I have used OpenNMT-py-1.0.0rc1 model for my image2text dataset. I have trained the model for 100K steps. Now when I am using that trained model demo-model_step_100000.pt to predict the testing dataset equations, it is only predicting the same equation, irrespective of the input dataset. Even if I change the model from demo-model_step_100000.pt to demo-model_step_80000.pt, it throws only one equation equation. Can anyone please help me what’s going on here? (I have installed opennmt-py using pip as mentioned here: GitHub - OpenNMT/OpenNMT-py at legacy )
λ − 1 = − λ − 1
λ − 1 = − λ − 1
λ − 1 = − λ − 1
λ − 1 = − λ − 1
λ − 1 = − λ − 1
λ − 1 = − λ − 1
This is the equation that it is predicting irrespective of any input eqaution.