Skip to content

Use torch.svd_lowrank for large matrices in resize_lora.py #2240

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Use torch.svd_lowrank for large matrices in resize_lora.py #2240

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 13 additions & 2 deletions networks/resize_lora.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,7 +87,14 @@ def index_sv_ratio(S, target):
# Modified from Kohaku-blueleaf's extract/merge functions
def extract_conv(weight, lora_rank, dynamic_method, dynamic_param, device, scale=1):
out_size, in_size, kernel_size, _ = weight.size()
U, S, Vh = torch.linalg.svd(weight.reshape(out_size, -1).to(device))
weight = weight.reshape(out_size, -1)
_in_size = in_size * kernel_size * kernel_size

if out_size > 2048 and _in_size > 2048:
U, S, V = torch.svd_lowrank(weight.to(device), q=min(2 * lora_rank, out_size, _in_size))
Vh = V.T
else:
U, S, Vh = torch.linalg.svd(weight.to(device))

param_dict = rank_resize(S, lora_rank, dynamic_method, dynamic_param, scale)
lora_rank = param_dict["new_rank"]
Expand All @@ -106,7 +113,11 @@ def extract_conv(weight, lora_rank, dynamic_method, dynamic_param, device, scale
def extract_linear(weight, lora_rank, dynamic_method, dynamic_param, device, scale=1):
out_size, in_size = weight.size()

U, S, Vh = torch.linalg.svd(weight.to(device))
if out_size > 2048 and in_size > 2048:
U, S, V = torch.svd_lowrank(weight.to(device), q=min(2 * lora_rank, out_size, in_size))
Vh = V.T
else:
U, S, Vh = torch.linalg.svd(weight.to(device))

param_dict = rank_resize(S, lora_rank, dynamic_method, dynamic_param, scale)
lora_rank = param_dict["new_rank"]
Expand Down