Spaces:
Runtime error
Runtime error
File size: 4,313 Bytes
3424266 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 |
# Copyright 2024 EPFL and Apple Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# DISCLAIMER: This code is strongly influenced by https://github.com/lucidrains/NWT-pytorch/
import torch
import torch.nn.functional as F
from torch import nn, einsum
from einops import rearrange, repeat
from einops.layers.torch import EinMix as Mix
class Memcodes(nn.Module):
def __init__(
self,
*,
dim,
codebook_size,
heads = 1,
temperature = 1.,
channel_last = False,
accept_image_fmap = True,
**kwargs,
):
super().__init__()
assert (dim % heads) == 0, 'dimension must be divisible by number of heads'
self.heads = heads
self.dim = dim
self.scale = (dim // heads) ** -0.5
self.temperature = temperature
self.codebook_size = codebook_size
self.accept_image_fmap = accept_image_fmap
self.channel_last = channel_last
num_codebooks = heads
codebook_dim = dim // heads
self.codes = nn.Parameter(torch.randn(num_codebooks, codebook_size, codebook_dim))
self.to_k = Mix('h n d -> h n c', weight_shape = 'h d c', h = heads, d = codebook_dim, c = codebook_dim)
self.to_v = Mix('h n d -> h n c', weight_shape = 'h d c', h = heads, d = codebook_dim, c = codebook_dim)
def indices_to_embedding(self, indices):
batch = indices.shape[0]
values = self.to_v(self.codes)
values = repeat(values, 'h n d -> b h n d', b = batch)
indices = repeat(indices, '... -> ... d', d = values.shape[-1]).squeeze(2)
if self.accept_image_fmap and len(indices.size())==4:
out = values.gather(2, indices)
out = rearrange(out, 'b h n d -> b (h d) n 1')
return out
else:
out = values.gather(2, indices.unsqueeze(2))
return rearrange(out, 'b h n d -> b n (h d)')
def forward(self, x):
need_transpose = not self.channel_last and not self.accept_image_fmap
if self.accept_image_fmap:
height, width = x.shape[-2:]
x = rearrange(x, 'b c h w -> b (h w) c')
if need_transpose:
x = rearrange(x, 'b d n -> b n d')
assert x.shape[-1] == self.dim
# split out heads
q = rearrange(x, 'b n (h d) -> b h n d', h = self.heads)
q = q * self.scale
# get key / values of codes
k, v = self.to_k(self.codes), self.to_v(self.codes)
# straight through gumbel softmax
logits = einsum('b h i d, h j d -> b h i j', q, k)
if self.training:
attn = F.gumbel_softmax(logits, tau = self.temperature, dim = -1, hard = True)
codebook_indices = attn.argmax(dim = -1)
else:
codebook_indices = logits.argmax(dim = -1)
attn = F.one_hot(codebook_indices, num_classes = self.codebook_size).float()
if self.heads == 1:
codebook_indices = codebook_indices.squeeze(1)
out = einsum('b h i j, h j d -> b h i d', attn, v)
# merge heads
out = rearrange(out, 'b h n d -> b n (h d)')
if need_transpose:
out = rearrange(out, 'b n d -> b d n')
if self.accept_image_fmap:
out = rearrange(out, 'b (h w) c -> b c h w', h = height, w = width)
if self.heads == 1:
codebook_indices = rearrange(codebook_indices, 'b (h w) -> b h w', h = height, w = width)
else:
codebook_indices = rearrange(codebook_indices, 'b n (h w) -> b n h w', h = height, w = width)
# Dummy codebook loss for compatibility with other types of quantizers
codebook_loss = torch.tensor([0.], device=x.device)
return out, codebook_loss, codebook_indices
|