | eta=0.0, 221 | x_T=img, 222 | quiet=True, 223 | mask_cond=self.cond, 224 | save_attn_maps=True) 225 | all_attn_maps = [item[0][0] for item in self.sampler.attn_maps['input_blocks.8.1.transformer_blocks.0.attn2']] 226 | avg_maps = [torch.mean(item, axis=0) for item in all_attn_maps] 227 | avg_maps = [rearrange(item, 'w h d -> d w h')[None,:] for item in avg_maps] 228 | avg_maps = rearrange(torch.cat(avg_maps,dim=0), 't word w h -> word t w h') 229 | return avg_maps 230 | 231 | 232 | def make_base(self, n_img): 233 | for i in range(n_img): 234 | img_ = torch.randn(self.shape).cuda() 235 | _blocks = image_to_blocks(img_.clone().cpu()) 236 | self.base['blocks'].append(torch.stack(_blocks, dim=0)) 237 | for p in range(len(self.prompt)): 238 | caption = [[self.prompt[p]]] 239 | w = self.combinations[p] 240 | 241 | avg_maps = self._get_attention(caption, img_) 242 | 243 | m = [avg_maps[2][0]] 244 | _scores = attention_to_score(m)[0] 245 | self.base[w[0]][w[1]] = torch.cat([self.base[w[0]][w[1]], score_normalize(torch.tensor(_scores))], dim=0) 246 | m = [avg_maps[5][0]] 247 | _scores = attention_to_score(m)[0] 248 | self.base[w[1]][w[0]] = torch.cat([self.base[w[1]][w[0]], score_normalize(torch.tensor(_scores))], dim=0) 249 | self.base['blocks'] = torch.cat(self.base['blocks'], dim=0) 250 | 251 | # 对于每个类别，算一次总分， a 在与 b，c，d。。等类别同时出现时的得分相加 252 | for k in self.labels: 253 | avg = 0 254 | counter = 0 255 | for w in self.base[k].keys(): 256 | counter = counter + 1 257 | avg = avg + self.base[k][w] 258 | self.base[k]['average'] = avg / counter 259 | 260 | def make_base_by_list(self, n_img, comb_list): 261 | # comb_list : ['a', 'b'] 262 | 263 | for i in trange(n_img): 264 | img_ = torch.randn(self.shape).cuda() 265 | _blocks = image_to_blocks(img_.clone().cpu()) 266 | self.base['blocks'].append(torch.stack(_blocks, dim=0)) 267 | for p in trange(len(comb_list)): 268 | caption = [['A ' + comb_list[p][0] + ' and a ' + comb_list[p][1]+ '.'] ] 269 | w = comb_list[p] 270 | avg_maps = self._get_attention(caption, img_) 271 | 272 | m = [avg_maps[2][0]] 273 | _scores = attention_to_score(m)[0] 274 | self.base[w[0]][w[1]] = torch.cat([self.base[w[0]][w[1]], score_normalize(torch.tensor(_scores))], dim=0) 275 | 276 | m = [avg_maps[5][0]] 277 | _scores = attention_to_score(m)[0] 278 | self.base[w[1]][w[0]] = torch.cat([self.base[w[1]][w[0]], score_normalize(torch.tensor(_scores))], dim=0) 279 | self.base['blocks'] = torch.cat(self.base['blocks'], dim=0) 280 | 281 | 282 | 283 | def generate_region_mask(self, regions): 284 | region_mask = [] 285 | for r in regions: 286 | z = torch.zeros(16,16) 287 | z[r[1]:r[3], r[0]:r[2]] = 1 288 | region_mask.append(z) 289 | region_mask = rearrange(torch.stack(region_mask, dim = 0), 'n w h -> n (w h)') 290 | if len(regions) == 2: 291 | mask_1 = torch.where(region_mask[0] == 1)[0] 292 | mask_2 = torch.where(region_mask[1] == 1)[0] 293 | if mask_1.shape[0] > mask_2.shape[0]: 294 | region_mask[0][intersection(mask_1, mask_2)] = 0 295 | else: 296 | region_mask[1][intersection(mask_1, mask_2)] = 0 297 | if len(regions) == 3: 298 | mask_1 = torch.where(region_mask[0] == 1)[0] 299 | mask_2 = torch.where(region_mask[1] == 1)[0] 300 | mask_3 = torch.where(region_mask[2] == 1)[0] 301 | if mask_1.shape[0] > mask_2.shape[0]: 302 | region_mask[0][intersection(mask_1, mask_2)] = 0 303 | else: 304 | region_mask[1][intersection(mask_1, mask_2)] = 0 305 | 306 | if mask_1.shape[0] > mask_3.shape[0]: 307 | region_mask[0][intersection(mask_1, mask_3)] = 0 308 | else: 309 | region_mask[2][intersection(mask_1, mask_3)] = 0 310 | 311 | if