diff options
Diffstat (limited to 'text_recognizer/networks')
-rw-r--r-- | text_recognizer/networks/quantizer/__init__.py | 0 | ||||
-rw-r--r-- | text_recognizer/networks/quantizer/codebook.py | 96 | ||||
-rw-r--r-- | text_recognizer/networks/quantizer/kmeans.py | 32 | ||||
-rw-r--r-- | text_recognizer/networks/quantizer/quantizer.py | 59 | ||||
-rw-r--r-- | text_recognizer/networks/quantizer/utils.py | 26 | ||||
-rw-r--r-- | text_recognizer/networks/vq_transformer.py | 84 | ||||
-rw-r--r-- | text_recognizer/networks/vqvae/__init__.py | 1 | ||||
-rw-r--r-- | text_recognizer/networks/vqvae/decoder.py | 93 | ||||
-rw-r--r-- | text_recognizer/networks/vqvae/encoder.py | 85 | ||||
-rw-r--r-- | text_recognizer/networks/vqvae/norm.py | 24 | ||||
-rw-r--r-- | text_recognizer/networks/vqvae/residual.py | 54 | ||||
-rw-r--r-- | text_recognizer/networks/vqvae/resize.py | 19 | ||||
-rw-r--r-- | text_recognizer/networks/vqvae/vqvae.py | 42 |
13 files changed, 0 insertions, 615 deletions
diff --git a/text_recognizer/networks/quantizer/__init__.py b/text_recognizer/networks/quantizer/__init__.py deleted file mode 100644 index e69de29..0000000 --- a/text_recognizer/networks/quantizer/__init__.py +++ /dev/null diff --git a/text_recognizer/networks/quantizer/codebook.py b/text_recognizer/networks/quantizer/codebook.py deleted file mode 100644 index cb9bc59..0000000 --- a/text_recognizer/networks/quantizer/codebook.py +++ /dev/null @@ -1,96 +0,0 @@ -"""Codebook module.""" -from typing import Tuple - -import attr -from einops import rearrange -import torch -from torch import nn, Tensor -import torch.nn.functional as F - -from text_recognizer.networks.quantizer.kmeans import kmeans -from text_recognizer.networks.quantizer.utils import ( - ema_inplace, - norm, - sample_vectors, -) - - -@attr.s(eq=False) -class CosineSimilarityCodebook(nn.Module): - """Cosine similarity codebook.""" - - dim: int = attr.ib() - codebook_size: int = attr.ib() - kmeans_init: bool = attr.ib(default=False) - kmeans_iters: int = attr.ib(default=10) - decay: float = attr.ib(default=0.8) - eps: float = attr.ib(default=1.0e-5) - threshold_dead: int = attr.ib(default=2) - - def __attrs_pre_init__(self) -> None: - super().__init__() - - def __attrs_post_init__(self) -> None: - if not self.kmeans_init: - embeddings = norm(torch.randn(self.codebook_size, self.dim)) - else: - embeddings = torch.zeros(self.codebook_size, self.dim) - self.register_buffer("initalized", Tensor([not self.kmeans_init])) - self.register_buffer("cluster_size", torch.zeros(self.codebook_size)) - self.register_buffer("embeddings", embeddings) - - def _initalize_embedding(self, data: Tensor) -> None: - embeddings, cluster_size = kmeans(data, self.codebook_size, self.kmeans_iters) - self.embeddings.data.copy_(embeddings) - self.cluster_size.data.copy_(cluster_size) - self.initalized.data.copy_(Tensor([True])) - - def _replace(self, samples: Tensor, mask: Tensor) -> None: - samples = norm(samples) - modified_codebook = torch.where( - mask[..., None], - sample_vectors(samples, self.codebook_size), - self.embeddings, - ) - self.embeddings.data.copy_(modified_codebook) - - def _replace_dead_codes(self, batch_samples: Tensor) -> None: - if self.threshold_dead == 0: - return - dead_codes = self.cluster_size < self.threshold_dead - if not torch.any(dead_codes): - return - batch_samples = rearrange(batch_samples, "... d -> (...) d") - self._replace(batch_samples, mask=dead_codes) - - def forward(self, x: Tensor) -> Tuple[Tensor, Tensor]: - """Quantizes tensor.""" - shape = x.shape - flatten = rearrange(x, "... d -> (...) d") - flatten = norm(flatten) - - if not self.initalized: - self._initalize_embedding(flatten) - - embeddings = norm(self.embeddings) - dist = flatten @ embeddings.t() - indices = dist.max(dim=-1).indices - one_hot = F.one_hot(indices, self.codebook_size).type_as(x) - indices = indices.view(*shape[:-1]) - - quantized = F.embedding(indices, self.embeddings) - - if self.training: - bins = one_hot.sum(0) - ema_inplace(self.cluster_size, bins, self.decay) - zero_mask = bins == 0 - bins = bins.masked_fill(zero_mask, 1.0) - - embed_sum = flatten.t() @ one_hot - embed_norm = (embed_sum / bins.unsqueeze(0)).t() - embed_norm = norm(embed_norm) - embed_norm = torch.where(zero_mask[..., None], embeddings, embed_norm) - ema_inplace(self.embeddings, embed_norm, self.decay) - self._replace_dead_codes(x) - - return quantized, indices diff --git a/text_recognizer/networks/quantizer/kmeans.py b/text_recognizer/networks/quantizer/kmeans.py deleted file mode 100644 index a34c381..0000000 --- a/text_recognizer/networks/quantizer/kmeans.py +++ /dev/null @@ -1,32 +0,0 @@ -"""K-means clustering for embeddings.""" -from typing import Tuple - -from einops import repeat -import torch -from torch import Tensor - -from text_recognizer.networks.quantizer.utils import norm, sample_vectors - - -def kmeans( - samples: Tensor, num_clusters: int, num_iters: int = 10 -) -> Tuple[Tensor, Tensor]: - """Compute k-means clusters.""" - D = samples.shape[-1] - - means = sample_vectors(samples, num_clusters) - - for _ in range(num_iters): - dists = samples @ means.t() - buckets = dists.max(dim=-1).indices - bins = torch.bincount(buckets, minlength=num_clusters) - zero_mask = bins == 0 - bins_min_clamped = bins.masked_fill(zero_mask, 1) - - new_means = buckets.new_zeros(num_clusters, D).type_as(samples) - new_means.scatter_add_(0, repeat(buckets, "n -> n d", d=D), samples) - new_means /= bins_min_clamped[..., None] - new_means = norm(new_means) - means = torch.where(zero_mask[..., None], means, new_means) - - return means, bins diff --git a/text_recognizer/networks/quantizer/quantizer.py b/text_recognizer/networks/quantizer/quantizer.py deleted file mode 100644 index 3e8f0b2..0000000 --- a/text_recognizer/networks/quantizer/quantizer.py +++ /dev/null @@ -1,59 +0,0 @@ -"""Implementation of a Vector Quantized Variational AutoEncoder. - -Reference: -https://github.com/AntixK/PyTorch-VAE/blob/master/models/vq_vae.py -""" -from typing import Tuple, Type - -import attr -from einops import rearrange -import torch -from torch import nn -from torch import Tensor -import torch.nn.functional as F - - -@attr.s(eq=False) -class VectorQuantizer(nn.Module): - """Vector quantizer.""" - - input_dim: int = attr.ib() - codebook: Type[nn.Module] = attr.ib() - commitment: float = attr.ib(default=1.0) - project_in: nn.Linear = attr.ib(default=None, init=False) - project_out: nn.Linear = attr.ib(default=None, init=False) - - def __attrs_pre_init__(self) -> None: - super().