-
Notifications
You must be signed in to change notification settings - Fork 272
Description
λ¬Έμ URL
μμ μ΄ νμν URLμ λ¨κ²¨μ£ΌμΈμ. (μ. https://tutorials.pytorch.kr/beginner/basics/intro.html)
λ³κ²½ μ¬ν
(1)μ΄λ€ λ¨μ΄ / λ¬Έμ₯ / λ΄μ©μ΄ (2)μ΄λ»κ² λ³κ²½λμ΄μΌ νλ€κ³ μκ°νμΈμ?
(1) λ΄μ©λ¬Έμλ λ²μμ΄ λμ΄ μμΌλ https://tutorials.pytorch.kr/recipes/recipes/warmstarting_model_using_parameters_from_a_different_model.html
PYTORCHμμ λ€λ₯Έ λͺ¨λΈμ λ§€κ°λ³μλ₯Ό μ¬μ©νμ¬ λΉ λ₯΄κ² λͺ¨λΈ μμνκΈ°(WARMSTART)
λͺ¨λΈμ λΆλΆμ μΌλ‘ λ‘λνκ±°λ λΆλΆ λͺ¨λΈμ λ‘λνμ¬ νλ ¨ νλ‘μΈμ€λ₯Ό λΉ λ₯΄κ² μμνλ κ²μ΄ μ²μλΆν° νλ ¨νλ κ²λ³΄λ€ ν¨μ¬ λΉ λ₯΄κ² λͺ¨λΈμ μλ ΄νλ λ° μ΄λ»κ² λμμ΄ λλμ§ μμ보μΈμ.
(2) μΈλ±μ€μ μλ λ΄μ©μ΄ λ²μλμ΄ μμ§ μμ΅λλ€.
https://github.com/PyTorchKorea/tutorials-kr/blob/master/recipes_source/recipes_index.rst
Warmstarting model using parameters from a different model in PyTorch
Learn how warmstarting the training process by partially loading a model or loading a partial model can help your model converge much faster than training from scratch.
μΆκ° μ 보
μμ κ°μ΄ μκ°νμ μ΄μ λλ λ€λ₯Έ μ°Έκ³ ν μ λ³΄κ° μλ€λ©΄ μλ €μ£ΌμΈμ.
λ€λ₯Έ λ¬Έμμμ μΌκ΄μ±μ μν΄ λ³Έλ¬Έμ΄ λ²μμ΄ λμ΄ μλ€λ©΄ μ λͺ©λ λ°μμ΄ λμ΄μμΌλ©΄ λ²μλ λ¬ΈμλΌλ κ²μ μμ μμ΅λλ€.
Metadata
Metadata
Assignees
Labels
Type
Projects
Status