forked from jiuyuan/CPM-9G-8B
38 lines
1.1 KiB
Python
38 lines
1.1 KiB
Python
import bmtrain as bmt
|
|
import torch
|
|
|
|
|
|
@torch.jit.script
|
|
def rms_layernorm(hidden: torch.Tensor, weight: torch.Tensor, eps: float):
|
|
old_dtype = hidden.dtype
|
|
variance = hidden.to(torch.float32).pow(2).mean(dim=-1, keepdim=True)
|
|
hidden = (hidden * torch.rsqrt(variance + eps)).to(old_dtype)
|
|
return hidden * weight
|
|
|
|
|
|
class LayerNorm(bmt.DistributedModule):
|
|
"""RMS LayerNorm"""
|
|
|
|
def __init__(
|
|
self,
|
|
dim_norm: int,
|
|
dtype: torch.dtype = torch.half,
|
|
eps: float = 1e-5,
|
|
init_var: float = 1.0,
|
|
):
|
|
super().__init__()
|
|
|
|
self.eps = eps
|
|
self.dim_norm = dim_norm
|
|
self.weight = bmt.DistributedParameter(torch.full((dim_norm,), init_var, dtype=dtype))
|
|
|
|
def forward(self, x: torch.Tensor):
|
|
"""
|
|
Args:
|
|
x (:obj:`torch.Tensor` of shape ``(batch_size, seq_len, dim_norm)``): Input tensor that need to be normalized.
|
|
Return:
|
|
:obj:`torch.Tensor` of shape ``(batch_size, seq_len, dim_norm)``: The layernorm output.
|
|
""" # noqa: E501
|
|
assert x.size(-1) == self.dim_norm
|
|
return rms_layernorm(x, self.weight, self.eps)
|