File size: 1,708 Bytes
ef4d689 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 |
<!--Copyright 2024 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.
-->
# κ°μ
λ
Έμ΄μ¦κ° λ§μ μΆλ ₯μμ μ μ μΆλ ₯μΌλ‘ λ§λλ κ³Όμ μΌλ‘ κ³ νμ§ μμ± λͺ¨λΈμ μΆλ ₯μ λ§λλ κ°κ°μ λ°λ³΅λλ μ€ν
μ λ§μ κ³μ°μ΄ νμν©λλ€. 𧨠Diffuserμ λͺ©ν μ€ νλλ λͺ¨λ μ¬λμ΄ μ΄ κΈ°μ μ λ리 μ΄μ©ν μ μλλ‘ νλ κ²μ΄λ©°, μ¬κΈ°μλ μλΉμ λ° νΉμ νλμ¨μ΄μμ λΉ λ₯Έ μΆλ‘ μ κ°λ₯νκ² νλ κ²μ ν¬ν¨ν©λλ€.
μ΄ μΉμ
μμλ μΆλ‘ μλλ₯Ό μ΅μ ννκ³ λ©λͺ¨λ¦¬ μλΉλ₯Ό μ€μ΄κΈ° μν λ°μ λ°(half-precision) κ°μ€μΉ λ° sliced attentionκ³Ό κ°μ νκ³Ό μλ Ήμ λ€λ£Ήλλ€. λν [`torch.compile`](https://pytorch.org/tutorials/intermediate/torch_compile_tutorial.html) λλ [ONNX Runtime](https://onnxruntime.ai/docs/)μ μ¬μ©νμ¬ PyTorch μ½λμ μλλ₯Ό λμ΄κ³ , [xFormers](https://facebookresearch.github.io/xformers/)λ₯Ό μ¬μ©νμ¬ memory-efficient attentionμ νμ±ννλ λ°©λ²μ λ°°μΈ μ μμ΅λλ€. Apple Silicon, Intel λλ Habana νλ‘μΈμμ κ°μ νΉμ νλμ¨μ΄μμ μΆλ‘ μ μ€ννκΈ° μν κ°μ΄λλ μμ΅λλ€. |