Seongsu Park
8b18cd8e7f
* 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 일부 단어 수정 * 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 * ✨ Add translation for write_own_pipeline.mdx * conditional-직역, 언컨디셔널 * unconditional_image_generation first draft * reviese * Update pipeline_overview.mdx * revise-2 * ♻️ translation fixed for write_own_pipeline.mdx * complete translate basic_training.mdx * other-formats.mdx 번역 완료 * fix tutorials/basic_training.mdx * other-formats 수정 * inpaint 한국어 번역 * depth2img translation * translate training/adapt-a-model.mdx * revised_all * feedback taken * using_safetensors.mdx_first_draft * custom_pipeline_examples.mdx_first_draft * img2img 한글번역 완료 * tutorial_overview edit * reusing_seeds * torch2.0 * translate complete * fix) 용어 통일 규약 반영 * [fix] 피드백을 반영해서 번역 보정 * 오탈자 정정 + 컨벤션 위배된 부분 정정 * typo, style fix * toctree update * copyright fix * toctree fix * Update _toctree.yml --------- Co-authored-by: Chanran Kim <seriousran@gmail.com> Co-authored-by: apolinário <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.
|
|
-->
|
|
|
|
# 개요
|
|
|
|
노이즈가 많은 출력에서 적은 출력으로 만드는 과정으로 고품질 생성 모델의 출력을 만드는 각각의 반복되는 스텝은 많은 계산이 필요합니다. 🧨 Diffuser의 목표 중 하나는 모든 사람이 이 기술을 널리 이용할 수 있도록 하는 것이며, 여기에는 소비자 및 특수 하드웨어에서 빠른 추론을 가능하게 하는 것을 포함합니다.
|
|
|
|
이 섹션에서는 추론 속도를 최적화하고 메모리 소비를 줄이기 위한 반정밀(half-precision) 가중치 및 sliced attention과 같은 팁과 요령을 다룹니다. 또한 [`torch.compile`](https://pytorch.org/tutorials/intermediate/torch_compile_tutorial.html) 또는 [ONNX Runtime](https://onnxruntime.ai/docs/)을 사용하여 PyTorch 코드의 속도를 높이고, [xFormers](https://facebookresearch.github.io/xformers/)를 사용하여 memory-efficient attention을 활성화하는 방법을 배울 수 있습니다. Apple Silicon, Intel 또는 Habana 프로세서와 같은 특정 하드웨어에서 추론을 실행하기 위한 가이드도 있습니다. |