def make_node(self, axis, *tensors): node = Join.make_node(self, axis, *tensors) return Apply(self, [node.inputs[0]] + map(as_gpuarray_variable, tensors), [GpuArrayType(broadcastable=node.outputs[0].broadcastable, dtype=node.outputs[0].dtype)()])
def make_node(self, axis, *tensors): node = Join.make_node(self, axis, *tensors) return Apply(self, [node.inputs[0]] + list(map(as_gpuarray_variable, tensors)), [GpuArrayType(broadcastable=node.outputs[0].broadcastable, dtype=node.outputs[0].dtype)()])
def make_node(self, axis, *tensors): node = Join.make_node(self, axis, *tensors) ctx_name = infer_context_name(*tensors) def agv(v): return as_gpuarray_variable(v, context_name=ctx_name) return Apply(self, [node.inputs[0]] + list(map(agv, tensors)), [GpuArrayType(broadcastable=node.outputs[0].broadcastable, dtype=node.outputs[0].dtype, context_name=ctx_name)()])
def make_node(self, *tensors): # Neet to check ndim and shape of all input tensors! for x in tensors: assert x.type.ndim == 4 node = Join.make_node(self, 1, *tensors) def agv(v): return as_tensor_variable(v) return Apply(self, list(map(agv, tensors)), [TensorType(dtype=node.outputs[0].dtype, broadcastable=node.outputs[0].broadcastable)()])
def make_node(self, *axis_and_tensors): return Join.make_node(self, *axis_and_tensors)