Weights for Camelyon16 - Githubissues

binli123 / dsmil-wsi

DSMIL: Dual-stream multiple instance learning networks for tumor detection in Whole Slide Image

MIT License

378 stars 88 forks source link

Open Ajaz-Ahmad opened 3 years ago

Ajaz-Ahmad commented 3 years ago

Hi,

I saw the thread regarding camelyon16 weights. I can see in the shared drive weights with embedder_v-1 and embedder_v-2. Can you please describe what each one is? Is the v1 for 5x and v2 for 20x or the other way ? https://drive.google.com/drive/folders/1jd5qbpZ0fdqJdH3FYQaTimxYHkzNa_bW

Thanks

binli123 commented 3 years ago

I have updated the link: https://drive.google.com/drive/folders/14pSKk2rnPJiJsGK2CQJXctP7fhRJZiyn?usp=sharing These are all 20x models but with different training settings (batch size 512-4096 with different training time (5-20days))

Ajaz-Ahmad commented 3 years ago

Can you please upload the weights for lower patches and model_v2 is corresponding to 4096 batch size?

Thanks

86ai68 commented 2 years ago

Can you specify v1 v2 v3 for that configuration?