mask_2.shape[0] > mask_3.shape[0]: 312 | region_mask[1][intersection(mask_2, mask_3)] = 0 313 | else: 314 | region_mask[2][intersection(mask_2, mask_3)] = 0 315 | 316 | bg_mask = 1 - region_mask.sum(axis=0) 317 | return region_mask, bg_mask 318 | 319 | def make_img(self, region_mask, bg_mask, obj_blocks, bg_blocks, recalibration): 320 | # input : masks indicating locations and blocks selected for corresponding contents 321 | img = torch.zeros([1,4,64,64]) 322 | for i in range(region_mask.shape[0]): 323 | num = torch.where(region_mask[i] != 0)[0].shape[0] 324 | r = rearrange(region_mask[i], '(w h) -> w h', w=16) 325 | total_num = obj_blocks[i].shape[0] 326 | if num > total_num: 327 | sampled_index = [] 328 | while num > total_num: 329 | sampled_index = sampled_index + random.sample(range(0, total_num), total_num) 330 | num = num - total_num 331 | sampled_index = sampled_index + random.sample(range(0, total_num), num) 332 | else: 333 | sampled_index = random.sample(range(0, total_num), num) 334 | positions = (r == 1).nonzero(as_tuple=False) 335 | selected_blocks_obj = obj_blocks[i][sampled_index] 336 | 337 | # recalibration 338 | if recalibration: 339 | print(selected_blocks_obj.mean()) 340 | print((selected_blocks_obj - selected_blocks_obj.mean()).pow(2).mean()) 341 | 342 | for j in range(len(positions)): 343 | p, q = positions[j] 344 | img[:, :, 4 * p : 4 * p + 4, 4 * q : 4 * q + 4] = selected_blocks_obj[j] 345 | 346 | bg_num = torch.where(bg_mask != 0)[0].shape[0] 347 | r = rearrange(bg_mask, '(w h) -> w h', w=16) 348 | bg_total_num = bg_blocks.shape[0] 349 | if bg_num > bg_total_num: 350 | sampled_index = [] 351 | while bg_num > bg_total_num: 352 | sampled_index = sampled_index + random.sample(range(0, bg_total_num), bg_total_num) 353 | bg_num = bg_num - bg_total_num 354 | sampled_index = sampled_index + random.sample(range(0, bg_total_num), bg_num) 355 | else: 356 | sampled_index = random.sample(range(0, bg_total_num), bg_num) 357 | bg_positions = (r == 1).nonzero(as_tuple=False) 358 | selected_blocks_bg = bg_blocks[sampled_index] 359 | for i in range(len(bg_positions)): 360 | p, q = bg_positions[i] 361 | img[:, :, 4 * p : 4 * p + 4, 4 * q : 4 * q + 4] = selected_blocks_bg[i] 362 | return img 363 | 364 | def product_image(self, words, regions, t_pos_1 = 0.5, t_bg_1 = 0.3, t_pos_2 = 0.3, t_neg_2 = 0.3, t_bg_2 = 0.1, recalibration = False): 365 | if len(words) == 1: 366 | # Fetch pre-collected pixel blocks and their scores 367 | word = words[0] 368 | scores = self.base[word]['average'] 369 | blocks = self.base['blocks'] 370 | # select blocks for obj 371 | # threshold [TO DO: sort?] 372 | blocks_index = torch.where(scores > t_pos_1)[0].numpy() 373 | # select blocks for bg 374 | blocks_index_bg = torch.where(scores < t_bg_1)[0].numpy() 375 | obj_blocks = [blocks[blocks_index]] 376 | bg_blocks = blocks[blocks_index_bg] 377 | region_mask, bg_mask = self.generate_region_mask(regions) 378 | img = self.make_img(region_mask, bg_mask, obj_blocks, bg_blocks, recalibration) 379 | 380 | elif len(words) == 2: 381 | score_1 = self.base[words[0]][words[1]] 382 | score_2 = self.base[words[1]][words[0]] 383 | blocks = self.base['blocks'] 384 | # for class 1 : 385 | blocks_index_1 = intersection(torch.where(score_1 > t_pos_2)[0].numpy(), torch.where(score_2 < t_neg_2)[0].numpy()) 386 | # for class 2 : 387 | blocks_index_2 = intersection(torch.where(score_2 > t_pos_2)[0].numpy(), torch.where(score_1 < t_neg_2)[0].numpy()) 388 | # for background : 389 | blocks_index_bg = intersection(torch.where(score_1 < t_bg_2)[0].numpy(), torch.where(score_2 < t_bg_2)[0].numpy()) 390 | obj_blocks = [blocks[blocks_index_1], blocks[blocks_index_2]] 391 | bg_blocks = blocks[blocks_index_bg] 392 | region_mask, bg_mask = self.generate_region_mask(regions) 393 | img = self.make_img(region_mask, bg_mask, obj_blocks, bg_blocks, recalibration) 394 | 395 | elif len(words) == 3: 396 | score_1_2 = self.base[words[0]][words[1]] 397 | score_1_3 = self.base[words[0]][words[2]] 398 | # for class 1 : 399 | blocks_index_1 = intersection(torch.where(score_1_2 > t_pos_2)[0].numpy(), torch.where(score_1_3 > t_pos_2)[0].numpy()) 400 | 401 | score_2_1 = self.base[words[1]][words[0]] 402 | score_2_3 = self.base[words[1]][words[2]] 403 | # for class 2 : 404 | blocks_index_2 = intersection(torch.where(score_2_1 > t_pos_2)[0].numpy(), torch.where(score_2_3 > t_pos_2)[0].numpy()) 405 | 406 | score_3_1 = self.base[words[2]][words[0]] 407 | score_3_2 = self.base[words[2]][words[1]] 408 | # for class 2 : 409 | blocks_index_3 = intersection(torch.where(score_3_1 > t_pos_2)[0].numpy(), torch.where(score_3_2 > t_pos_2)[0].numpy()) 410 | 411 | blocks = self.base['blocks'] 412 | score_1 = self.base[words[0]]['average'] 413 | score_2 = self.base[words[1]]['average'] 414 | score_3 = self.base[words[2]]['average'] 415 | 416 | # for background : 417 | blocks_index_bg = intersection(torch.where(score_1 < t_bg_2)[0].numpy(), torch.where(score_2 < t_bg_2)[0].numpy()) 418 | blocks_index_bg = intersection(blocks_index_bg, torch.where(score_3 < t_bg_2)[0].numpy()) 419 | 420 | obj_blocks = [blocks[blocks_index_1], blocks[blocks_index_2], blocks[blocks_index_3]] 421 | bg_blocks = blocks[blocks_index_bg] 422 | region_mask, bg_mask = self.generate_region_mask(regions) 423 | img = self.make_img(region_mask, bg_mask, obj_blocks, bg_blocks, recalibration) 424 | return img, region_mask -------------------------------------------------------------------------------- /ldm/modules/x_transformer.py: -------------------------------------------------------------------------------- 1 | """shout-out to https://github.com/lucidrains/x-transformers/tree/main/x_transformers""" 2 | import torch 3 | from torch import nn, einsum 4 | import torch.nn.functional as F 5 | from functools import partial 6 | from inspect import isfunction 7 | from collections import namedtuple 8 | from einops import rearrange, repeat, reduce 9 | 10 | # constants 11 | 12 | DEFAULT_DIM_HEAD = 64 13 | 14 | Intermediates = namedtuple('Intermediates', [ 15 | 'pre_softmax_attn', 16 | 'post_softmax_attn' 17 | ]) 18 | 19 | LayerIntermediates = namedtuple('Intermediates', [ 20 | 'hiddens', 21 | 'attn_intermediates' 22 | ]) 23 | 24 | 25 | class AbsolutePositionalEmbedding(nn.Module): 26 | def __init__(self, dim, max_seq_len): 27 | super().__init__() 28 | self.emb = nn.Embedding(max_seq_len, dim) 29 | self.init_() 30 | 31 | def init_(self): 32 | nn.init.normal_(self.emb.weight, std=0.02) 33 | 34 | def forward(self, x): 35 | n = torch.arange(x.shape[1], device=x.device) 36 | return self.emb(n)[None, :, :] 37 | 38 | 39 | class FixedPositionalEmbedding(nn.Module): 40 | def __init__(self, dim): 41 | super().