217 lines
6.8 KiB
Python
217 lines
6.8 KiB
Python
"""Custom normalization layers."""
|
|
from typing import Optional, Tuple, Union
|
|
|
|
import torch
|
|
import torch.nn as nn
|
|
|
|
from vllm._C import ops
|
|
|
|
|
|
class RMSNorm(nn.Module):
|
|
"""Root mean square normalization.
|
|
|
|
Computes x -> w * x / sqrt(E[x^2] + eps) where w is the learned weight.
|
|
Refer to https://arxiv.org/abs/1910.07467
|
|
"""
|
|
|
|
def __init__(
|
|
self,
|
|
hidden_size: int,
|
|
eps: float = 1e-6,
|
|
) -> None:
|
|
super().__init__()
|
|
self.weight = nn.Parameter(torch.ones(hidden_size))
|
|
self.variance_epsilon = eps
|
|
|
|
def _forward(
|
|
self,
|
|
x: torch.Tensor,
|
|
residual: Optional[torch.Tensor] = None,
|
|
) -> Union[torch.Tensor, Tuple[torch.Tensor, torch.Tensor]]:
|
|
"""PyTorch-native implementation equivalent to forward()."""
|
|
orig_dtype = x.dtype
|
|
x = x.to(torch.float32)
|
|
if residual is not None:
|
|
x = x + residual.to(torch.float32)
|
|
residual = x.to(orig_dtype)
|
|
|
|
variance = x.pow(2).mean(dim=-1, keepdim=True)
|
|
x = x * torch.rsqrt(variance + self.variance_epsilon)
|
|
x = x.to(orig_dtype) * self.weight
|
|
if residual is None:
|
|
return x
|
|
else:
|
|
return x, residual
|
|
|
|
def forward(
|
|
self,
|
|
x: torch.Tensor,
|
|
residual: Optional[torch.Tensor] = None,
|
|
scale: float = 1.0,
|
|
) -> Union[torch.Tensor, Tuple[torch.Tensor, torch.Tensor]]:
|
|
if residual is not None:
|
|
ops.fused_add_rms_norm(
|
|
x,
|
|
residual,
|
|
self.weight.data,
|
|
self.variance_epsilon,
|
|
scale,
|
|
)
|
|
return x, residual
|
|
out = torch.empty_like(x)
|
|
ops.rms_norm(
|
|
out,
|
|
x,
|
|
self.weight.data,
|
|
self.variance_epsilon,
|
|
)
|
|
return out
|
|
|
|
|
|
# ↓ add for smoothquant
|
|
class RMSNormQuant(nn.Module):
|
|
"""Root mean square normalization.
|
|
|
|
Computes x -> w * x / sqrt(E[x^2] + eps) where w is the learned weight.
|
|
Refer to https://arxiv.org/abs/1910.07467
|
|
"""
|
|
|
|
def __init__(
|
|
self,
|
|
hidden_size: int,
|
|
eps: float = 1e-6,
|
|
) -> None:
|
|
super().__init__()
|
|
self.weight = nn.Parameter(torch.ones(hidden_size))
|
|
self.variance_epsilon = eps
|
|
|
|
def forward(
|
|
self,
|
|
x: torch.Tensor,
|
|
) -> Union[torch.Tensor, Tuple[torch.Tensor, torch.Tensor]]:
|
|
out = torch.empty_like(x, dtype=torch.int8)
|
|
ops.rms_norm_quant(
|
|
out,
|
|
x,
|
|
self.weight.data,
|
|
self.variance_epsilon,
|
|
)
|
|
return out
|
|
|
|
|
|
class AddResidualRMSNormQuant(nn.Module):
|
|
"""Root mean square normalization.
|
|
Computes x -> w * x / sqrt(E[x^2] + eps) where w is the learned weight.
|
|
Refer to https://arxiv.org/abs/1910.07467
|
|
"""
|
|
|
|
def __init__(self,
|
|
hidden_size: int,
|
|
eps: float = 1e-6) -> None:
|
|
super().__init__()
|
|
self.weight = nn.Parameter(torch.ones(hidden_size))
|
|
self.variance_epsilon = eps
|
|
|
|
def forward(self,
|
|
x: torch.Tensor,
|
|
residual: torch.Tensor,
|
|
scale: torch.Tensor = None) -> torch.Tensor:
|
|
out = torch.empty_like(x, dtype=torch.int8)
|
|
ops.fused_add_rms_norm_quant(out, x, residual, self.weight.data, self.variance_epsilon)
|
|
return out, residual
|
|
|
|
|
|
class DequantAddResidualRMSNormQuant(nn.Module):
|
|
"""Root mean square normalization.
