File size: 30,506 Bytes
3b609b9 |
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 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 |
# Copyright 2023-present the HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import annotations
import math
import warnings
from typing import Any, Optional, Union
import torch
import torch.nn as nn
import torch.nn.functional as F
from peft.tuners.tuners_utils import BaseTunerLayer, check_adapters_to_merge
class MultiplicativeDropoutLayer(nn.Module):
"""
Implements the multiplicative dropout layer for OFT.
"""
def __init__(self, p=0.0):
"""
Initializes the multiplicative dropout layer.
Parameters:
p (float): The probability of dropping out a block. Defaults to 0.0.
"""
super().__init__()
self.p = p
def forward(self, x):
"""
Applies multiplicative dropout to the input tensor.
Parameters:
x (Tensor): The input tensor of shape (D, H, H), where `D` represents
the number of OFT blocks, and `H` is the size of the square blocks along the last two dimensions,
the block size in OFT.
"""
if self.training:
# Ensure the last two dimensions are the same
if x.shape[-1] != x.shape[-2]:
raise ValueError("The last two dimensions of input should be the same!")
D, H, _ = x.shape
# If block share, skip the multiplicative dropout
if D == 1:
return x
num_to_replace = int(self.p * D)
num_zeros = D - num_to_replace
mask = torch.cat([torch.ones(num_to_replace, device=x.device), torch.zeros(num_zeros, device=x.device)])
mask = mask[torch.randperm(D)].view(D, 1, 1)
eye_matrix = torch.eye(H, device=x.device).repeat(D, 1, 1)
x = (1 - mask) * x + mask * eye_matrix
return x
class OFTLayer(BaseTunerLayer):
"""
Implements the OFT layer.
"""
# All names of layers that may contain adapter weights
adapter_layer_names = ("oft_r", "oft_s")
# other_param_names is defined on parent class
other_param_names = ("r", "oft_block_size", "oft_dropout")
def __init__(self, base_layer: nn.Module, **kwargs) -> None:
"""
Initializes the OFT layer.
Note, currently only support linear layer and convolutional layer, with further support for other layers to be
added soon.
Parameters:
base_layer: the pretrained model layer
"""
self.base_layer = base_layer
# OFT info
self.oft_r = nn.ParameterDict({})
self.oft_s = nn.ParameterDict({})
self.r = {}
self.oft_block_size = {}
self.oft_dropout = nn.ModuleDict({})
self.coft = {}
self.eps = {}
self.block_share = {}
# Mark the weight as unmerged
self._disable_adapters = False
self.merged_adapters = []
self.kwargs = kwargs
base_layer = self.get_base_layer()
if isinstance(base_layer, nn.Linear):
in_features, out_features = base_layer.in_features, base_layer.out_features
elif isinstance(base_layer, nn.Conv2d):
in_features, out_features = base_layer.in_channels, base_layer.out_channels
else:
raise ValueError(f"Unsupported layer type {type(base_layer)}")
self.in_features = in_features
self.out_features = out_features
@property
def _available_adapters(self) -> set[str]:
return {*self.oft_r}
def set_scale(self, adapter, scale):
if adapter not in self.scaling:
# Ignore the case where the adapter is not in the layer
return
warnings.warn("Scaling operation for OFT not supported! Automatically set scale to 1.")
def scale_layer(self, scale: float) -> None:
if scale == 1:
return
for active_adapter in self.active_adapters:
if active_adapter not in self.oft_r.keys():
continue
warnings.warn("Scaling operation for OFT not supported! Automatically set scale to 1.")
def unscale_layer(self, scale=None) -> None:
for active_adapter in self.active_adapters:
if active_adapter not in self.oft_r.keys():
continue
warnings.warn("Unscaling operation for OFT not supported! Keeping scale to 1.")
def update_layer(self, adapter_name, r, oft_block_size, module_dropout, coft, eps, block_share, init_weights):
"""
Update the linear layer with trainable OFT weights. Override for other layer types.
"""
"""Internal function to create oft adapter
Args:
adapter_name (`str`): Name for the adapter to add.
r (`int`): Rank for the added adapter.
oft_block_size (`int`): The block size for added adapter.
module_dropout (`float`):
The multiplicative dropout probability for disabling adapter blocks during training.
coft (`bool`): Whether to use the constrained variant of OFT or not.
eps (`float`):
The control strength of COFT. The freedom of rotation. Only has an effect if `coft` is set to True.
block_share (`bool`): Whether to share the OFT parameters between blocks or not.
init_weights (`bool`): Whether to initialize weights.
