yfzhang114
commented
1 month ago Hi,
It's a common PyTorch issue. Please refer to https://github.com/pytorch/pytorch/issues/123553. Additionally, you can try this function instead of nn.trunc_normal_:
def _trunc_normal_(tensor, mean, std, a, b):
# rewrite timm trunc normal
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)
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_() # NOTE: deleted as "erfinv_cuda" not implemented for 'BFloat16'
# 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
Luo-Z13
Hello, I encounter the error "RuntimeError: 'erfinv_cuda' not implemented for 'BFloat16'" when I try to fine-tune based on the SliME-Vicuna-7B weight. Could you please provide some suggestions? My script:
My environment:
The detail of the error information: