RuntimeError: CUDA error: device-side assert triggered

RuntimeError: CUDA error: device-side assert triggered 

/pytorch/aten/src/THCUNN/BCECriterion.cu:57: void bce_updateOutput_no_reduce_functor<Dtype, Acctype>::operator()(const Dtype *, const Dtype *, Dtype *) [with Dtype = float, Acctype = float]: block: [5,0,0], thread: [183,0,0] Assertion `*input >= 0. && *input <= 1.` failed.

 

Assertion `*input >= 0. && *input <= 1.` failed.

# 抛出异常

我的错误定位到了

full_edge_accr_layers = [self.hit(full_logit_layer, 1-full_edge[:, 0].long()) for full_logit_layer in full_logit_layers]
query_edge_accr_layers = [torch.sum(full_edge_accr_layer * query_edge_mask * evaluation_mask) / torch.sum(query_edge_mask * evaluation_mask) for full_edge_accr_layer in full_edge_accr_layers]
#pdb.set_trace()
# compute node loss & accuracy (num_tasks x n