"""
# Initialize the MultiplicativeDropoutLayer for module_dropout > 0.0.
if module_dropout > 0.0:
oft_dropout_layer = MultiplicativeDropoutLayer(p=module_dropout)
else:
oft_dropout_layer = nn.Identity()
self.oft_dropout.update(nn.ModuleDict({adapter_name: oft_dropout_layer}))
if r == 0 and oft_block_size != 0:
if self.in_features % oft_block_size != 0 or oft_block_size > self.in_features:
old_oft_block_size = oft_block_size
oft_block_size = self.adjust_oft_parameters(self.in_features, oft_block_size)
warnings.warn(
f"Invalid `oft_block_size` ({old_oft_block_size})! Adjusted `oft_block_size` to ({oft_block_size})."
)
r = int(self.in_features // oft_block_size)
elif r != 0 and oft_block_size == 0:
if self.in_features % r != 0 or r > self.in_features:
old_r = r
r = self.adjust_oft_parameters(self.in_features, r)
warnings.warn(f"Invalid `r` ({old_r})! Adjusted `r` to ({r}).")
oft_block_size = int(self.in_features // r)
else:
raise ValueError(
"Something went wrong, please report this error: https://github.com/huggingface/peft/issues"
)
self.coft[adapter_name] = coft
self.block_share[adapter_name] = block_share
self.eps[adapter_name] = eps * math.ceil(self.out_features / r) * math.ceil(self.out_features / r)
# Create weights with provided shape
if block_share:
self.oft_r[adapter_name] = nn.Parameter(
torch.empty(1, math.ceil(self.in_features / r), math.ceil(self.in_features / r))
)
else:
self.oft_r[adapter_name] = nn.Parameter(
torch.empty(r, math.ceil(self.in_features / r), math.ceil(self.in_features / r))
)
self.oft_s[adapter_name] = nn.Parameter(torch.empty(int(self.out_features), 1))
# Initialize weights
self.reset_oft_parameters(adapter_name, init_weights)
# set oft r and block size
self.r[adapter_name] = r
self.oft_block_size[adapter_name] = oft_block_size
# Move new weights to device
self._move_adapter_to_device_of_base_layer(adapter_name)
self.set_adapter(self.active_adapters)
def reset_oft_parameters(self, adapter_name, init_weights):
"""
Reset the OFT parameters.
"""
if init_weights is False:
nn.init.normal_(self.oft_r[adapter_name], mean=0.0, std=0.1)
nn.init.normal_(self.oft_s[adapter_name], mean=1.0, std=0.1)
return
if adapter_name in self.oft_r.keys():
if init_weights is True:
# initialize oft_r to zero
nn.init.zeros_(self.oft_r[adapter_name])
nn.init.ones_(self.oft_s[adapter_name])
else:
raise ValueError(f"Unknown initialization {init_weights=}")
def _cayley_batch(self, data: torch.Tensor) -> torch.Tensor:
"""
Perform the Cayley parametrization on a batch of skew-symmetric matrices.
Args:
data: A batch of skew-symmetric matrices of shape (b, r, c).
"""
b, r, c = data.shape
# Ensure the input matrix is skew-symmetric
skew_mat = 0.5 * (data - data.transpose(1, 2))
id_mat = torch.eye(r, device=data.device).unsqueeze(0).expand(b, r, c) # noqa: E741
# Perform the Cayley parametrization
Q = torch.linalg.solve(id_mat + skew_mat, id_mat - skew_mat, left=False)
return Q
# Copied from https://github.com/Zeju1997/oft/blob/84cebb965df69781e3d9c3c875f5980b421eaf24/oft-control/oft.py#L155
def _block_diagonal(self, oft_r: torch.Tensor, rank: int) -> torch.Tensor:
if oft_r.shape[0] == 1:
# block share
blocks = [oft_r[0, ...] for i in range(rank)]
else:
blocks = [oft_r[i, ...] for i in range(rank)]
# Use torch.block_diag to create the block diagonal matrix
A = torch.block_diag(*blocks)
return A
# Copied from https://github.com/Zeju1997/oft/blob/84cebb965df69781e3d9c3c875f5980b421eaf24/oft-control/oft.py#L52
def _project_batch(self, oft_r, eps=1e-5):
# scaling factor for each of the smaller block matrix
eps = eps * 1 / torch.sqrt(torch.tensor(oft_r.shape[0]))
I = ( # noqa: E741
torch.zeros((oft_r.size(1), oft_r.size(1)), device=oft_r.device, dtype=oft_r.dtype)
.unsqueeze(0)
.expand_as(oft_r)
)
diff = oft_r - I
norm_diff = torch.norm(oft_r - I, dim=(1, 2), keepdim=True)
mask = (norm_diff <= eps).bool()
out = torch.where(mask, oft_r, I + eps * (diff / norm_diff))
return out
def adjust_oft_parameters(self, in_features, params):
"""
Adjust the OFT parameters to be divisible by the in_features dimension.