__init__() - - def __attrs_post_init__(self) -> None: - require_projection = self.codebook.dim != self.input_dim - self.project_in = ( - nn.Linear(self.input_dim, self.codebook.dim) - if require_projection - else nn.Identity() - ) - self.project_out = ( - nn.Linear(self.codebook.dim, self.input_dim) - if require_projection - else nn.Identity() - ) - - def forward(self, x: Tensor) -> Tuple[Tensor, Tensor, Tensor]: - """Quantizes latent vectors.""" - H, W = x.shape[-2:] - x = rearrange(x, "b d h w -> b (h w) d") - x = self.project_in(x) - - quantized, indices = self.codebook(x) - - if self.training: - commitment_loss = F.mse_loss(quantized.detach(), x) * self.commitment - quantized = x + (quantized - x).detach() - else: - commitment_loss = torch.tensor([0.0]).type_as(x) - - quantized = self.project_out(quantized) - quantized = rearrange(quantized, "b (h w) d -> b d h w", h=H, w=W) - - return quantized, indices, commitment_loss diff --git a/text_recognizer/networks/quantizer/utils.py b/text_recognizer/networks/quantizer/utils.py deleted file mode 100644 index 0502d49..0000000 --- a/text_recognizer/networks/quantizer/utils.py +++ /dev/null @@ -1,26 +0,0 @@ -"""Helper functions for quantization.""" -from typing import Tuple - -import torch -from torch import Tensor -import torch.nn.functional as F - - -def sample_vectors(samples: Tensor, num: int) -> Tensor: - """Subsamples a set of vectors.""" - B, device = samples.shape[0], samples.device - if B >= num: - indices = torch.randperm(B, device=device)[:num] - else: - indices = torch.randint(0, B, (num,), device=device)[:num] - return samples[indices] - - -def norm(t: Tensor) -> Tensor: - """Applies L2-normalization.""" - return F.normalize(t, p=2, dim=-1) - - -def ema_inplace(moving_avg: Tensor, new: Tensor, decay: float) -> None: - """Applies exponential moving average.""" - moving_avg.data.mul_(decay).add_(new, alpha=(1 - decay)) diff --git a/text_recognizer/networks/vq_transformer.py b/text_recognizer/networks/vq_transformer.py deleted file mode 100644 index a2bd81b..0000000 --- a/text_recognizer/networks/vq_transformer.py +++ /dev/null @@ -1,84 +0,0 @@ -"""Vector quantized encoder, transformer decoder.""" -from typing import Optional, Tuple, Type - -from torch import nn, Tensor - -from text_recognizer.networks.conv_transformer import ConvTransformer -from text_recognizer.networks.quantizer.quantizer import VectorQuantizer -from text_recognizer.networks.transformer.layers import Decoder - - -class VqTransformer(ConvTransformer): - """Convolutional encoder and transformer decoder network.""" - - def __init__( - self, - input_dims: Tuple[int, int, int], - hidden_dim: int, - num_classes: int, - pad_index: Tensor, - encoder: nn.Module, - decoder: Decoder, - pixel_pos_embedding: Type[nn.Module], - quantizer: VectorQuantizer, - token_pos_embedding: Optional[Type[nn.Module]] = None, - ) -> None: - super().__init__( - input_dims=input_dims, - hidden_dim=hidden_dim, - num_classes=num_classes, - pad_index=pad_index, - encoder=encoder, - decoder=decoder, - pixel_pos_embedding=pixel_pos_embedding, - token_pos_embedding=token_pos_embedding, - ) - self.quantizer = quantizer - - def encode(self, x: Tensor) -> Tuple[Tensor, Tensor]: - """Encodes an image into a discrete (VQ) latent representation. - - Args: - x (Tensor): Image tensor. - - Shape: - - x: :math: `(B, C, H, W)` - - z: :math: `(B, Sx, E)` - - where Sx is the length of the flattened feature maps projected from - the encoder. E latent dimension for each pixel in the projected - feature maps. - - Returns: - Tensor: A Latent embedding of the image. - """ - z = self.encoder(x) - z = self.conv(z) - z, _, commitment_loss = self.quantizer(z) - z = self.pixel_pos_embedding(z) - z = z.flatten(start_dim=2) - - # Permute tensor from [B, E, Ho * Wo] to [B, Sx, E] - z = z.permute(0, 2, 1) - return z, commitment_loss - - def forward(self, x: Tensor, context: Tensor) -> Tensor: - """Encodes images into word piece logtis. - - Args: - x (Tensor): Input image(s). - context (Tensor): Target word embeddings. - - Shapes: - - x: :math: `(B, C, H, W)` - - context: :math: `(B, Sy, T)` - - where B is the batch size, C is the number of input channels, H is - the image height and W is the image width. - - Returns: - Tensor: Sequence of logits. - """ - z, commitment_loss = self.encode(x) - logits = self.decode(z, context) - return logits, commitment_loss diff --git a/text_recognizer/networks/vqvae/__init__.py b/text_recognizer/networks/vqvae/__init__.py deleted file mode 100644 index e1f05fa..0000000 --- a/text_recognizer/networks/vqvae/__init__.py +++ /dev/null @@ -1 +0,0 @@ -"""VQ-VAE module.""" diff --git a/text_recognizer/networks/vqvae/decoder.py b/text_recognizer/networks/vqvae/decoder.py deleted file mode 100644 index 7734a5a..0000000 --- a/text_recognizer/networks/vqvae/decoder.py +++ /dev/null @@ -1,93 +0,0 @@ -"""CNN decoder for the VQ-VAE.""" -from typing import Sequence - -from torch import nn -from torch import Tensor - -from text_recognizer.networks.util import activation_function -from text_recognizer.networks.vqvae.norm import Normalize -from text_recognizer.networks.vqvae.residual import Residual - - -class Decoder(nn.Module): - """A CNN encoder network.""" - - def __init__( - self, - out_channels: int, - hidden_dim: int, - channels_multipliers: Sequence[int], - dropout_rate: float, - activation: str = "mish", - use_norm: bool = False, - num_residuals: int = 4, - residual_channels: int = 32, - ) -> None: - super().__init__() - self.out_channels = out_channels - self.hidden_dim = hidden_dim - self.channels_multipliers = tuple(channels_multipliers) - self.activation = activation - self.dropout_rate = dropout_rate - self.use_norm = use_norm - self.num_residuals = num_residuals - self.residual_channels = residual_channels - self.decoder = self._build_decompression_block() - - def _build_decompression_block(self,) -> nn.Sequential: - decoder = [] - in_channels = self.hidden_dim * self.channels_multipliers[0] - for _ in range(self.num_residuals): - decoder += [ - Residual( - in_channels=in_channels, - residual_channels=self.residual_channels, - use_norm=self.use_norm, - activation=self.activation, - ), - ] - - activation_fn = activation_function(self.activation) - out_channels_multipliers = self.channels_multipliers + (1,) - num_blocks = len(self.channels_multipliers) - - for i in range(num_blocks): - in_channels = self.hidden_dim * self.channels_multipliers[i] - out_channels = self.hidden_dim * out_channels_multipliers[i + 1] - if self.use_norm: - decoder += [ - Normalize(num_channels=in_channels,), - ] - decoder += [ - activation_fn, - nn.ConvTranspose2d( - in_channels=in_channels, - out_channels=out_channels, - kernel_size=4, - stride=2, - padding=1, - ), - ] - - if self.use_norm: - decoder += [ - Normalize( - num_channels=self.hidden_dim * out_channels_multipliers[-1], - num_groups=self.hidden_dim * out_channels_multipliers[-1] // 4, - ), - ] - - decoder += [ - nn.Conv2d( - in_channels=self.hidden_dim * out_channels_multipliers[-1], - out_channels=self.out_channels, - kernel_size=3, - stride=1, - padding=1, - ), - ] - return