Checkpoint file issue with semantic segmentation for inferencing beit model

microsoft / unilm

Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities

https://aka.ms/GeneralAI

MIT License

19.12k stars 2.44k forks source link

Checkpoint file issue with semantic segmentation for inferencing beit model #1230

Open alekhyaparimala opened 11 months ago

alekhyaparimala commented 11 months ago

Model I am using is unilm, I'm trying to inference beit model, after loading the checkpoint file, i'm getting error as Checkpoint does not have 'meta' key File "/home/T7724A1/Documents/unilm/beit/semantic_segmentation/tools/test.py", line 139, in main model.CLASSES = checkpoint['meta']['CLASSES'] KeyError: 'meta'

The check point file I'm using here is vbeit_large_patch16_224_pt22k_ft22k.pth

addf400 commented 10 months ago

You are currently using a checkpoint that has not been fine-tuned. You may consider trying a fine-tuned checkpoint, such as the one provided in this link to beit_large_patch16_640_pt22k_ft22ktoade20k.