ziweipolaris / atec2018-nlp Goto Github PK

您好！
看了您的代码，收获很大。请问词向量是你们在这个数据集上训练的吗？有没有去停用词或者其他处理呢？我用的 gensim 只去掉标点，window_size = 5 训练的，只使用了 decom 的几个融合的提交测试了下。在第一赛季的B榜上（新人学习赛）只有 0.53+，按照 README 应该有 0.6+ 所以向您请教下是不是我哪里处理有问题。
多谢了！

您好，刚在公众号上看到了这篇推送，去官网之后发现无权下载数据，能否分享一下数据，谢谢！
我的邮箱为：[email protected]

你好；
看了您的代码很有启发，我自己在跑这个模型的时候，只用了siamase网络这个，然后用的字向量，用的网上的。我有几个问题想请教一下哈？
1.首先就是我这边用的10万条数据训练的，但是有数据不平衡问题，想请教一下是怎么解决的，
2.还有就是我在siamase网络上的验证集F1只有0.4几，提交了一次后是0.51，我想问一下这个结果正常吗？
3.还有就是想问一下用多少个epoch比较合适

当用字向量siamase网络训练时出现OOM错误, GPU内存5G, tensorflow版本是1.13.1, keras版本是2.2.4, 请问是我的GPU内存不够吗？如果是，请问你的机器GPU内存是多少？谢谢
Train on 99915 samples, validate on 2562 samples
Epoch 1/2
Traceback (most recent call last):
File "pai_train.py", line 888, in
train_all_models(index=0)
File "pai_train.py", line 811, in train_all_models
train_model(model, swa_model, cfg)
File "pai_train.py", line 794, in train_model
fit()
File "pai_train.py", line 786, in fit
epochs=n_epoch,verbose=2)
File "/home/cc/.local/lib/python3.6/site-packages/keras/engine/training.py", line 1039, in fit
validation_steps=validation_steps)
File "/home/cc/.local/lib/python3.6/site-packages/keras/engine/training_arrays.py", line 199, in fit_loop
outs = f(ins_batch)
File "/home/cc/.local/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py", line 2715, in call
return self._call(inputs)
File "/home/cc/.local/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py", line 2675, in _call
fetched = self._callable_fn(*array_vals)
File "/home/cc/.local/lib/python3.6/site-packages/tensorflow/python/client/session.py", line 1439, in call
run_metadata_ptr)
File "/home/cc/.local/lib/python3.6/site-packages/tensorflow/python/framework/errors_impl.py", line 528, in exit
c_api.TF_GetCode(self.status.status))
tensorflow.python.framework.errors_impl.ResourceExhaustedError: OOM when allocating tensor with shape[635974,300] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc
[[{{node training/Adam/mul_1}}]]
Hint: If you want to see a list of allocated tensors when OOM happens, add report_tensor_allocations_upon_oom to RunOptions for current allocation info.

 [[{{node metrics/acc/Mean_1}}]]

Hint: If you want to see a list of allocated tensors when OOM happens, add report_tensor_allocations_upon_oom to RunOptions for current allocation info.