self trained model param mismatch

[Shiyao-Xu]

Dear all, I tried to use train_mono.py on nyu to do the training, and I chose DPT_BEiT_B_384 as my base. After training done, I wanted to test my model by using evaluate.py, but got param mismatch error, just like that below. I guess in the training process, the original zoeD_N model was overwrite, right? What should I do prevent this mismatch mistake?

[Muhammad-Aqib453]

i think you are using checkpoint wrong. you should use zoedepth_m12_n only other wise it will not work. as far as i know