eaglew / describing_a_knowledge_base Goto Github PK

（Why do the main functions cand and ref in the eval_final file have two descriptions generated_description1 and generated_description2?？Do we need two values to compare？？）
为什么eval_final文件的main函数cand和ref里面有两个，分别为generated_description1和generated_description2？需要有两个值对比吗

首先十分感谢您的回复 python main.py --cuda --mode 3 报错BrokenPipeError: [Errno 32] Broken pipe
--mode 1预测单个实体，GPU运行了一个晚上，比训练时间都慢吗？一直处于Finish build vocabulary
Finish build word dictionary
number of training examples: 250186
model restored
Read test data
Type index from (0 to 29981) to continue: 这个状态
遗憾的是，这里上传不了屏幕截图 ，

亲爱的楼主，不好意思又麻烦您，我运行python main.py --cuda --mode 3 报错BrokenPipeError: [Errno 32] Broken pipe 从网上找了解决方案都是修改num_workers =0，可是我们没有调用torch.utils.data.DataLoader()函数

Hello,

Great repo on NLG.
May I know how to test the trained model with custom input.

Thanks

def preeval_batch(self, dataset): return cands, refs 请问 cands和refs 的区别？分别代表是参考内容和结果？

Hi, Im trying to run main.py for the animal dataset.. However im unable to do getting below mentioned error. Kindly assist thanks.

File "main.py", line 92, in
model = Seq2seq(encoder, decoder).to(device)

File "C:\Users\Anaconda3\lib\site-packages\torch\cuda_init_.py", line 210, in _lazy_init
raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Torch not compiled with CUDA enabled