"""
if params < in_features:
higher_params = params
while higher_params <= in_features and in_features % higher_params != 0:
higher_params += 1
else:
return in_features
lower_params = params
while lower_params > 1 and in_features % lower_params != 0:
lower_params -= 1
if (params - lower_params) <= (higher_params - params):
return lower_params
else:
return higher_params
class Linear(nn.Module, OFTLayer):
"""OFT implemented in Linear layer"""
def __init__(
self,
base_layer,
adapter_name: str,
r: int = 8,
oft_block_size: int = 0,
module_dropout: float = 0.0,
coft: bool = False,
eps: float = 6e-5,
block_share: bool = False,
fan_in_fan_out: bool = False, # Set this to True if the layer to replace stores weight like (fan_in, fan_out)
init_weights: Union[bool, str] = True,
is_target_conv_1d_layer: bool = False,
**kwargs,
) -> None:
super().__init__()
OFTLayer.__init__(self, base_layer, **kwargs)
self.fan_in_fan_out = fan_in_fan_out
self._active_adapter = adapter_name
self.update_layer(adapter_name, r, oft_block_size, module_dropout, coft, eps, block_share, init_weights)
self.is_target_conv_1d_layer = is_target_conv_1d_layer
def merge(self, safe_merge: bool = False, adapter_names: Optional[list[str]] = None) -> None:
"""
Merge the active adapter weights into the base weights
Args:
safe_merge (`bool`, *optional*):
If `True`, the merge operation will be performed in a copy of the original weights and check for NaNs
before merging the weights. This is useful if you want to check if the merge operation will produce
NaNs. Defaults to `False`.
adapter_names (`List[str]`, *optional*):
The list of adapter names that should be merged. If `None`, all active adapters will be merged.
Defaults to `None`.
"""
adapter_names = check_adapters_to_merge(self, adapter_names)
if not adapter_names:
# no adapter to merge
return
for active_adapter in adapter_names:
if active_adapter in self._available_adapters:
base_layer = self.get_base_layer()
if safe_merge:
# Note that safe_merge will be slower than the normal merge
# because of the copy operation.
orig_weights = base_layer.weight.data
oft_mat, oft_s = self.get_delta_weight(active_adapter)
orig_weights = torch.transpose(orig_weights, 0, 1)
orig_weights = torch.mm(oft_mat, orig_weights)
orig_weights = torch.transpose(orig_weights, 0, 1)
orig_weights = orig_weights * oft_s
if not torch.isfinite(orig_weights).all():
raise ValueError(
f"NaNs detected in the merged weights. The adapter {active_adapter} seems to be broken"
)
base_layer.weight.data = orig_weights.contiguous()
else:
oft_mat, oft_s = self.get_delta_weight(active_adapter)
orig_weights = base_layer.weight.data
orig_weights = torch.transpose(orig_weights, 0, 1)
orig_weights = torch.mm(oft_mat, orig_weights)
orig_weights = torch.transpose(orig_weights, 0, 1)
orig_weights = orig_weights * oft_s
base_layer.weight.data = orig_weights.contiguous()
self.merged_adapters.append(active_adapter)
def unmerge(self) -> None:
"""
This method unmerges all merged adapter layers from the base weights.
"""
if not self.merged:
warnings.warn("Already unmerged. Nothing to do.")
return
while len(self.merged_adapters) > 0:
active_adapter = self.merged_adapters.pop()
if active_adapter in self.oft_r.keys():
oft_mat, oft_s = self.get_delta_weight(active_adapter)
orig_weights = self.get_base_layer().weight.data
orig_weights = torch.transpose(orig_weights, 0, 1)
orig_weights = torch.mm(oft_mat.t(), orig_weights)
orig_weights = torch.transpose(orig_weights, 0, 1)
self.get_base_layer().weight.data = orig_weights * (1 / oft_s)
def get_delta_weight(self, adapter_name) -> tuple[torch.Tensor, torch.Tensor]:
"""
Compute the delta weight for the given adapter.
