CloserLookFewShot
CloserLookFewShot copied to clipboard
How to fine-tune the Matching Network
When I checked the fine-tuning code of the Matching network on the novel dataset, I found that only the preliminary embedding (for example: backbone.Conv4) was called, and the FCE network trained by the Matching Network was not called. I am very confused about the fine-tuning steps of Matching Network. Looking forward to your reply!!! Thank you!!!
Can you access to pretrained weights?