NEWS-TEXT-CLASSIFICATION icon indicating copy to clipboard operation
NEWS-TEXT-CLASSIFICATION copied to clipboard

零基础入门NLP - 新闻文本分类 正式赛第一名方案

Results 6 NEWS-TEXT-CLASSIFICATION issues
Sort by recently updated
recently updated
newest added

如题,能提供下新的下载密码嘛?万分感谢

详细错误是: ``` session.py 1356 _do_call return fn(*args) session.py 1339 _run_fn self._extend_graph() session.py 1374 _extend_graph tf_session.ExtendSession(self._session) tensorflow.python.framework.errors_impl.InvalidArgumentError ------------------------------------------------------------------------------- pretraining.py 164 callbacks=[checkpoint, csv_logger], training.py 649 fit validation_freq=validation_freq) training_distributed.py 143 fit_distributed steps_name='steps_per_epoch') training_arrays.py...

大佬 用keras_bert 的预训练模型微调效果和这个一样吗

我看了下天池给的方案,是 bert + lstm + attn,您的方案是直接bert + atten。想问下,您选择这样的方案的原因是啥呢?这个和 加了 lstm 后的差异点是啥