Args:
adapter (str):
The name of the adapter for which the delta weight should be computed.
"""
oft_r = self.oft_r[adapter_name]
oft_s = self.oft_s[adapter_name]
rank = self.r[adapter_name]
coft = self.coft[adapter_name]
eps = self.eps[adapter_name]
if coft:
with torch.no_grad():
oft_r.copy_(self._project_batch(oft_r, eps=eps))
orth_rotate = self._cayley_batch(oft_r)
weight = self._block_diagonal(orth_rotate, rank)
return weight, oft_s
def forward(self, x: torch.Tensor, *args, **kwargs) -> torch.Tensor:
previous_dtype = x.dtype
if self.disable_adapters:
if self.merged:
self.unmerge()
result = self.base_layer(x, *args, **kwargs)
elif self.merged:
result = self.base_layer(x, *args, **kwargs)
else:
oft_rotation = torch.eye(self.in_features, device=x.device, dtype=previous_dtype)
oft_scale = torch.ones((int(self.out_features), 1), device=x.device, dtype=previous_dtype)
for active_adapter in self.active_adapters:
if active_adapter not in self.oft_r.keys():
continue
oft_r = self.oft_r[active_adapter]
oft_s = self.oft_s[active_adapter]
dropout = self.oft_dropout[active_adapter]
rank = self.r[active_adapter]
coft = self.coft[active_adapter]
eps = self.eps[active_adapter]
if coft:
with torch.no_grad():
oft_r.copy_(self._project_batch(oft_r, eps=eps))
orth_rotate = self._cayley_batch(oft_r)
orth_rotate = dropout(orth_rotate)
oft_mat = self._block_diagonal(orth_rotate, rank)
oft_rotation = oft_mat @ oft_rotation
oft_scale = oft_s * oft_scale
x = x.to(self.get_base_layer().weight.data.dtype)
orig_weight = self.get_base_layer().weight.data
orig_weight = torch.transpose(orig_weight, 0, 1)
oft_rotation = oft_rotation.to(previous_dtype)
orig_weight = orig_weight.to(previous_dtype)
rotated_weight = torch.mm(oft_rotation, orig_weight)
rotated_weight = torch.transpose(rotated_weight, 0, 1)
scaled_rotated_weight = rotated_weight * oft_scale
scaled_rotated_weight = scaled_rotated_weight.to(previous_dtype)
bias = self.get_base_layer().bias.to(previous_dtype) if self.get_base_layer().bias is not None else None
result = F.linear(input=x, weight=scaled_rotated_weight, bias=bias)
result = result.to(previous_dtype)
return result
def __repr__(self) -> str:
rep = super().__repr__()
return "oft." + rep
class Conv2d(nn.Module, OFTLayer):
"""OFT implemented in Conv2d layer"""
def __init__(
self,
base_layer: nn.Module,
adapter_name: str,
r: int = 8,
oft_block_size: int = 0,
fan_in_fan_out: bool = False, # Set this to True if the layer to replace stores weight like (fan_in, fan_out)
module_dropout: float = 0.0,
coft: bool = False,
eps: float = 6e-5,
block_share: bool = False,
init_weights: Union[bool, str] = True,
**kwargs,
) -> None:
super().__init__()
OFTLayer.__init__(self, base_layer)
self.fan_in_fan_out = fan_in_fan_out
self._active_adapter = adapter_name
# Create adapter and set it active
self.update_layer(adapter_name, r, oft_block_size, module_dropout, coft, eps, block_share, init_weights)
def update_layer(self, adapter_name, r, oft_block_size, module_dropout, coft, eps, block_share, init_weights):
"""
Update the conv2d layer with trainable OFT weights.
