Spaces:
Sleeping
Sleeping
import torch | |
import math | |
import warnings | |
from torch.nn.init import _calculate_fan_in_and_fan_out | |
def _no_grad_trunc_normal_(tensor, mean, std, a, b): | |
# Cut & paste from PyTorch official master until it's in a few official releases - RW | |
# Method based on https://people.sc.fsu.edu/~jburkardt/presentations/truncated_normal.pdf | |
def norm_cdf(x): | |
# Computes standard normal cumulative distribution function | |
return (1. + math.erf(x / math.sqrt(2.))) / 2. | |
if (mean < a - 2 * std) or (mean > b + 2 * std): | |
warnings.warn("mean is more than 2 std from [a, b] in nn.init.trunc_normal_. " | |
"The distribution of values may be incorrect.", | |
stacklevel=2) | |
with torch.no_grad(): | |
# Values are generated by using a truncated uniform distribution and | |
# then using the inverse CDF for the normal distribution. | |
# Get upper and lower cdf values | |
l = norm_cdf((a - mean) / std) | |
u = norm_cdf((b - mean) / std) | |
# Uniformly fill tensor with values from [l, u], then translate to | |
# [2l-1, 2u-1]. | |
tensor.uniform_(2 * l - 1, 2 * u - 1) | |
# Use inverse cdf transform for normal distribution to get truncated | |
# standard normal | |
tensor.erfinv_() | |
# Transform to proper mean, std | |
tensor.mul_(std * math.sqrt(2.)) | |
tensor.add_(mean) | |
# Clamp to ensure it's in the proper range | |
tensor.clamp_(min=a, max=b) | |
return tensor | |
def trunc_normal_(tensor, mean=0., std=1., a=-2., b=2.): | |
# type: (Tensor, float, float, float, float) -> Tensor | |
r"""Fills the input Tensor with values drawn from a truncated | |
normal distribution. The values are effectively drawn from the | |
normal distribution :math:`\mathcal{N}(\text{mean}, \text{std}^2)` | |
with values outside :math:`[a, b]` redrawn until they are within | |
the bounds. The method used for generating the random values works | |
best when :math:`a \leq \text{mean} \leq b`. | |
Args: | |
tensor: an n-dimensional `torch.Tensor` | |
mean: the mean of the normal distribution | |
std: the standard deviation of the normal distribution | |
a: the minimum cutoff value | |
b: the maximum cutoff value | |
Examples: | |
>>> w = torch.empty(3, 5) | |
>>> nn.init.trunc_normal_(w) | |
""" | |
return _no_grad_trunc_normal_(tensor, mean, std, a, b) | |
def variance_scaling_(tensor, scale=1.0, mode='fan_in', distribution='normal'): | |
fan_in, fan_out = _calculate_fan_in_and_fan_out(tensor) | |
if mode == 'fan_in': | |
denom = fan_in | |
elif mode == 'fan_out': | |
denom = fan_out | |
elif mode == 'fan_avg': | |
denom = (fan_in + fan_out) / 2 | |
variance = scale / denom | |
if distribution == "truncated_normal": | |
# constant is stddev of standard normal truncated to (-2, 2) | |
trunc_normal_(tensor, std=math.sqrt(variance) / .87962566103423978) | |
elif distribution == "normal": | |
tensor.normal_(std=math.sqrt(variance)) | |
elif distribution == "uniform": | |
bound = math.sqrt(3 * variance) | |
tensor.uniform_(-bound, bound) | |
else: | |
raise ValueError(f"invalid distribution {distribution}") | |
def lecun_normal_(tensor): | |
variance_scaling_(tensor, mode='fan_in', distribution='truncated_normal') | |