mirror of
https://github.com/huggingface/diffusers.git
synced 2026-01-29 07:22:12 +03:00
* feat) optimization kr translation * fix) typo, italic setting * feat) dreambooth, text2image kr * feat) lora kr * fix) LoRA * fix) fp16 fix * fix) doc-builder style * fix) fp16 ์ผ๋ถ ๋จ์ด ์์ * fix) fp16 style fix * fix) opt, training docs update * merge conflict * Fix community pipelines (#3266) * Allow disabling torch 2_0 attention (#3273) * Allow disabling torch 2_0 attention * make style * Update src/diffusers/models/attention.py * Release: v0.16.1 * feat) toctree update * feat) toctree update * Fix custom releases (#3708) * Fix custom releases * make style * Fix loading if unexpected keys are present (#3720) * Fix loading * make style * Release: v0.17.0 * opt_overview * commit * Create pipeline_overview.mdx * unconditional_image_generatoin_1stDraft * โจ Add translation for write_own_pipeline.mdx * conditional-์ง์ญ, ์ธ์ปจ๋์ ๋ * unconditional_image_generation first draft * reviese * Update pipeline_overview.mdx * revise-2 * โป๏ธ translation fixed for write_own_pipeline.mdx * complete translate basic_training.mdx * other-formats.mdx ๋ฒ์ญ ์๋ฃ * fix tutorials/basic_training.mdx * other-formats ์์ * inpaint ํ๊ตญ์ด ๋ฒ์ญ * depth2img translation * translate training/adapt-a-model.mdx * revised_all * feedback taken * using_safetensors.mdx_first_draft * custom_pipeline_examples.mdx_first_draft * img2img ํ๊ธ๋ฒ์ญ ์๋ฃ * tutorial_overview edit * reusing_seeds * torch2.0 * translate complete * fix) ์ฉ์ด ํต์ผ ๊ท์ฝ ๋ฐ์ * [fix] ํผ๋๋ฐฑ์ ๋ฐ์ํด์ ๋ฒ์ญ ๋ณด์ * ์คํ์ ์ ์ + ์ปจ๋ฒค์ ์๋ฐฐ๋ ๋ถ๋ถ ์ ์ * typo, style fix * toctree update * copyright fix * toctree fix * Update _toctree.yml --------- Co-authored-by: Chanran Kim <seriousran@gmail.com> Co-authored-by: apolinรกrio <joaopaulo.passos@gmail.com> Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com> Co-authored-by: Lee, Hongkyu <75282888+howsmyanimeprofilepicture@users.noreply.github.com> Co-authored-by: hyeminan <adios9709@gmail.com> Co-authored-by: movie5 <oyh5800@naver.com> Co-authored-by: idra79haza <idra79haza@github.com> Co-authored-by: Jihwan Kim <cuchoco@naver.com> Co-authored-by: jungwoo <boonkoonheart@gmail.com> Co-authored-by: jjuun0 <jh061993@gmail.com> Co-authored-by: szjung-test <93111772+szjung-test@users.noreply.github.com> Co-authored-by: idra79haza <37795618+idra79haza@users.noreply.github.com> Co-authored-by: howsmyanimeprofilepicture <howsmyanimeprofilepicture@gmail.com> Co-authored-by: hoswmyanimeprofilepicture <hoswmyanimeprofilepicture@gmail.com>
17 lines
1.7 KiB
Plaintext
17 lines
1.7 KiB
Plaintext
<!--Copyright 2023 The HuggingFace Team. All rights reserved.
|
|
|
|
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.
|
|
-->
|
|
|
|
# ๊ฐ์
|
|
|
|
๋
ธ์ด์ฆ๊ฐ ๋ง์ ์ถ๋ ฅ์์ ์ ์ ์ถ๋ ฅ์ผ๋ก ๋ง๋๋ ๊ณผ์ ์ผ๋ก ๊ณ ํ์ง ์์ฑ ๋ชจ๋ธ์ ์ถ๋ ฅ์ ๋ง๋๋ ๊ฐ๊ฐ์ ๋ฐ๋ณต๋๋ ์คํ
์ ๋ง์ ๊ณ์ฐ์ด ํ์ํฉ๋๋ค. ๐งจ Diffuser์ ๋ชฉํ ์ค ํ๋๋ ๋ชจ๋ ์ฌ๋์ด ์ด ๊ธฐ์ ์ ๋๋ฆฌ ์ด์ฉํ ์ ์๋๋ก ํ๋ ๊ฒ์ด๋ฉฐ, ์ฌ๊ธฐ์๋ ์๋น์ ๋ฐ ํน์ ํ๋์จ์ด์์ ๋น ๋ฅธ ์ถ๋ก ์ ๊ฐ๋ฅํ๊ฒ ํ๋ ๊ฒ์ ํฌํจํฉ๋๋ค.
|
|
|
|
์ด ์น์
์์๋ ์ถ๋ก ์๋๋ฅผ ์ต์ ํํ๊ณ ๋ฉ๋ชจ๋ฆฌ ์๋น๋ฅผ ์ค์ด๊ธฐ ์ํ ๋ฐ์ ๋ฐ(half-precision) ๊ฐ์ค์น ๋ฐ sliced attention๊ณผ ๊ฐ์ ํ๊ณผ ์๋ น์ ๋ค๋ฃน๋๋ค. ๋ํ [`torch.compile`](https://pytorch.org/tutorials/intermediate/torch_compile_tutorial.html) ๋๋ [ONNX Runtime](https://onnxruntime.ai/docs/)์ ์ฌ์ฉํ์ฌ PyTorch ์ฝ๋์ ์๋๋ฅผ ๋์ด๊ณ , [xFormers](https://facebookresearch.github.io/xformers/)๋ฅผ ์ฌ์ฉํ์ฌ memory-efficient attention์ ํ์ฑํํ๋ ๋ฐฉ๋ฒ์ ๋ฐฐ์ธ ์ ์์ต๋๋ค. Apple Silicon, Intel ๋๋ Habana ํ๋ก์ธ์์ ๊ฐ์ ํน์ ํ๋์จ์ด์์ ์ถ๋ก ์ ์คํํ๊ธฐ ์ํ ๊ฐ์ด๋๋ ์์ต๋๋ค. |