Reinforcement Q-learning issue on GPU #36
Description
when running on GPU, torch.cuda.is_available() == True, I run into the following problem, in training loop:
TypeError Traceback (most recent call last)
in ()
50 for t in count():
51 # Select and perform an action
---> 52 action = select_action(state)
53 _, reward, done, _ = env.step(action[0,0])
54 reward = torch.Tensor([reward])
in select_action(state)
27 steps_done += 1
28 if sample > eps_threshold:
---> 29 return model(Variable(state, volatile=True)).data.max(1)[1].cpu()
30 else:
31 return torch.LongTensor([[random.randrange(2)]])
/home/ubuntu/.local/lib/python2.7/site-packages/torch/nn/modules/module.pyc in call(self, *input, **kwargs)
200
201 def call(self, *input, **kwargs):
--> 202 result = self.forward(*input, **kwargs)
203 for hook in self._forward_hooks.values():
204 hook_result = hook(self, input, result)
in forward(self, x)
11
12 def forward(self, x):
---> 13 x = F.relu(self.bn1(self.conv1(x)))
14 x = F.relu(self.bn2(self.conv2(x)))
15 x = F.relu(self.bn3(self.conv3(x)))
/home/ubuntu/.local/lib/python2.7/site-packages/torch/nn/modules/module.pyc in call(self, *input, **kwargs)
200
201 def call(self, *input, **kwargs):
--> 202 result = self.forward(*input, **kwargs)
203 for hook in self._forward_hooks.values():
204 hook_result = hook(self, input, result)
/home/ubuntu/.local/lib/python2.7/site-packages/torch/nn/modules/conv.pyc in forward(self, input)
235 def forward(self, input):
236 return F.conv2d(input, self.weight, self.bias, self.stride,
--> 237 self.padding, self.dilation, self.groups)
238
239
/home/ubuntu/.local/lib/python2.7/site-packages/torch/nn/functional.pyc in conv2d(input, weight, bias, stride, padding, dilation, groups)
35 f = ConvNd(_pair(stride), _pair(padding), _pair(dilation), False,
36 _pair(0), groups)
---> 37 return f(input, weight, bias) if bias is not None else f(input, weight)
38
39
/home/ubuntu/.local/lib/python2.7/site-packages/torch/nn/_functions/conv.pyc in forward(self, input, weight, bias)
33 if k == 3:
34 input, weight = _view4d(input, weight)
---> 35 output = self._update_output(input, weight, bias)
36 if k == 3:
37 output, = _view3d(output)
/home/ubuntu/.local/lib/python2.7/site-packages/torch/nn/_functions/conv.pyc in _update_output(self, input, weight, bias)
95
96 self._bufs = [[] for g in range(self.groups)]
---> 97 return self._thnn('update_output', input, weight, bias)
98
99 def _grad_input(self, input, weight, grad_output):
/home/ubuntu/.local/lib/python2.7/site-packages/torch/nn/_functions/conv.pyc in _thnn(self, fn_name, input, weight, *args)
154 impl = _thnn_convs[self.thnn_class_name(input)]
155 if self.groups == 1:
--> 156 return impl[fn_name](self, self._bufs[0], input, weight, *args)
157 else:
158 res = []
/home/ubuntu/.local/lib/python2.7/site-packages/torch/nn/_functions/conv.pyc in call_update_output(self, bufs, input, weight, bias)
232 args = parse_arguments(self, fn.arguments[5:], bufs, kernel_size)
233 getattr(backend, fn.name)(backend.library_state, input, output, weight,
--> 234 bias, *args)
235 return output
236 return call_update_output
TypeError: FloatSpatialConvolutionMM_updateOutput received an invalid combination of arguments - got (int, torch.FloatTensor, torch.FloatTensor, torch.cuda.FloatTensor, torch.cuda.FloatTensor, torch.FloatTensor, torch.FloatTensor, long, long, int, int, int, int), but expected (int state, torch.FloatTensor input, torch.FloatTensor output, torch.FloatTensor weight, [torch.FloatTensor bias or None], torch.FloatTensor finput, torch.FloatTensor fgradInput, int kW, int kH, int dW, int dH, int padW, int padH)