nn.Sequential(*decoder) - - def forward(self, z_q: Tensor) -> Tensor: - """Reconstruct input from given codes.""" - return self.decoder(z_q) diff --git a/text_recognizer/networks/vqvae/encoder.py b/text_recognizer/networks/vqvae/encoder.py deleted file mode 100644 index 4761486..0000000 --- a/text_recognizer/networks/vqvae/encoder.py +++ /dev/null @@ -1,85 +0,0 @@ -"""CNN encoder for the VQ-VAE.""" -from typing import List, Tuple - -from torch import nn -from torch import Tensor - -from text_recognizer.networks.util import activation_function -from text_recognizer.networks.vqvae.norm import Normalize -from text_recognizer.networks.vqvae.residual import Residual - - -class Encoder(nn.Module): - """A CNN encoder network.""" - - def __init__( - self, - in_channels: int, - hidden_dim: int, - channels_multipliers: List[int], - dropout_rate: float, - activation: str = "mish", - use_norm: bool = False, - num_residuals: int = 4, - residual_channels: int = 32, - ) -> None: - super().__init__() - self.in_channels = in_channels - self.hidden_dim = hidden_dim - self.channels_multipliers = tuple(channels_multipliers) - self.activation = activation - self.dropout_rate = dropout_rate - self.use_norm = use_norm - self.num_residuals = num_residuals - self.residual_channels = residual_channels - self.encoder = self._build_compression_block() - - def _build_compression_block(self) -> nn.Sequential: - """Builds encoder network.""" - num_blocks = len(self.channels_multipliers) - channels_multipliers = (1,) + self.channels_multipliers - activation_fn = activation_function(self.activation) - - encoder = [ - nn.Conv2d( - in_channels=self.in_channels, - out_channels=self.hidden_dim, - kernel_size=3, - stride=1, - padding=1, - ), - ] - - for i in range(num_blocks): - in_channels = self.hidden_dim * channels_multipliers[i] - out_channels = self.hidden_dim * channels_multipliers[i + 1] - if self.use_norm: - encoder += [ - Normalize(num_channels=in_channels,), - ] - encoder += [ - activation_fn, - nn.Conv2d( - in_channels=in_channels, - out_channels=out_channels, - kernel_size=4, - stride=2, - padding=1, - ), - ] - - for _ in range(self.num_residuals): - encoder += [ - Residual( - in_channels=out_channels, - residual_channels=self.residual_channels, - use_norm=self.use_norm, - activation=self.activation, - ) - ] - - return nn.Sequential(*encoder) - - def forward(self, x: Tensor) -> Tuple[Tensor, Tensor]: - """Encodes input into a discrete representation.""" - return self.encoder(x) diff --git a/text_recognizer/networks/vqvae/norm.py b/text_recognizer/networks/vqvae/norm.py deleted file mode 100644 index d73f9f8..0000000 --- a/text_recognizer/networks/vqvae/norm.py +++ /dev/null @@ -1,24 +0,0 @@ -"""Normalizer block.""" -import attr -from torch import nn, Tensor - - -@attr.s(eq=False) -class Normalize(nn.Module): - num_channels: int = attr.ib() - num_groups: int = attr.ib(default=32) - norm: nn.GroupNorm = attr.ib(init=False) - - def __attrs_post_init__(self) -> None: - """Post init configuration.""" - super().