VictorMylle/Kohya-ss-sd-scripts
1
0
Fork 1
mirror of https://github.com/kohya-ss/sd-scripts.git synced 2026-04-09 14:45:19 +00:00
Code Issues Packages Projects Releases Wiki Activity

レイヤー数変更(hako-mikan/sd-webui-lora-block-weight参考)

Browse Source
This commit is contained in:
u-haru
2023-04-02 04:02:34 +09:00
parent 786971d443
commit 058e442072
1 changed files with 22 additions and 30 deletions
Download Patch File Download Diff File Expand all files Collapse all files

52
networks/lora.py
View File

@@ -337,6 +337,7 @@ class LoRANetwork(torch.nn.Module):
self.up_lr_weight:list[float] = None self.up_lr_weight:list[float] = None
self.down_lr_weight:list[float] = None self.down_lr_weight:list[float] = None
self.mid_lr_weight:float = None self.mid_lr_weight:float = None
self.stratified_lr = False
def set_multiplier(self, multiplier): def set_multiplier(self, multiplier):
self.multiplier = multiplier self.multiplier = multiplier
@@ -434,10 +435,7 @@ class LoRANetwork(torch.nn.Module):
# 層別学習率用に層ごとの学習率に対する倍率を定義する # 層別学習率用に層ごとの学習率に対する倍率を定義する
def set_stratified_lr_weight(self, up_lr_weight:list[float]|str=None, mid_lr_weight:float=None, down_lr_weight:list[float]|str=None, zero_threshold:float=0.0): def set_stratified_lr_weight(self, up_lr_weight:list[float]|str=None, mid_lr_weight:float=None, down_lr_weight:list[float]|str=None, zero_threshold:float=0.0):
max_len = 3 # attentions -> attentions -> attentions で3個のModuleに対して定義 max_len=12 # フルモデル相当でのup,downの層の数
if self.apply_to_conv2d_3x3:
max_len = 10 # (resnets -> {up,down}sampler -> attentions) x3 -> resnets で10個のModuleに対して定義
def get_list(name) -> list[float]: def get_list(name) -> list[float]:
import math import math
if name=="cosine": if name=="cosine":
@@ -469,6 +467,7 @@ class LoRANetwork(torch.nn.Module):
up_lr_weight = up_lr_weight + [1.0] * (max_len - len(up_lr_weight)) up_lr_weight = up_lr_weight + [1.0] * (max_len - len(up_lr_weight))
if (up_lr_weight != None) or (mid_lr_weight != None) or (down_lr_weight != None): if (up_lr_weight != None) or (mid_lr_weight != None) or (down_lr_weight != None):
print("層別学習率を適用します。") print("層別学習率を適用します。")
self.stratified_lr = True
if (down_lr_weight != None): if (down_lr_weight != None):
self.down_lr_weight = [w if w > zero_threshold else 0 for w in down_lr_weight[:max_len]] self.down_lr_weight = [w if w > zero_threshold else 0 for w in down_lr_weight[:max_len]]
print("down_lr_weight(浅い層->深い層):",self.down_lr_weight) print("down_lr_weight(浅い層->深い層):",self.down_lr_weight)
@@ -483,31 +482,22 @@ class LoRANetwork(torch.nn.Module):
def get_stratified_lr_weight(self, lora:LoRAModule) -> float: def get_stratified_lr_weight(self, lora:LoRAModule) -> float:
m = RE_UPDOWN.search(lora.lora_name) m = RE_UPDOWN.search(lora.lora_name)
if m: if m:
idx = 0
g = m.groups() g = m.groups()
i = int(g[1]) i = int(g[1])
if self.apply_to_conv2d_3x3: j = int(g[3])
if g[2]=="resnets": if g[2]=="resnets":
idx=3*i idx=3*i + j
elif g[2]=="attentions": elif g[2]=="attentions":
if g[0]=="down": idx=3*i + j
idx=3*i + 2 elif g[2]=="upsamplers" or g[2]=="downsamplers":
else: idx=3*i + 2
idx=3*i - 1
elif g[2]=="upsamplers" or g[2]=="downsamplers":
idx=3*i + 1
else:
idx=i
if g[0]=="up":
idx=i-1
if (g[0]=="up") and (self.up_lr_weight != None): if (g[0]=="down") and (self.down_lr_weight != None):
return self.down_lr_weight[idx+1]
elif (g[0]=="up") and (self.up_lr_weight != None):
return self.up_lr_weight[idx] return self.up_lr_weight[idx]
elif (g[0]=="down") and (self.down_lr_weight != None): elif ("mid_block_" in lora.lora_name) and (self.mid_lr_weight != None): # idx=12
return self.down_lr_weight[idx] return self.mid_lr_weight
elif ("mid_block_" in lora.lora_name) and (self.mid_lr_weight != None):
return self.mid_lr_weight
# print({'params': lora.parameters(), 'lr':alpha*lr})
return 1 return 1
def prepare_optimizer_params(self, text_encoder_lr, unet_lr , default_lr): def prepare_optimizer_params(self, text_encoder_lr, unet_lr , default_lr):
@@ -525,13 +515,15 @@ class LoRANetwork(torch.nn.Module):
if self.unet_loras: if self.unet_loras:
for lora in self.unet_loras: for lora in self.unet_loras:
param_data={} param_data = {'params': lora.parameters()}
if unet_lr is not None: if unet_lr is not None:
param_data = {'params': lora.parameters(), 'lr':self.get_stratified_lr_weight(lora)*unet_lr} param_data['lr'] = unet_lr
elif default_lr is not None: elif default_lr is not None:
param_data = {'params': lora.parameters(), 'lr':self.get_stratified_lr_weight(lora)*default_lr} param_data['lr'] = default_lr
if param_data["lr"]==0: if self.stratified_lr and ('lr' in param_data):
continue param_data['lr'] = param_data['lr'] * self.get_stratified_lr_weight(lora)
if (param_data['lr']==0):
continue
all_params.append(param_data) all_params.append(param_data)
return all_params return all_params