__init__() 42 | inv_freq = 1. / (10000 ** (torch.arange(0, dim, 2).float() / dim)) 43 | self.register_buffer('inv_freq', inv_freq) 44 | 45 | def forward(self, x, seq_dim=1, offset=0): 46 | t = torch.arange(x.shape[seq_dim], device=x.device).type_as(self.inv_freq) + offset 47 | sinusoid_inp = torch.einsum('i , j -> i j', t, self.inv_freq) 48 | emb = torch.cat((sinusoid_inp.sin(), sinusoid_inp.cos()), dim=-1) 49 | return emb[None, :, :] 50 | 51 | 52 | # helpers 53 | 54 | def exists(val): 55 | return val is not None 56 | 57 | 58 | def default(val, d): 59 | if exists(val): 60 | return val 61 | return d() if isfunction(d) else d 62 | 63 | 64 | def always(val): 65 | def inner(*args, **kwargs): 66 | return val 67 | return inner 68 | 69 | 70 | def not_equals(val): 71 | def inner(x): 72 | return x != val 73 | return inner 74 | 75 | 76 | def equals(val): 77 | def inner(x): 78 | return x == val 79 | return inner 80 | 81 | 82 | def max_neg_value(tensor): 83 | return -torch.finfo(tensor.dtype).max 84 | 85 | 86 | # keyword argument helpers 87 | 88 | def pick_and_pop(keys, d): 89 | values = list(map(lambda key: d.pop(key), keys)) 90 | return dict(zip(keys, values)) 91 | 92 | 93 | def group_dict_by_key(cond, d): 94 | return_val = [dict(), dict()] 95 | for key in d.keys(): 96 | match = bool(cond(key)) 97 | ind = int(not match) 98 | return_val[ind][key] = d[key] 99 | return (*return_val,) 100 | 101 | 102 | def string_begins_with(prefix, str): 103 | return str.startswith(prefix) 104 | 105 | 106 | def group_by_key_prefix(prefix, d): 107 | return group_dict_by_key(partial(string_begins_with, prefix), d) 108 | 109 | 110 | def groupby_prefix_and_trim(prefix, d): 111 | kwargs_with_prefix, kwargs = group_dict_by_key(partial(string_begins_with, prefix), d) 112 | kwargs_without_prefix = dict(map(lambda x: (x[0][len(prefix):], x[1]), tuple(kwargs_with_prefix.items()))) 113 | return kwargs_without_prefix, kwargs 114 | 115 | 116 | # classes 117 | class Scale(nn.Module): 118 | def __init__(self, value, fn): 119 | super().__init__() 120 | self.value = value 121 | self.fn = fn 122 | 123 | def forward(self, x, **kwargs): 124 | x, *rest = self.fn(x, **kwargs) 125 | return (x * self.value, *rest) 126 | 127 | 128 | class Rezero(nn.Module): 129 | def __init__(self, fn): 130 | super().__init__() 131 | self.fn = fn 132 | self.g = nn.Parameter(torch.zeros(1)) 133 | 134 | def forward(self, x, **kwargs): 135 | x, *rest = self.fn(x, **kwargs) 136 | return (x * self.g, *rest) 137 | 138 | 139 | class ScaleNorm(nn.Module): 140 | def __init__(self, dim, eps=1e-5): 141 | super().__init__() 142 | self.scale = dim ** -0.5 143 | self.eps = eps 144 | self.g = nn.Parameter(torch.ones(1)) 145 | 146 | def forward(self, x): 147 | norm = torch.norm(x, dim=-1, keepdim=True) * self.scale 148 | return x / norm.clamp(min=self.eps) * self.g 149 | 150 | 151 | class RMSNorm(nn.Module): 152 | def __init__(self, dim, eps=1e-8): 153 | super().__init__() 154 | self.scale = dim ** -0.5 155 | self.eps = eps 156 | self.g = nn.Parameter(torch.ones(dim)) 157 | 158 | def forward(self, x): 159 | norm = torch.norm(x, dim=-1, keepdim=True) * self.scale 160 | return x / norm.clamp(min=self.eps) * self.g 161 | 162 | 163 | class Residual(nn.Module): 164 | def forward(self, x, residual): 165 | return x + residual 166 | 167 | 168 | class GRUGating(nn.Module): 169 | def __init__(self, dim): 170 | super().__init__() 171 | self.gru = nn.GRUCell(dim, dim) 172 | 173 | def forward(self, x, residual): 174 | gated_output = self.gru( 175 | rearrange(x, 'b n d -> (b n) d'), 176 | rearrange(residual, 'b n d -> (b n) d') 177 | ) 178 | 179 | return gated_output.reshape_as(x) 180 | 181 | 182 | # feedforward 183 | 184 | class GEGLU(nn.Module): 185 | def __init__(self, dim_in, dim_out): 186 | super().__init__() 187 | self.proj = nn.Linear(dim_in, dim_out * 2) 188 | 189 | def forward(self, x): 190 | x, gate = self.proj(x).chunk(2, dim=-1) 191 | return x * F.gelu(gate) 192 | 193 | 194 | class FeedForward(nn.Module): 195 | def __init__(self, dim, dim_out=None, mult=4, glu=False, dropout=0.): 196 | super().