VictorMylle/Kohya-ss-sd-scripts
1
0
Fork 1
mirror of https://github.com/kohya-ss/sd-scripts.git synced 2026-04-09 06:45:09 +00:00
Code Issues Packages Projects Releases Wiki Activity

refactor optimizer selection for bnb

Browse Source
This commit is contained in:
Kohya S
2023-07-30 13:43:29 +09:00
parent b62185b821
commit a296654c1b
1 changed files with 44 additions and 50 deletions
Download Patch File Download Diff File Expand all files Collapse all files

94
library/train_util.py
View File

@@ -65,6 +65,7 @@ import safetensors.torch
from library.lpw_stable_diffusion import StableDiffusionLongPromptWeightingPipeline from library.lpw_stable_diffusion import StableDiffusionLongPromptWeightingPipeline
import library.model_util as model_util import library.model_util as model_util
import library.huggingface_util as huggingface_util import library.huggingface_util as huggingface_util
# from library.attention_processors import FlashAttnProcessor # from library.attention_processors import FlashAttnProcessor
# from library.hypernetwork import replace_attentions_for_hypernetwork # from library.hypernetwork import replace_attentions_for_hypernetwork
from library.original_unet import UNet2DConditionModel from library.original_unet import UNet2DConditionModel
@@ -2165,7 +2166,7 @@ def cache_batch_latents(
info.latents = latent info.latents = latent
if flip_aug: if flip_aug:
info.latents_flipped = flipped_latent info.latents_flipped = flipped_latent
if torch.cuda.is_available(): if torch.cuda.is_available():
torch.cuda.empty_cache() torch.cuda.empty_cache()
@@ -3184,32 +3185,9 @@ def get_optimizer(args, trainable_params):
# print("optkwargs:", optimizer_kwargs) # print("optkwargs:", optimizer_kwargs)
lr = args.learning_rate lr = args.learning_rate
optimizer = None
if optimizer_type == "AdamW8bit".lower(): if optimizer_type == "Lion".lower():
try:
import bitsandbytes as bnb
except ImportError:
raise ImportError("No bitsand bytes / bitsandbytesがインストールされていないようです")
print(f"use 8-bit AdamW optimizer | {optimizer_kwargs}")
optimizer_class = bnb.optim.AdamW8bit
optimizer = optimizer_class(trainable_params, lr=lr, **optimizer_kwargs)
elif optimizer_type == "SGDNesterov8bit".lower():
try:
import bitsandbytes as bnb
except ImportError:
raise ImportError("No bitsand bytes / bitsandbytesがインストールされていないようです")
print(f"use 8-bit SGD with Nesterov optimizer | {optimizer_kwargs}")
if "momentum" not in optimizer_kwargs:
print(
f"8-bit SGD with Nesterov must be with momentum, set momentum to 0.9 / 8-bit SGD with Nesterovはmomentum指定が必須のため0.9に設定します"
)
optimizer_kwargs["momentum"] = 0.9
optimizer_class = bnb.optim.SGD8bit
optimizer = optimizer_class(trainable_params, lr=lr, nesterov=True, **optimizer_kwargs)
elif optimizer_type == "Lion".lower():
try: try:
import lion_pytorch import lion_pytorch
except ImportError: except ImportError:
@@ -3217,37 +3195,53 @@ def get_optimizer(args, trainable_params):
print(f"use Lion optimizer | {optimizer_kwargs}") print(f"use Lion optimizer | {optimizer_kwargs}")
optimizer_class = lion_pytorch.Lion optimizer_class = lion_pytorch.Lion
optimizer = optimizer_class(trainable_params, lr=lr, **optimizer_kwargs) optimizer = optimizer_class(trainable_params, lr=lr, **optimizer_kwargs)
elif optimizer_type.endswith("8bit".lower()): elif optimizer_type.endswith("8bit".lower()):
try: try:
import bitsandbytes as bnb import bitsandbytes as bnb
except ImportError: except ImportError:
raise ImportError("No bitsandbytes / bitsandbytesがインストールされていないようです") raise ImportError("No bitsandbytes / bitsandbytesがインストールされていないようです")
if optimizer_type == "Lion8bit".lower(): if optimizer_type == "AdamW8bit".lower():
