FUTO Swipe – A new swipe typing model
摘要
该系统由三类模型组成：Encoder 是跨语言、跨键盘布局的通用模型，用于基础滑动输入预测，但精度不算最前沿；ContextLM 是按单一语言训练的轻量语言模型，仅用文本数据即可训练，用于结合上下文过滤不合理词；Decoder 则同时依赖语言与键盘布局，是专门学习某一布局特性以达到最高精度的模型，但需要对应布局的滑动输入数据，因此目前仅支持 QWERTY 英文版本。三者结合，并使用 beam width 300，在测试集上 top-4 fail rate 约为 4%；在忽略 OOV（词表外）情况后，错误率低于 1%。
荐读理由
FUTO Swipe 模型用3种模型（通用Encoder、单语言ContextLM、特定QWERTY解码器）+300 beam width，测试集top-4 fail率仅4%，OOV忽略下错误率<1%，能直接迁移到AI工程项目
原文
Models
Our architecture includes three model types.
The Encoder model is a universal layout-agnostic and language-agnostic, and is used for making swipe typing predictions in the general case. However, it does not offer cutting-edge accuracy.
The ContextLM model is a very small language model that is trained for a single language. It's used to improve the quality of predictions by eliminating nonsensical words given the preceding words in the sentence. It only requires text data for training.
Finally, the decoder is a language-specific and layout-specific model that learns layout's peculiarities and achieves leading accuracy. As it requires swipe typing data for a specific layout and language for training, we only have a QWERTY English decoder for now.
With all 3 models and with a beam width of 300, we achieve a top-4 fail rate of only ~4% on our test set. Ignoring out-of-vocabulary cases, the error rate is below 1%.
这条对你有帮助吗？