Open argman opened 6 years ago
Yes, you are right, attention-based encoder-decoder should be better than lstm+ctc when modeling the language model. Generating Chinese dataset is more complicated than you think, but I will share my simplified code soon.
大神,很像知道你的训练数据如何将背景与字符进行合成,拉伸等操作的?能提供下脚本拜读么?
什么时候开放数据生成代码 @senlinuc
想测试一下效果编译都不过谁能给发个编译好的 393777485@qq.com
Really nice project btw!
the reason bi-lstm does not affect accuracy because lstm is more likely to model the language model(especially works for English words), but in Chinese, it depends on how you generate your data.