"""
# Initialize the MultiplicativeDropoutLayer for module_dropout > 0.0.
if module_dropout > 0.0:
oft_dropout_layer = MultiplicativeDropoutLayer(p=module_dropout)
else:
oft_dropout_layer = nn.Identity()
self.oft_dropout.update(nn.ModuleDict({adapter_name: oft_dropout_layer}))
# layer information from the base layer
base_layer = self.get_base_layer()
conv_filter_dim = self.in_features * base_layer.kernel_size[0] * base_layer.kernel_size[0]
if r == 0 and oft_block_size != 0:
if conv_filter_dim % oft_block_size != 0 or oft_block_size > conv_filter_dim:
old_oft_block_size = oft_block_size
oft_block_size = self.adjust_oft_parameters(conv_filter_dim, oft_block_size)
warnings.warn(
f"Invalid `oft_block_size` ({old_oft_block_size})! Adjusted `oft_block_size` to ({oft_block_size})."
)
r = int(conv_filter_dim // oft_block_size)
elif r != 0 and oft_block_size == 0:
if conv_filter_dim % r != 0 or r > conv_filter_dim:
old_r = r
r = self.adjust_oft_parameters(conv_filter_dim, r)
warnings.warn(f"Invalid `r` ({old_r})! Adjusted `r` to ({r}).")
oft_block_size = int(conv_filter_dim // r)
else:
raise ValueError(
"Something went wrong, please report this error: https://github.com/huggingface/peft/issues"
)
self.coft[adapter_name] = coft
self.block_share[adapter_name] = block_share
self.eps[adapter_name] = eps * math.ceil(self.out_features / r) * math.ceil(self.out_features / r)
# Create weights with provided shape
if block_share:
self.oft_r[adapter_name] = nn.Parameter(
torch.empty(1, math.ceil(conv_filter_dim / r), math.ceil(conv_filter_dim / r))
)
else:
self.oft_r[adapter_name] = nn.Parameter(
torch.empty(r, math.ceil(conv_filter_dim / r), math.ceil(conv_filter_dim / r))
)
self.oft_s[adapter_name] = nn.Parameter(torch.empty(int(self.out_features), 1))
# Initialize weights
self.reset_oft_parameters(adapter_name, init_weights)
# set oft r and block size
self.r[adapter_name] = r
self.oft_block_size[adapter_name] = oft_block_size
# Move new weights to device
self._move_adapter_to_device_of_base_layer(adapter_name)
self.set_adapter(self.active_adapters)
def merge(self, safe_merge: bool = False, adapter_names: Optional[list[str]] = None) -> None:
"""
Merge the active adapter weights into the base weights
Args:
safe_merge (`bool`, *optional*):
If True, the merge operation will be performed in a copy of the original weights and check for NaNs
before merging the weights. This is useful if you want to check if the merge operation will produce
NaNs. Defaults to `False`.
adapter_names (`List[str]`, *optional*):
The list of adapter names that should be merged. If None, all active adapters will be merged. Defaults
to `None`.
"""
adapter_names = check_adapters_to_merge(self, adapter_names)
if not adapter_names:
# no adapter to merge
return
for active_adapter in adapter_names:
if active_adapter in self.oft_r.keys():
base_layer = self.get_base_layer()
if safe_merge:
# Note that safe_merge will be slower than the normal merge
# because of the copy operation.
orig_weights = base_layer.weight.data.clone()
oft_mat, oft_s = self.get_delta_weight(active_adapter)
orig_weights = orig_weights.view(
self.out_features, self.in_features * base_layer.kernel_size[0] * base_layer.kernel_size[0]
)
orig_weights = torch.transpose(orig_weights, 0, 1)
orig_weights = torch.mm(oft_mat, orig_weights)
orig_weights = torch.transpose(orig_weights, 0, 1)
orig_weights = orig_weights * oft_s
orig_weights = orig_weights.view(
self.out_features, self.in_features, base_layer.kernel_size[0], base_layer.kernel_size[0]
)
base_layer.weight.data = orig_weights.contiguous()
else:
oft_mat, oft_s = self.get_delta_weight(active_adapter)
orig_weights = base_layer.weight.data.clone()
orig_weights = orig_weights.view(
self.out_features, self.in_features * base_layer.kernel_size[0] * base_layer.kernel_size[0]
)
orig_weights = torch.transpose(orig_weights, 0, 1)
orig_weights = torch.mm(oft_mat, orig_weights)
orig_weights = torch.transpose(orig_weights, 0, 1)
orig_weights = orig_weights * oft_s
orig_weights = orig_weights.view(
self.out_features, self.in_features, base_layer.kernel_size[0], base_layer.kernel_size[0]
)
base_layer.weight.data = orig_weights.contiguous()
self.merged_adapters.append(active_adapter)
def unmerge(self) -> None:
"""
This method unmerges all merged adapter layers from the base weights.
