StableSR fp16 or 32?

"Global exception Input type (struct c10::Half) and bias type (float) should be the same" who can help me please.

Oct 24 '23 08:10 cardsellsteve

Hi. Which script are you running?

Oct 24 '23 08:10 IceClear

Hi. Which script are you running?

I only have a CPU, not a Nvidia GPU, and when I run app.py, it encounters an error. layers.0.bias', 'fea_tran.3.in_layers.2.weight', 'fea_tran.3.in_layers.2.bias', 'fea_tran.3.emb_layers.1.weight', 'feayers.0.weight', 'fea_ta_tran.3.emb_layers.1.bias', 'fea_tran.3.out_layers.0.weight', 'fea_tran.3.out_layers.0.bias', 'fea_tran.3.out_layerson.weight', 'fea_tran.3.3.weight', 'fea_tran.3.out_layers.3.bias', 'fea_tran.3.skip_connection.weight', 'fea_tran.3.skip_connection.bias'] Global exception Input type (struct c10::Half) and bias type (float) should be the same

Oct 25 '23 06:10 cardsellsteve

Hi. Which script are you running?

After I submitted one image, a warning was reported "Global exception Input type (struct c10::Half) and bias type (float) should be the same", and it was not during the execution of app.py that the warning occurred.

Oct 25 '23 06:10 cardsellsteve

I did not test on the CPU. I am afraid that the current code may not support CPU.

Oct 25 '23 08:10 IceClear

i have GPU 4090 but i have this [StbaleSR] in StableSR init - dtype: torch.float16, device: cpu

Feb 08 '24 10:02 ffdown