Botao Chen 52a21ce78f Free up memory after post training finishes (#770) ... ## context Currently, the GPU memory will be continuously occupied after the training finishes. In this PR, we explicitly delete the reference and clean up the memory after training finishes. ## test Before the change, after training a llama 3.2 3B model, >6GB GPU memory is still occupied After the change, after training a llama 3.2 3B model, the GPU memory drops to ~1GB <img width="156" alt="Screenshot 2025-01-14 at 6 05 17 PM" src="https://github.com/user-attachments/assets/45d212b1-a651-49f3-aad9-1c0a27fcebcf" />		2025-01-14 19:19:38 -08:00
..
common	[post training] define llama stack post training dataset format (#717)	2025-01-14 12:48:49 -08:00
datasets	[post training] define llama stack post training dataset format (#717)	2025-01-14 12:48:49 -08:00
recipes	Free up memory after post training finishes (#770)	2025-01-14 19:19:38 -08:00
__init__.py	[1/n] torchtune <> llama-stack integration skeleton (#540)	2024-12-13 11:05:35 -08:00
config.py	[1/n] torchtune <> llama-stack integration skeleton (#540)	2024-12-13 11:05:35 -08:00
post_training.py	[remove import *] clean up import *'s (#689)	2024-12-27 15:45:44 -08:00