"""
if not self.merged:
warnings.warn("Already unmerged. Nothing to do.")
return
while len(self.merged_adapters) > 0:
active_adapter = self.merged_adapters.pop()
if active_adapter in self.oft_r.keys():
oft_mat, oft_s = self.get_delta_weight(active_adapter)
orig_weights = self.get_base_layer().weight.data.clone()
orig_weights = orig_weights.view(
self.out_features,
self.in_features * self.get_base_layer().kernel_size[0] * self.get_base_layer().kernel_size[0],
)
orig_weights = torch.transpose(orig_weights, 0, 1)
orig_weights = torch.mm(oft_mat.t(), orig_weights)
orig_weights = torch.transpose(orig_weights, 0, 1)
orig_weights = orig_weights * (1 / oft_s)
orig_weights = orig_weights.view(
self.out_features,
self.in_features,
self.get_base_layer().kernel_size[0],
self.get_base_layer().kernel_size[0],
)
self.get_base_layer().weight.data = orig_weights
def get_delta_weight(self, adapter_name) -> tuple[torch.Tensor, torch.Tensor]:
"""
Compute the delta weight for the given adapter.
Args:
adapter (str):
The name of the adapter for which the delta weight should be computed.
"""
oft_r = self.oft_r[adapter_name]
oft_s = self.oft_s[adapter_name]
rank = self.r[adapter_name]
coft = self.coft[adapter_name]
eps = self.eps[adapter_name]
if coft:
with torch.no_grad():
oft_r.copy_(self._project_batch(oft_r, eps=eps))
orth_rotate = self._cayley_batch(oft_r)
weight = self._block_diagonal(orth_rotate, rank)
return weight, oft_s
def forward(self, x: torch.Tensor, *args: Any, **kwargs: Any) -> torch.Tensor:
previous_dtype = x.dtype
if self.disable_adapters:
if self.merged:
self.unmerge()
result = self.base_layer(x, *args, **kwargs)
elif self.merged:
result = self.base_layer(x, *args, **kwargs)
else:
oft_rotation = torch.eye(
self.in_features * self.get_base_layer().kernel_size[0] * self.get_base_layer().kernel_size[0],
device=x.device,
dtype=previous_dtype,
)
oft_scale = torch.ones((int(self.out_features), 1), device=x.device, dtype=previous_dtype)
for active_adapter in self.active_adapters:
if active_adapter not in self.oft_r.keys():
continue
oft_r = self.oft_r[active_adapter]
oft_s = self.oft_s[active_adapter]
dropout = self.oft_dropout[active_adapter]
rank = self.r[active_adapter]
coft = self.coft[active_adapter]
eps = self.eps[active_adapter]
if coft:
with torch.no_grad():
oft_r.copy_(self._project_batch(oft_r, eps=eps))
orth_rotate = self._cayley_batch(oft_r)
orth_rotate = dropout(orth_rotate)
oft_mat = self._block_diagonal(orth_rotate, rank)
oft_rotation = oft_mat @ oft_rotation
oft_scale = oft_s * oft_scale
x = x.to(self.get_base_layer().weight.data.dtype)
orig_weights = self.base_layer.weight.data
orig_weights = orig_weights.view(
self.out_features,
self.in_features * self.get_base_layer().kernel_size[0] * self.get_base_layer().kernel_size[0],
)
orig_weights = torch.transpose(orig_weights, 0, 1)
oft_rotation = oft_rotation.to(previous_dtype)
orig_weights = orig_weights.to(previous_dtype)
rotated_weight = torch.mm(oft_rotation, orig_weights)
rotated_weight = torch.transpose(rotated_weight, 0, 1)
scaled_rotated_weight = rotated_weight * oft_scale
scaled_rotated_weight = scaled_rotated_weight.view(
self.out_features,
self.in_features,
self.get_base_layer().kernel_size[0],
self.get_base_layer().kernel_size[0],
)
result = F.conv2d(
input=x,
weight=scaled_rotated_weight,
bias=self.get_base_layer().bias,
padding=self.get_base_layer().padding[0],
stride=self.get_base_layer().stride[0],
)
result = result.to(previous_dtype)
return result
def __repr__(self) -> str:
rep = super().__repr__()
return "oft." + rep
|