__init__() 197 | inner_dim = int(dim * mult) 198 | dim_out = default(dim_out, dim) 199 | project_in = nn.Sequential( 200 | nn.Linear(dim, inner_dim), 201 | nn.GELU() 202 | ) if not glu else GEGLU(dim, inner_dim) 203 | 204 | self.net = nn.Sequential( 205 | project_in, 206 | nn.Dropout(dropout), 207 | nn.Linear(inner_dim, dim_out) 208 | ) 209 | 210 | def forward(self, x): 211 | return self.net(x) 212 | 213 | 214 | # attention. 215 | class Attention(nn.Module): 216 | def __init__( 217 | self, 218 | dim, 219 | dim_head=DEFAULT_DIM_HEAD, 220 | heads=8, 221 | causal=False, 222 | mask=None, 223 | talking_heads=False, 224 | sparse_topk=None, 225 | use_entmax15=False, 226 | num_mem_kv=0, 227 | dropout=0., 228 | on_attn=False 229 | ): 230 | super().__init__() 231 | if use_entmax15: 232 | raise NotImplementedError("Check out entmax activation instead of softmax activation!") 233 | self.scale = dim_head ** -0.5 234 | self.heads = heads 235 | self.causal = causal 236 | self.mask = mask 237 | 238 | inner_dim = dim_head * heads 239 | 240 | self.to_q = nn.Linear(dim, inner_dim, bias=False) 241 | self.to_k = nn.Linear(dim, inner_dim, bias=False) 242 | self.to_v = nn.Linear(dim, inner_dim, bias=False) 243 | self.dropout = nn.Dropout(dropout) 244 | 245 | # talking heads 246 | self.talking_heads = talking_heads 247 | if talking_heads: 248 | self.pre_softmax_proj = nn.Parameter(torch.randn(heads, heads)) 249 | self.post_softmax_proj = nn.Parameter(torch.randn(heads, heads)) 250 | 251 | # explicit topk sparse attention 252 | self.sparse_topk = sparse_topk 253 | 254 | # entmax 255 | #self.attn_fn = entmax15 if use_entmax15 else F.softmax 256 | self.attn_fn = F.softmax 257 | 258 | # add memory key / values 259 | self.num_mem_kv = num_mem_kv 260 | if num_mem_kv > 0: 261 | self.mem_k = nn.Parameter(torch.randn(heads, num_mem_kv, dim_head)) 262 | self.mem_v = nn.Parameter(torch.randn(heads, num_mem_kv, dim_head)) 263 | 264 | # attention on attention 265 | self.attn_on_attn = on_attn 266 | self.to_out = nn.Sequential(nn.Linear(inner_dim, dim * 2), nn.GLU()) if on_attn else nn.Linear(inner_dim, dim) 267 | 268 | def forward( 269 | self, 270 | x, 271 | context=None, 272 | mask=None, 273 | context_mask=None, 274 | rel_pos=None, 275 | sinusoidal_emb=None, 276 | prev_attn=None, 277 | mem=None 278 | ): 279 | b, n, _, h, talking_heads, device = *x.shape, self.heads, self.talking_heads, x.device 280 | kv_input = default(context, x) 281 | 282 | q_input = x 283 | k_input = kv_input 284 | v_input = kv_input 285 | 286 | if exists(mem): 287 | k_input = torch.cat((mem, k_input), dim=-2) 288 | v_input = torch.cat((mem, v_input), dim=-2) 289 | 290 | if exists(sinusoidal_emb): 291 | # in shortformer, the query would start at a position offset depending on the past cached memory 292 | offset = k_input.shape[-2] - q_input.shape[-2] 293 | q_input = q_input + sinusoidal_emb(q_input, offset=offset) 294 | k_input = k_input + sinusoidal_emb(k_input) 295 | 296 | q = self.to_q(q_input) 297 | k = self.to_k(k_input) 298 | v = self.to_v(v_input) 299 | 300 | q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> b h n d', h=h), (q, k, v)) 301 | 302 | input_mask = None 303 | if any(map(exists, (mask, context_mask))): 304 | q_mask = default(mask, lambda: torch.ones((b, n), device=device).bool()) 305 | k_mask = q_mask if not exists(context) else context_mask 306 | k_mask = default(k_mask, lambda: torch.ones((b, k.shape[-2]), device=device).bool()) 307 | q_mask = rearrange(q_mask, 'b i -> b () i ()') 308 | k_mask = rearrange(k_mask, 'b j -> b () () j') 309 | input_mask = q_mask * k_mask 310 | 311 | if self.num_mem_kv > 0: 312 | mem_k, mem_v = map(lambda t: repeat(t, 'h n d -> b h n d', b=b), (self.mem_k, self.mem_v)) 313 | k = torch.cat((mem_k, k), dim=-2) 314 | v = torch.cat((mem_v, v), dim=-2) 315 | if exists(input_mask): 316 | input_mask = F.pad(input_mask, (self.num_mem_kv, 0), value=True) 317 | 318 | dots = einsum('b h i d, b h j d -> b h i j', q, k) * self.scale 319 | mask_value = max_neg_value(dots) 320 | 321 | if exists(prev_attn): 322 | dots = dots + prev_attn 323 | 324 | pre_softmax_attn = dots 325 | 326 | if talking_heads: 327 | dots = einsum('b h i j, h k -> b k i j', dots, self.pre_softmax_proj).contiguous() 328 | 329 | if exists(rel_pos): 330 | dots = rel_pos(dots) 331 | 332 | if exists(input_mask): 333 | dots.masked_fill_(~input_mask, mask_value) 334 | del input_mask 335 | 336 | if self.causal: 337 | i, j = dots.shape[-2:] 338 | r = torch.arange(i, device=device) 339 | mask = rearrange(r, 'i -> () () i ()') < rearrange(r, 'j -> () () () j') 340 | mask = F.pad(mask, (j - i, 0), value=False) 341 | dots.masked_fill_(mask, mask_value) 342 | del mask 343 | 344 | if exists(self.sparse_topk) and self.sparse_topk < dots.shape[-1]: 345 | top, _ = dots.topk(self.sparse_topk, dim=-1) 346 | vk = top[..., -1].unsqueeze(-1).expand_as(dots) 347 | mask = dots < vk 348 | dots.masked_fill_(mask, mask_value) 349 | del mask 350 | 351 | attn = self.attn_fn(dots, dim=-1) 352 | post_softmax_attn = attn 353 | 354 | attn = self.dropout(attn) 355 | 356 | if talking_heads: 357 | attn = einsum('b h i j, h k -> b k i j', attn, self.post_softmax_proj).contiguous() 358 | 359 | out = einsum('b h i j, b h j d -> b h i d', attn, v) 360 | out = rearrange(out, 'b h n d -> b n (h d)') 361 | 362 | intermediates = Intermediates( 363 | pre_softmax_attn=pre_softmax_attn, 364 | post_softmax_attn=post_softmax_attn 365 | ) 366 | 367 | return self.to_out(out), intermediates 368 | 369 | 370 | class AttentionLayers(nn.Module): 371 | def __init__( 372 | self, 373 | dim, 374 | depth, 375 | heads=8, 376 | causal=False, 377 | cross_attend=False, 378 | only_cross=False, 379 | use_scalenorm=False, 380 | use_rmsnorm=False, 381 | use_rezero=False, 382 | rel_pos_num_buckets=32, 383 | rel_pos_max_distance=128, 384 | position_infused_attn=False, 385 | custom_layers=None, 386 | sandwich_coef=None, 387 | par_ratio=None, 388 | residual_attn=False, 389 | cross_residual_attn=False, 390 | macaron=False, 391 | pre_norm=True, 392 | gate_residual=False, 393 | **kwargs 394 | ): 395 | super().__init__() 396 | ff_kwargs, kwargs = groupby_prefix_and_trim('ff_', kwargs) 397 | attn_kwargs, _ = groupby_prefix_and_trim('attn_', kwargs) 398 | 399 | dim_head = attn_kwargs.get('dim_head', DEFAULT_DIM_HEAD) 400 | 401 | self.dim = dim 402 | self.depth = depth 403 | self.layers = nn.ModuleList([]) 404 | 405 | self.has_pos_emb = position_infused_attn 406 | self.pia_pos_emb = FixedPositionalEmbedding(dim) if position_infused_attn else None 407 | self.rotary_pos_emb = always(None) 408 | 409 | assert rel_pos_num_buckets <= rel_pos_max_distance, 'number of relative position buckets must be less than the relative position max distance' 410 | self.rel_pos = None 411 | 412 | self.pre_norm = pre_norm 413 | 414 | self.residual_attn = residual_attn 415 | self.cross_residual_attn = cross_residual_attn 416 | 417 | norm_class = ScaleNorm if use_scalenorm else nn.LayerNorm 418 | norm_class = RMSNorm if use_rmsnorm else norm_class 419 | norm_fn = partial(norm_class, dim) 420 | 421 | norm_fn = nn.Identity if use_rezero else norm_fn 422 | branch_fn = Rezero if use_rezero else None 423 | 424 | if