|
|
Computes x -> w * x / sqrt(E[x^2] + eps) where w is the learned weight.
|
|
Refer to https://arxiv.org/abs/1910.07467
|
|
"""
|
|
|
|
# TODO(Zhang Ying): use_per_token_dequant
|
|
def __init__(self,
|
|
hidden_size: int,
|
|
dequant_scale: float = 1.0,
|
|
use_per_token_dequant: bool = True,
|
|
eps: float = 1e-6) -> None:
|
|
super().__init__()
|
|
self.weight = nn.Parameter(torch.ones(hidden_size))
|
|
self.variance_epsilon = eps
|
|
self.register_parameter(
|
|
"dequant_scale",
|
|
torch.nn.Parameter(torch.tensor(dequant_scale,dtype=torch.float32,requires_grad=False))
|
|
)
|
|
self.use_per_token_dequant = use_per_token_dequant
|
|
|
|
def _apply(self, fn):
|
|
super()._apply(fn)
|
|
self.dequant_scale.data = self.dequant_scale.cpu()
|
|
return self
|
|
|
|
def to(self, *args, **kwargs):
|
|
super().to(*args, **kwargs)
|
|
self.dequant_scale.data = self.dequant_scale.to(*args, **kwargs)
|
|
self.dequant_scale.data = self.dequant_scale.to(torch.float32)
|
|
return self
|
|
|
|
def forward(self,
|
|
x: torch.Tensor,
|
|
residual: torch.Tensor,
|
|
scale: torch.Tensor = None) -> torch.Tensor:
|
|
out = torch.empty_like(x, dtype=torch.int8)
|
|
if self.use_per_token_dequant and scale is not None:
|
|
ops.dequant_fused_add_rms_norm_quant(
|
|
out, x, residual, self.weight.data,self.variance_epsilon,
|
|
scale, self.dequant_scale.item())
|
|
else:
|
|
ops.dequant_fused_add_rms_norm_quant(
|
|
out, x, residual, self.weight.data, self.variance_epsilon,
|
|
None, self.dequant_scale.item())
|
|
return out, residual
|
|
|
|
|
|
class DequantAddResidual(nn.Module):
|
|
def __init__(self,
|
|
dequant_scale: float = 1.0,
|
|
use_per_token_dequant: bool = True) -> None:
|
|
super().__init__()
|
|
self.register_parameter(
|
|
"dequant_scale",
|
|
torch.nn.Parameter(torch.tensor(dequant_scale,dtype=torch.float32,requires_grad=False))
|
|
)
|
|
self.use_per_token_dequant = use_per_token_dequant
|
|
|
|
def _apply(self, fn):
|
|
super()._apply(fn)
|
|
self.dequant_scale.data = self.dequant_scale.cpu()
|
|
return self
|
|
|
|
def to(self, *args, **kwargs):
|
|
super().to(*args, **kwargs)
|
|
self.dequant_scale.data = self.dequant_scale.to(*args, **kwargs)
|
|
self.dequant_scale.data = self.dequant_scale.to(torch.float32)
|
|
return self
|
|
|
|
def forward(self,
|
|
x: torch.Tensor,
|
|
residual: torch.Tensor,
|
|
scale: torch.Tensor = None) -> torch.Tensor:
|
|
out = torch.empty_like(residual)
|
|
if self.use_per_token_dequant and scale is not None:
|
|
ops.dequant_add_residual(out, x, residual, scale, self.dequant_scale.item())
|
|
else:
|
|
ops.dequant_add_residual(out, x, residual, None, self.dequant_scale.item())
|
|
return out
|
|
|
|
|
|
class AddResidual(DequantAddResidual):
|
|
def __init__(self,
|
|
dequant_scale: float = 1.0,
|
|
use_per_token_dequant: bool = True):
|
|
super().__init__(dequant_scale,use_per_token_dequant)
|
|
|
|
def forward(self,
|
|
x: torch.Tensor,
|
|
residual: torch.Tensor,
|
|
scale: torch.Tensor = None) -> torch.Tensor:
|
|
return x + residual
|