How to resume training CT-CLIP

ibrahimethemhamamci / CT-CLIP

Developing Generalist Foundation Models from a Multimodal Dataset for 3D Computed Tomography

197 stars 21 forks source link

Closed qic999 closed 6 months ago

qic999 commented 7 months ago

Thank you for sharing the excellent work. How can one resume from a pre-training stage and continue training?

sezginerr commented 7 months ago

Hello @qic999,

You can add the following line after you define clip in run_train.py

clip.load("path_to_pretrained_model")

This should continue pre-training stage. There is no scheduler in the zero-shot training so no need to load the optimizer state.