mirror of
https://github.com/kohya-ss/sd-scripts.git
synced 2026-10-12 02:17:16 +00:00
kohya-ssandClaude Fable 5.1
fa356a9990
Drop the attn_implementation="eager" pin for T5-XXL / byT5: the SDPA path that transformers 5.6+ selects by default is as accurate as eager against fp32 and should be faster, so accept the slight bf16 output difference and document it in the README instead. Update the transformers pin to 5.17.0. Co-Authored-By: Claude Fable 5.1 <[email protected]>
59 lines
1.8 KiB
Plaintext
59 lines
1.8 KiB
Plaintext
accelerate==1.15.0
|
|
transformers==5.17.0
|
|
diffusers[torch]==0.40.0
|
|
ftfy==6.3.1
|
|
# albumentations==1.3.0
|
|
# opencv-python has no wheel for Windows on ARM64 (e.g. NVIDIA RTX Spark PCs), so it is
|
|
# skipped there. It is optional: without it a Pillow/NumPy-based stub in library/_cv2_stub
|
|
# takes over automatically. Some tools under tools/ (canny.py, detect_face_rotate.py, and
|
|
# the ControlNet canny preprocessor) still require the real opencv-python.
|
|
opencv-python==4.10.0.84; platform_machine != "ARM64"
|
|
einops==0.7.0
|
|
# pytorch-lightning==1.9.0
|
|
bitsandbytes
|
|
lion-pytorch==0.2.3
|
|
schedulefree==1.4.1
|
|
pytorch-optimizer==3.10.0
|
|
prodigy-plus-schedule-free==1.9.2
|
|
prodigyopt==1.1.2
|
|
# tensorboard 2.x depends on grpcio, which has no wheel for Windows on ARM64 (pip would
|
|
# silently fall back to the ancient tensorboard 1.10). tensorboardX is a pure-Python
|
|
# drop-in that accelerate uses automatically for --log_with tensorboard; view the logs
|
|
# with TensorBoard on another machine.
|
|
tensorboard; platform_machine != "ARM64"
|
|
tensorboardX; platform_machine == "ARM64"
|
|
safetensors==0.8.0
|
|
# gradio==3.16.2
|
|
# altair==4.2.2
|
|
# easygui==0.98.3
|
|
toml==0.10.2
|
|
voluptuous==0.15.2
|
|
huggingface-hub==1.32.0
|
|
# for Image utils
|
|
imagesize==1.4.1
|
|
numpy
|
|
# <=2.0
|
|
# for BLIP captioning
|
|
# requests==2.28.2
|
|
# timm==0.6.12
|
|
# fairscale==0.4.13
|
|
# for WD14 captioning (tensorflow)
|
|
# tensorflow==2.10.1
|
|
# for WD14 captioning (onnx)
|
|
# onnx==1.15.0
|
|
# onnxruntime-gpu==1.17.1
|
|
# onnxruntime==1.17.1
|
|
# for cuda 12.1(default 11.8)
|
|
# onnxruntime-gpu --extra-index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/
|
|
|
|
# this is for onnx:
|
|
# protobuf==3.20.3
|
|
# open clip for SDXL
|
|
# open-clip-torch==2.20.0
|
|
# For logging
|
|
rich==14.1.0
|
|
# for T5XXL tokenizer (SD3/FLUX)
|
|
sentencepiece==0.2.1
|
|
# for kohya_ss library
|
|
-e .
|