File size: 2,963 Bytes
4a1f918 |
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 |
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.
import torch
import torch.nn as nn
from typing import Type
from .svd_layers import SVDLinear
# from .SALT_layers_please_work import SALTLinear
from .SALT_layers_3 import SALTLinear , SALTConv2d
from .lora_layers import LoRAConv2D, LoRALinear
class MLPBlock(nn.Module):
def __init__(
self,
embedding_dim: int,
mlp_dim: int,
act: Type[nn.Module] = nn.GELU,
mlp_transform=False,
use_lora = False
) -> None:
super().__init__()
if use_lora:
self.lin1 = LoRALinear(embedding_dim, mlp_dim)
self.lin2 = LoRALinear(mlp_dim, embedding_dim)
else:
# self.lin1 = SVDLinear(embedding_dim, mlp_dim, mlp_transform=mlp_transform)
# self.lin2 = SVDLinear(mlp_dim, embedding_dim, mlp_transform=mlp_transform)
rank_value = 500
# print("\nEmbedding dim in MLP Block is" ,embedding_dim)
# print("\n no need for MLP transform" , mlp_transform)
self.lin1 = SALTLinear(embedding_dim, mlp_dim, rank=rank_value , r_lora=256 , rsLora=False,alpha=1)
self.lin2 = SALTLinear(mlp_dim, embedding_dim, rank=rank_value , r_lora=256 , rsLora=False,alpha=1)
self.act = act()
def forward(self, x: torch.Tensor, output_loss=True) -> torch.Tensor:
out, reg_loss1 = self.lin1(x)
out, reg_loss2 = self.lin2(self.act(out))
if output_loss:
return out, (reg_loss1+reg_loss2)
else:
return out
class MLPBlock2(nn.Module):
def __init__(
self,
embedding_dim: int,
mlp_dim: int,
act: Type[nn.Module] = nn.GELU,
) -> None:
super().__init__()
self.lin1 = nn.Linear(embedding_dim, mlp_dim)
self.lin2 = nn.Linear(mlp_dim, embedding_dim)
self.act = act()
def forward(self, x: torch.Tensor) -> torch.Tensor:
out = self.lin1(x)
out = self.lin2(self.act(out))
return out
# From https://github.com/facebookresearch/detectron2/blob/main/detectron2/layers/batch_norm.py # noqa
# Itself from https://github.com/facebookresearch/ConvNeXt/blob/d1fa8f6fef0a165b27399986cc2bdacc92777e40/models/convnext.py#L119 # noqa
class LayerNorm2d(nn.Module):
def __init__(self, num_channels: int, eps: float = 1e-6) -> None:
super().__init__()
self.weight = nn.Parameter(torch.ones(num_channels))
self.bias = nn.Parameter(torch.zeros(num_channels))
self.eps = eps
def forward(self, x: torch.Tensor) -> torch.Tensor:
u = x.mean(1, keepdim=True)
s = (x - u).pow(2).mean(1, keepdim=True)
x = (x - u) / torch.sqrt(s + self.eps)
x = self.weight[:, None, None] * x + self.bias[:, None, None]
return x
|