cross_attend and not only_cross: 425 | default_block = ('a', 'c', 'f') 426 | elif cross_attend and only_cross: 427 | default_block = ('c', 'f') 428 | else: 429 | default_block = ('a', 'f') 430 | 431 | if macaron: 432 | default_block = ('f',) + default_block 433 | 434 | if exists(custom_layers): 435 | layer_types = custom_layers 436 | elif exists(par_ratio): 437 | par_depth = depth * len(default_block) 438 | assert 1 < par_ratio <= par_depth, 'par ratio out of range' 439 | default_block = tuple(filter(not_equals('f'), default_block)) 440 | par_attn = par_depth // par_ratio 441 | depth_cut = par_depth * 2 // 3 # 2 / 3 attention layer cutoff suggested by PAR paper 442 | par_width = (depth_cut + depth_cut // par_attn) // par_attn 443 | assert len(default_block) <= par_width, 'default block is too large for par_ratio' 444 | par_block = default_block + ('f',) * (par_width - len(default_block)) 445 | par_head = par_block * par_attn 446 | layer_types = par_head + ('f',) * (par_depth - len(par_head)) 447 | elif exists(sandwich_coef): 448 | assert sandwich_coef > 0 and sandwich_coef <= depth, 'sandwich coefficient should be less than the depth' 449 | layer_types = ('a',) * sandwich_coef + default_block * (depth - sandwich_coef) + ('f',) * sandwich_coef 450 | else: 451 | layer_types = default_block * depth 452 | 453 | self.layer_types = layer_types 454 | self.num_attn_layers = len(list(filter(equals('a'), layer_types))) 455 | 456 | for layer_type in self.layer_types: 457 | if layer_type == 'a': 458 | layer = Attention(dim, heads=heads, causal=causal, **attn_kwargs) 459 | elif layer_type == 'c': 460 | layer = Attention(dim, heads=heads, **attn_kwargs) 461 | elif layer_type == 'f': 462 | layer = FeedForward(dim, **ff_kwargs) 463 | layer = layer if not macaron else Scale(0.5, layer) 464 | else: 465 | raise Exception(f'invalid layer type {layer_type}') 466 | 467 | if isinstance(layer, Attention) and exists(branch_fn): 468 | layer = branch_fn(layer) 469 | 470 | if gate_residual: 471 | residual_fn = GRUGating(dim) 472 | else: 473 | residual_fn = Residual() 474 | 475 | self.layers.append(nn.ModuleList([ 476 | norm_fn(), 477 | layer, 478 | residual_fn 479 | ])) 480 | 481 | def forward( 482 | self, 483 | x, 484 | context=None, 485 | mask=None, 486 | context_mask=None, 487 | mems=None, 488 | return_hiddens=False 489 | ): 490 | hiddens = [] 491 | intermediates = [] 492 | prev_attn = None 493 | prev_cross_attn = None 494 | 495 | mems = mems.copy() if exists(mems) else [None] * self.num_attn_layers 496 | 497 | for ind, (layer_type, (norm, block, residual_fn)) in enumerate(zip(self.layer_types, self.layers)): 498 | is_last = ind == (len(self.layers) - 1) 499 | 500 | if layer_type == 'a': 501 | hiddens.append(x) 502 | layer_mem = mems.pop(0) 503 | 504 | residual = x 505 | 506 | if self.pre_norm: 507 | x = norm(x) 508 | 509 | if layer_type == 'a': 510 | out, inter = block(x, mask=mask, sinusoidal_emb=self.pia_pos_emb, rel_pos=self.rel_pos, 511 | prev_attn=prev_attn, mem=layer_mem) 512 | elif layer_type == 'c': 513 | out, inter = block(x, context=context, mask=mask, context_mask=context_mask, prev_attn=prev_cross_attn) 514 | elif layer_type == 'f': 515 | out = block(x) 516 | 517 | x = residual_fn(out, residual) 518 | 519 | if layer_type in ('a', 'c'): 520 | intermediates.append(inter) 521 | 522 | if layer_type == 'a' and self.residual_attn: 523 | prev_attn = inter.pre_softmax_attn 524 | elif layer_type == 'c' and self.cross_residual_attn: 525 | prev_cross_attn = inter.pre_softmax_attn 526 | 527 | if not self.pre_norm and not is_last: 528 | x = norm(x) 529 | 530 | if return_hiddens: 531 | intermediates = LayerIntermediates( 532 | hiddens=hiddens, 533 | attn_intermediates=intermediates 534 | ) 535 | 536 | return x, intermediates 537 | 538 | return x 539 | 540 | 541 | class Encoder(AttentionLayers): 542 | def __init__(self, **kwargs): 543 | assert 'causal' not in kwargs, 'cannot set causality on encoder' 544 | super().