does anyone encounter issues like the cycle_A loss suddenly becomes very high after a random iteration? my loss log shows that in epoch 78, the discrimintor loss suddenly decrease to close to zero and my generator loss goes from 0.2ish to almost 1 and 2. The cycle_A loss also looks like being suddenly very high.
The inferenced image is all black now after these severeal losses exploded.
Hi, @junyanz @taesungp
does anyone encounter issues like the cycle_A loss suddenly becomes very high after a random iteration? my loss log shows that in epoch 78, the discrimintor loss suddenly decrease to close to zero and my generator loss goes from 0.2ish to almost 1 and 2. The cycle_A loss also looks like being suddenly very high.
The inferenced image is all black now after these severeal losses exploded.