File size: 6,334 Bytes
9b9e0ee |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 |
"""
Various utilities for neural networks.
"""
import math
import torch as th
import torch.nn as nn
import torch.nn.functional as F
class GroupNorm32(nn.GroupNorm):
def __init__(self, num_groups, num_channels, swish, eps=1e-5):
super().__init__(num_groups=num_groups, num_channels=num_channels, eps=eps)
self.swish = swish
def forward(self, x):
y = super().forward(x.float()).to(x.dtype)
if self.swish == 1.0:
y = F.silu(y)
elif self.swish:
y = y * F.sigmoid(y * float(self.swish))
return y
def conv_nd(dims, *args, **kwargs):
"""
Create a 1D, 2D, or 3D convolution module.
"""
if dims == 1:
return nn.Conv1d(*args, **kwargs)
elif dims == 2:
return nn.Conv2d(*args, **kwargs)
elif dims == 3:
return nn.Conv3d(*args, **kwargs)
raise ValueError(f"unsupported dimensions: {dims}")
def linear(*args, **kwargs):
"""
Create a linear module.
"""
return nn.Linear(*args, **kwargs)
def avg_pool_nd(dims, *args, **kwargs):
"""
Create a 1D, 2D, or 3D average pooling module.
"""
if dims == 1:
return nn.AvgPool1d(*args, **kwargs)
elif dims == 2:
return nn.AvgPool2d(*args, **kwargs)
elif dims == 3:
return nn.AvgPool3d(*args, **kwargs)
raise ValueError(f"unsupported dimensions: {dims}")
def update_ema(target_params, source_params, rate=0.99):
"""
Update target parameters to be closer to those of source parameters using
an exponential moving average.
:param target_params: the target parameter sequence.
:param source_params: the source parameter sequence.
:param rate: the EMA rate (closer to 1 means slower).
"""
for targ, src in zip(target_params, source_params):
targ.detach().mul_(rate).add_(src, alpha=1 - rate)
def zero_module(module):
"""
Zero out the parameters of a module and return it.
"""
for p in module.parameters():
p.detach().zero_()
return module
def scale_module(module, scale):
"""
Scale the parameters of a module and return it.
"""
for p in module.parameters():
p.detach().mul_(scale)
return module
def mean_flat(tensor):
"""
Take the mean over all non-batch dimensions.
"""
return tensor.mean(dim=list(range(1, len(tensor.shape))))
def normalization(channels, swish=0.0):
"""
Make a standard normalization layer, with an optional swish activation.
:param channels: number of input channels.
:return: an nn.Module for normalization.
"""
return GroupNorm32(num_channels=channels, num_groups=32, swish=swish)
# def timestep_embedding(timesteps, dim, max_period=10000):
# """
# Create sinusoidal timestep embeddings.
# :param timesteps: a 1-D Tensor of N indices, one per batch element.
# These may be fractional.
# :param dim: the dimension of the output.
# :param max_period: controls the minimum frequency of the embeddings.
# :return: an [N x dim] Tensor of positional embeddings.
# """
# half = dim // 2
# freqs = th.exp(
# -math.log(max_period) * th.arange(start=0, end=half, dtype=th.float32) / half
# ).to(device=timesteps.device)
# args = timesteps[:, None].float() * freqs[None]
# embedding = th.cat([th.cos(args), th.sin(args)], dim=-1)
# if dim % 2:
# embedding = th.cat([embedding, th.zeros_like(embedding[:, :1])], dim=-1)
# return embedding
def timestep_embedding(timesteps, dim, max_period=10000, repeat_only=False):
"""
Create sinusoidal timestep embeddings.
:param timesteps: a 1-D Tensor of N indices, one per batch element.
These may be fractional.
:param dim: the dimension of the output.
:param max_period: controls the minimum frequency of the embeddings.
:return: an [N x dim] Tensor of positional embeddings.
"""
if not repeat_only:
half = dim // 2
freqs = th.exp(
-math.log(max_period)
* th.arange(start=0, end=half, dtype=th.float32)
/ half
).to(device=timesteps.device)
args = timesteps[:, None].float() * freqs[None]
embedding = th.cat([th.cos(args), th.sin(args)], dim=-1)
if dim % 2:
embedding = th.cat([embedding, th.zeros_like(embedding[:, :1])], dim=-1)
else:
embedding = repeat(timesteps, "b -> b d", d=dim)
return embedding
def checkpoint(func, inputs, params, flag):
"""
Evaluate a function without caching intermediate activations, allowing for
reduced memory at the expense of extra compute in the backward pass.
:param func: the function to evaluate.
:param inputs: the argument sequence to pass to `func`.
:param params: a sequence of parameters `func` depends on but does not
explicitly take as arguments.
:param flag: if False, disable gradient checkpointing.
"""
# flag = False
if flag:
args = tuple(inputs) + tuple(params)
return CheckpointFunction.apply(func, len(inputs), *args)
else:
return func(*inputs)
class CheckpointFunction(th.autograd.Function):
@staticmethod
def forward(ctx, run_function, length, *args):
ctx.run_function = run_function
ctx.input_tensors = list(args[:length])
ctx.input_params = list(args[length:])
with th.no_grad():
output_tensors = ctx.run_function(*ctx.input_tensors)
return output_tensors
@staticmethod
def backward(ctx, *output_grads):
ctx.input_tensors = [x.detach().requires_grad_(True) for x in ctx.input_tensors]
with th.enable_grad():
# Fixes a bug where the first op in run_function modifies the
# Tensor storage in place, which is not allowed for detach()'d
# Tensors.
shallow_copies = [x.view_as(x) for x in ctx.input_tensors]
output_tensors = ctx.run_function(*shallow_copies)
input_grads = th.autograd.grad(
output_tensors,
ctx.input_tensors + ctx.input_params,
output_grads,
allow_unused=True,
)
del ctx.input_tensors
del ctx.input_params
del output_tensors
return (None, None) + input_grads
|