gaussic / text-classification-cnn-rnn

CNN-RNN中文文本分类,基于TensorFlow
MIT License
4.16k stars 1.47k forks source link

运行完后的模型在哪呢? #142

Closed flitdu closed 4 years ago

flitdu commented 4 years ago

可否单独加载呢??

gaussic commented 4 years ago

请认真阅读run_cnn.py里面的test部分。

flitdu commented 4 years ago

如下,验证集波动性很大,怎么回事呢?? Epoch: 9 Iter: 328, Train Loss: 0.014, Train Acc: 100.00%, Val Loss: 8.0, Val Acc: 50.00%, Time: 0:00:23 Iter: 330, Train Loss: 3.7e-09, Train Acc: 100.00%, Val Loss: 4.8, Val Acc: 61.76%, Time: 0:00:23 Iter: 332, Train Loss: 0.18, Train Acc: 96.88%, Val Loss: 2.8, Val Acc: 69.12%, Time: 0:00:23 Iter: 334, Train Loss: 3.4e-08, Train Acc: 100.00%, Val Loss: 1.4, Val Acc: 83.82%, Time: 0:00:23 Iter: 336, Train Loss: 1.5e-08, Train Acc: 100.00%, Val Loss: 1.1, Val Acc: 86.76%, Time: 0:00:23 Iter: 338, Train Loss: 0.39, Train Acc: 93.75%, Val Loss: 0.95, Val Acc: 89.71%, Time: 0:00:24 Iter: 340, Train Loss: 7.5e-09, Train Acc: 100.00%, Val Loss: 0.93, Val Acc: 86.76%, Time: 0:00:24 Iter: 342, Train Loss: 7.4e-06, Train Acc: 100.00%, Val Loss: 1.5, Val Acc: 80.88%, Time: 0:00:24 Iter: 344, Train Loss: 1.6e-05, Train Acc: 100.00%, Val Loss: 2.6, Val Acc: 67.65%, Time: 0:00:24 Iter: 346, Train Loss: 0.004, Train Acc: 100.00%, Val Loss: 3.6, Val Acc: 61.76%, Time: 0:00:24 Iter: 348, Train Loss: 3.2e-05, Train Acc: 100.00%, Val Loss: 4.4, Val Acc: 57.35%, Time: 0:00:24 Iter: 350, Train Loss: 0.00024, Train Acc: 100.00%, Val Loss: 3.9, Val Acc: 58.82%, Time: 0:00:24 Iter: 352, Train Loss: 0.00019, Train Acc: 100.00%, Val Loss: 1.5, Val Acc: 77.94%, Time: 0:00:24 Iter: 354, Train Loss: 5.6e-07, Train Acc: 100.00%, Val Loss: 0.45, Val Acc: 92.65%, Time: 0:00:25 Iter: 356, Train Loss: 0.069, Train Acc: 93.75%, Val Loss: 0.12, Val Acc: 91.18%, Time: 0:00:25 Iter: 358, Train Loss: 0.35, Train Acc: 93.75%, Val Loss: 0.17, Val Acc: 92.65%, Time: 0:00:25 Iter: 360, Train Loss: 3.7e-07, Train Acc: 100.00%, Val Loss: 1.1, Val Acc: 82.35%, Time: 0:00:25 Iter: 362, Train Loss: 0.18, Train Acc: 96.88%, Val Loss: 2.5, Val Acc: 63.24%, Time: 0:00:25 Iter: 364, Train Loss: 7.5e-09, Train Acc: 100.00%, Val Loss: 3.3, Val Acc: 55.88%, Time: 0:00:25 Iter: 366, Train Loss: 3.7e-09, Train Acc: 100.00%, Val Loss: 3.9, Val Acc: 51.47%, Time: 0:00:25 Iter: 368, Train Loss: 1.3e-08, Train Acc: 100.00%, Val Loss: 4.2, Val Acc: 50.00%, Time: 0:00:26

flitdu commented 4 years ago

你这到底能不能迁移???

gaussic commented 4 years ago

如果你是带着学习的心态来的,那我很乐意解答你的问题。 验证集的波动很大。不从各个层面找原因,来这里吐口水,这是傲慢和无理的,这也充分看出了你急于求成的表现。

这样的态度这里不欢迎,麻烦右转,谢谢。