when training a new pioneer model:
python pioneer_train.py
When exporting checkpoints as an onnx model, change code as the following:
-
line 11: device = torch.device("cpu") #device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-
line 247: #def train(self, replay_buffer, batch_data ,batch_size,writer): { ``` #}
-
line 241: #out=action[0,ind].cpu().data.numpy().flatten() #out=torch.tensor(out) out=action[0,ind]
After completing the changes , select the appropriate checkpoints file and run it within the terminal
python pionner.py