__init__(causal=False, **kwargs) 545 | 546 | 547 | 548 | class TransformerWrapper(nn.Module): 549 | def __init__( 550 | self, 551 | *, 552 | num_tokens, 553 | max_seq_len, 554 | attn_layers, 555 | emb_dim=None, 556 | max_mem_len=0., 557 | emb_dropout=0., 558 | num_memory_tokens=None, 559 | tie_embedding=False, 560 | use_pos_emb=True 561 | ): 562 | super().__init__() 563 | assert isinstance(attn_layers, AttentionLayers), 'attention layers must be one of Encoder or Decoder' 564 | 565 | dim = attn_layers.dim 566 | emb_dim = default(emb_dim, dim) 567 | 568 | self.max_seq_len = max_seq_len 569 | self.max_mem_len = max_mem_len 570 | self.num_tokens = num_tokens 571 | 572 | self.token_emb = nn.Embedding(num_tokens, emb_dim) 573 | self.pos_emb = AbsolutePositionalEmbedding(emb_dim, max_seq_len) if ( 574 | use_pos_emb and not attn_layers.has_pos_emb) else always(0) 575 | self.emb_dropout = nn.Dropout(emb_dropout) 576 | 577 | self.project_emb = nn.Linear(emb_dim, dim) if emb_dim != dim else nn.Identity() 578 | self.attn_layers = attn_layers 579 | self.norm = nn.LayerNorm(dim) 580 | 581 | self.init_() 582 | 583 | self.to_logits = nn.Linear(dim, num_tokens) if not tie_embedding else lambda t: t @ self.token_emb.weight.t() 584 | 585 | # memory tokens (like [cls]) from Memory Transformers paper 586 | num_memory_tokens = default(num_memory_tokens, 0) 587 | self.num_memory_tokens = num_memory_tokens 588 | if num_memory_tokens > 0: 589 | self.memory_tokens = nn.Parameter(torch.randn(num_memory_tokens, dim)) 590 | 591 | # let funnel encoder know number of memory tokens, if specified 592 | if hasattr(attn_layers, 'num_memory_tokens'): 593 | attn_layers.num_memory_tokens = num_memory_tokens 594 | 595 | def init_(self): 596 | nn.init.normal_(self.token_emb.weight, std=0.02) 597 | 598 | def forward( 599 | self, 600 | x, 601 | return_embeddings=False, 602 | mask=None, 603 | return_mems=False, 604 | return_attn=False, 605 | mems=None, 606 | **kwargs 607 | ): 608 | b, n, device, num_mem = *x.shape, x.device, self.num_memory_tokens 609 | x = self.token_emb(x) 610 | x += self.pos_emb(x) 611 | x = self.emb_dropout(x) 612 | 613 | x = self.project_emb(x) 614 | 615 | if num_mem > 0: 616 | mem = repeat(self.memory_tokens, 'n d -> b n d', b=b) 617 | x = torch.cat((mem, x), dim=1) 618 | 619 | # auto-handle masking after appending memory tokens 620 | if exists(mask): 621 | mask = F.pad(mask, (num_mem, 0), value=True) 622 | 623 | x, intermediates = self.attn_layers(x, mask=mask, mems=mems, return_hiddens=True, **kwargs) 624 | x = self.norm(x) 625 | 626 | mem, x = x[:, :num_mem], x[:, num_mem:] 627 | 628 | out = self.to_logits(x) if not return_embeddings else x 629 | 630 | if return_mems: 631 | hiddens = intermediates.hiddens 632 | new_mems = list(map(lambda pair: torch.cat(pair, dim=-2), zip(mems, hiddens))) if exists(mems) else hiddens 633 | new_mems = list(map(lambda t: t[..., -self.max_mem_len:, :].detach(), new_mems)) 634 | return out, new_mems 635 | 636 | if return_attn: 637 | attn_maps = list(map(lambda t: t.post_softmax_attn, intermediates.attn_intermediates)) 638 | return out, attn_maps 639 | 640 | return out 641 | 642 | --------------------------------------------------------------------------------