-
我在自己的数据集上试了一下,bert的效果大概是85%,textCNN是79%, 然后用蒸馏大概只有77.8%.
蒸馏相关的两个参数都是按照你代码里的:
self.T = 10 # 调整温度
self.alpha = 0.9 # 调整soft_target loss 和 hard_target loss 比重
-
Hi, I get the following error while running with yelp_review_polarity, I'm quite new to ML. Kindly look into the error and help me, thanks in advance! I'm using tensorflow: 1.13.1 and python 3.7
2…
-
为什么BERT后面接了CNN / LSTM效果更好呢?是因为BERT encode能力不够吗?(显然不是)这其中有什么原理吗?谢谢
-
报错:OSError: Unable to create file (unable to open file: name = './model/best_model\model_01-0.93.hdf5', errno = 2, error message = 'No such file or directory', flags = 13, o_flags = 302)
发现是源文件中./tex…
-
wykdg updated
4 years ago
-
Hi,
Thanks for the excellent piece of work . But I'm getting the following error while I try to initiate the function call in this line . Not sure whats wrong .
Message : Expected int32 but got…
-
File "/home/wt/桌面/class/text_classification-master/a02_TextCNN/other_experiement/data_util_zhihu.py", line 28, in create_voabulary
model=word2vec.load(word2vec_model_path,kind='bin')
File "/us…
-
请问一下博主知道有ELMO+ESIM的组合开源项目吗?我搜了目前网上开源的项目,包括提出ELMO+ESIM这个思路的原论文Deep contextualized word representations也没有开源这个组合的代码,我想问一下博主您在这个项目中用到的ELMO+textCNN是不是也是类似ELMO+ESIM的这个架构?多谢了
-
你好。很感谢您提供的代码教程。目前发现transformer训练准确度低。只有不到60%,大大低于TextCNN的85%,请问您当时测试也是如此吗?若是提高,可以给点建议吗?谢谢
text_classifier/models/transformer.py
-
ValueError: Error when checking input: expected input_22 to have 2 dimensions, but got array with shape (9598, 61, 300)
根据你的textcnn,在model.fit传入参数时,由于我的train_data是三维的[N, max_len, w2v_dim],而你的是二维的,产生了…