__init__() - self.norm = nn.GroupNorm( - num_groups=self.num_groups, - num_channels=self.num_channels, - eps=1.0e-6, - affine=True, - ) - - def forward(self, x: Tensor) -> Tensor: - """Applies group normalization.""" - return self.norm(x) diff --git a/text_recognizer/networks/vqvae/residual.py b/text_recognizer/networks/vqvae/residual.py deleted file mode 100644 index bdff9eb..0000000 --- a/text_recognizer/networks/vqvae/residual.py +++ /dev/null @@ -1,54 +0,0 @@ -"""Residual block.""" -import attr -from torch import nn -from torch import Tensor - -from text_recognizer.networks.util import activation_function -from text_recognizer.networks.vqvae.norm import Normalize - - -@attr.s(eq=False) -class Residual(nn.Module): - in_channels: int = attr.ib() - residual_channels: int = attr.ib() - use_norm: bool = attr.ib(default=False) - activation: str = attr.ib(default="relu") - - def __attrs_post_init__(self) -> None: - """Post init configuration.""" - super().__init__() - self.block = self._build_res_block() - - def _build_res_block(self) -> nn.Sequential: - """Build residual block.""" - block = [] - activation_fn = activation_function(activation=self.activation) - - if self.use_norm: - block.append(Normalize(num_channels=self.in_channels)) - - block += [ - activation_fn, - nn.Conv2d( - self.in_channels, - self.residual_channels, - kernel_size=3, - padding=1, - bias=False, - ), - ] - - if self.use_norm: - block.append(Normalize(num_channels=self.residual_channels)) - - block += [ - activation_fn, - nn.Conv2d( - self.residual_channels, self.in_channels, kernel_size=1, bias=False - ), - ] - return nn.Sequential(*block) - - def forward(self, x: Tensor) -> Tensor: - """Apply the residual forward pass.""" - return x + self.block(x) diff --git a/text_recognizer/networks/vqvae/resize.py b/text_recognizer/networks/vqvae/resize.py deleted file mode 100644 index 8d67d02..0000000 --- a/text_recognizer/networks/vqvae/resize.py +++ /dev/null @@ -1,19 +0,0 @@ -"""Up and down-sample with linear interpolation.""" -from torch import nn, Tensor -import torch.nn.functional as F - - -class Upsample(nn.Module): - """Upsamples by a factor 2.""" - - def forward(self, x: Tensor) -> Tensor: - """Applies upsampling.""" - return F.interpolate(x, scale_factor=2.0, mode="nearest") - - -class Downsample(nn.Module): - """Downsampling by a factor 2.""" - - def forward(self, x: Tensor) -> Tensor: - """Applies downsampling.""" - return F.avg_pool2d(x, kernel_size=2, stride=2) diff --git a/text_recognizer/networks/vqvae/vqvae.py b/text_recognizer/networks/vqvae/vqvae.py deleted file mode 100644 index 5560e12..0000000 --- a/text_recognizer/networks/vqvae/vqvae.py +++ /dev/null @@ -1,42 +0,0 @@ -"""The VQ-VAE.""" -from typing import Tuple - -from torch import nn -from torch import Tensor - -from text_recognizer.networks.quantizer.quantizer import VectorQuantizer - - -class VQVAE(nn.Module): - """Vector Quantized Variational AutoEncoder.""" - - def __init__( - self, - encoder: nn.Module, - decoder: nn.Module, - quantizer: VectorQuantizer, - ) -> None: - super().__init__() - self.encoder = encoder - self.decoder = decoder - self.quantizer = quantizer - - def encode(self, x: Tensor) -> Tensor: - """Encodes input to a latent code.""" - return self.encoder(x) - - def quantize(self, z_e: Tensor) -> Tuple[Tensor, Tensor]: - """Quantizes the encoded latent vectors.""" - z_q, _, commitment_loss = self.quantizer(z_e) - return z_q, commitment_loss - - def decode(self, z_q: Tensor) -> Tensor: - """Reconstructs input from latent codes.""" - return self.decoder(z_q) - - def forward(self, x: Tensor) -> Tuple[Tensor, Tensor]: - """Compresses and decompresses input.""" - z_e = self.encode(x) - z_q, commitment_loss = self.quantize(z_e) - x_hat = self.decode(z_q) - return x_hat, commitment_loss |