print(f"use 8-bit Lion optimizer | {optimizer_kwargs}") print(f"use 8-bit AdamW optimizer | {optimizer_kwargs}")
try: optimizer_class = bnb.optim.AdamW8bit
optimizer_class = bnb.optim.Lion8bit optimizer = optimizer_class(trainable_params, lr=lr, **optimizer_kwargs)
except AttributeError:
raise AttributeError( elif optimizer_type == "SGDNesterov8bit".lower():
"No Lion8bit. The version of bitsandbytes installed seems to be old. Please install 0.38.0 or later. / Lion8bitが定義されていません。インストールされているbitsandbytesのバージョンが古いようです。0.38.0以上をインストールしてください" print(f"use 8-bit SGD with Nesterov optimizer | {optimizer_kwargs}")
) if "momentum" not in optimizer_kwargs:
print(
f"8-bit SGD with Nesterov must be with momentum, set momentum to 0.9 / 8-bit SGD with Nesterovはmomentum指定が必須のため0.9に設定します"
)
optimizer_kwargs["momentum"] = 0.9
optimizer_class = bnb.optim.SGD8bit
optimizer = optimizer_class(trainable_params, lr=lr, nesterov=True, **optimizer_kwargs)
elif optimizer_type == "Lion8bit".lower():
print(f"use 8-bit Lion optimizer | {optimizer_kwargs}")
try:
optimizer_class = bnb.optim.Lion8bit
except AttributeError:
raise AttributeError(
"No Lion8bit. The version of bitsandbytes installed seems to be old. Please install 0.38.0 or later. / Lion8bitが定義されていません。インストールされているbitsandbytesのバージョンが古いようです。0.38.0以上をインストールしてください"
)
elif optimizer_type == "PagedAdamW8bit".lower(): elif optimizer_type == "PagedAdamW8bit".lower():
print(f"use 8-bit PagedAdamW optimizer | {optimizer_kwargs}") print(f"use 8-bit PagedAdamW optimizer | {optimizer_kwargs}")
try: try:
optimizer_class = bnb.optim.PagedAdamW8bit optimizer_class = bnb.optim.PagedAdamW8bit
except AttributeError: except AttributeError:
raise AttributeError( raise AttributeError(
"No PagedAdamW8bit. The version of bitsandbytes installed seems to be old. Please install 0.39.0 or later. / PagedAdamW8bitが定義されていません。インストールされているbitsandbytesのバージョンが古いようです。0.39.0以上をインストールしてください" "No PagedAdamW8bit. The version of bitsandbytes installed seems to be old. Please install 0.39.0 or later. / PagedAdamW8bitが定義されていません。インストールされているbitsandbytesのバージョンが古いようです。0.39.0以上をインストールしてください"
) )
elif optimizer_type == "PagedLion8bit".lower(): elif optimizer_type == "PagedLion8bit".lower():
print(f"use 8-bit Paged Lion optimizer | {optimizer_kwargs}") print(f"use 8-bit Paged Lion optimizer | {optimizer_kwargs}")
try: try:
optimizer_class = bnb.optim.PagedLion8bit optimizer_class = bnb.optim.PagedLion8bit
except AttributeError: except AttributeError:
raise AttributeError( raise AttributeError(
"No PagedLion8bit. The version of bitsandbytes installed seems to be old. Please install 0.39.0 or later. / PagedLion8bitが定義されていません。インストールされているbitsandbytesのバージョンが古いようです。0.39.0以上をインストールしてください" "No PagedLion8bit. The version of bitsandbytes installed seems to be old. Please install 0.39.0 or later. / PagedLion8bitが定義されていません。インストールされているbitsandbytesのバージョンが古いようです。0.39.0以上をインストールしてください"
) )
optimizer = optimizer_class(trainable_params, lr=lr, **optimizer_kwargs) optimizer = optimizer_class(trainable_params, lr=lr, **optimizer_kwargs)
@@ -3379,7 +3373,7 @@ def get_optimizer(args, trainable_params):
optimizer_class = torch.optim.AdamW optimizer_class = torch.optim.AdamW
optimizer = optimizer_class(trainable_params, lr=lr, **optimizer_kwargs) optimizer = optimizer_class(trainable_params, lr=lr, **optimizer_kwargs)
else: if optimizer is None:
# 任意のoptimizerを使う # 任意のoptimizerを使う
optimizer_type = args.optimizer_type # lowerでないやつ（微妙） optimizer_type = args.optimizer_type # lowerでないやつ（微妙）
print(f"use {optimizer_type} | {optimizer_kwargs}") print(f"use {optimizer_type} | {optimizer_kwargs}")