结果跑出来准确率并不是很高啊，什么原因

[wangtao666666]

training : total 19 nodes. LSTM_only 0 ----test: 0.1 | dev: 0.1 training : total 19 nodes. LSTM_only 1 ----test: 0.3 | dev: 0.3 LSTM pretrain OK epsilon 0.05 training : total 19 nodes. RL pretrain 0 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 1 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 2 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 3 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 4 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 5 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 6 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 7 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 8 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 9 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 10 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 11 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 12 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 13 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 14 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 15 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 16 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 17 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 18 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 19 ----test: 0.3 | dev: 0.3 RL pretrain OK training : total 19 nodes. epoch 0 ----test: 0.3 | dev: 0.3 training : total 19 nodes. epoch 1 ----test: 0.3 | dev: 0.3 training : total 19 nodes. epoch 2 ----test: 0.3 | dev: 0.3 training : total 19 nodes. epoch 3 ----test: 0.3 | dev: 0.3 training : total 19 nodes. epoch 4 ----test: 0.0 | dev: 0.0

Process finished with exit code 0

[wangtao666666]

AG数据集合，哪里出问题啦嘛

[wangtao666666]

数据集 AG:https://drive.google.com/open?id=1becf7pzfuLL7qgWqv4q-TyDYjSzodWfR 词向量：GLOVE 300维词向量语料，为什么差这么多

[scofield7419]

AG数据集合，哪里出问题啦嘛

我的跑出来的是符合论文中的指标的。 你使用了fasttest，试试把： parser.add_argument('--fasttest', type=int, default=0, choices=[0, 1]) 设为0，使用全量数据训练.

[wangtao666666]

AG数据集合，哪里出问题啦嘛

我的跑出来的是符合论文中的指标的。 你使用了fasttest，试试把： parser.add_argument('--fasttest', type=int, default=0, choices=[0, 1]) 设为0，使用全量数据训练.

👌的，谢谢，还有一个问题action定义为1是保留操作吧,定义为0是删除操作

[YMMY22]

你好，我想问一下代码里，datamanager.py文件中get_vector()函数的参数，vector.25dim文件你是从哪里找到的，我找了好久都没找到。如果可以的话，麻烦发给我一下，我的Email：659640300@qq.com，十分感谢！

[MENGHAH]

training : total 19 nodes. LSTM_only 0 ----test: 0.1 | dev: 0.1 training : total 19 nodes. LSTM_only 1 ----test: 0.3 | dev: 0.3 LSTM pretrain OK epsilon 0.05 training : total 19 nodes. RL pretrain 0 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 1 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 2 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 3 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 4 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 5 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 6 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 7 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 8 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 9 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 10 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 11 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 12 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 13 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 14 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 15 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 16 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 17 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 18 ----test: 0.3 | dev: 0.3 training : total 19 nodes. RL pretrain 19 ----test: 0.3 | dev: 0.3 RL pretrain OK training : total 19 nodes. epoch 0 ----test: 0.3 | dev: 0.3 training : total 19 nodes. epoch 1 ----test: 0.3 | dev: 0.3 training : total 19 nodes. epoch 2 ----test: 0.3 | dev: 0.3 training : total 19 nodes. epoch 3 ----test: 0.3 | dev: 0.3 training : total 19 nodes. epoch 4 ----test: 0.0 | dev: 0.0

Process finished with exit code 0 你好，我想问一下你是否明白代码Actor.py里面的input_l和input_d都代表什么吗？

[MENGHAH]

Traceback (most recent call last): File "datamanager.py", line 106, in train_data, test_data, dev_data = datamanager.getdata(2, 200) File "datamanager.py", line 74, in getdata solution = one_hot_vector(int(sent['rating'])) File "datamanager.py", line 53, in one_hot_vector s[r] += 1.0 IndexError: index 3 is out of bounds for axis 0 with size 2 我在执行datamanager.py的文件时出现了以上错误，不知道前辈们有没有遇到过，是怎么解决的？

[Maigewm]

您好，请问您的问题最终解决了没有？关于代码部分我有一些疑惑，请问可以联系您解答吗？我的邮箱是411466357@qq.com，万分感谢！