FUTO Swipe – 一款全新的滑行输入法

FUTO Swipe – 一款全新的滑行输入法
FUTO Swipe – A new swipe typing model

FUTO Swipe 是一款开源、注重隐私的滑动输入系统，旨在为那些会收集数据的专有键盘提供高性能的替代方案。该系统集成在 FUTO Keyboard 应用中，完全在本地离线运行，确保用户数据不出设备。该项目采用轻量级架构，由编码器、上下文感知语言模型和解码器组成。这些模型协同工作，在保持约 250 万个参数的极小体积的同时，实现了业界领先的准确率（排除词汇表外情况，错误率低于 1%）。这种高效性使其即使在低端设备上也能实现低延迟的性能表现。为了支持社区，FUTO 已在 HuggingFace 上发布了一个包含 100 万个遵循 MIT 协议的滑动轨迹数据集，并提供了 C++ 推理库。通过开源这些模型和算法，FUTO 旨在去中心化高质量的滑动输入技术，邀请开发者在正确署名的前提下将其集成到自己的项目中。FUTO Swipe 证明了强大且精准的文本输入无需云端追踪，展现了先进的机器学习技术既可以平易近人，又可以尊重用户隐私。

FUTO 为其 Android 键盘发布了一款全新的滑动输入模型，在 Hacker News 上引发了广泛关注与讨论。用户将其视为 Gboard 和 SwiftKey 等主流键盘的隐私友好型开源替代方案。 **主要亮点：** * **性能表现：** 早期用户普遍认为新模型有显著改进，性能已接近 Gboard 的水平。许多人对其语音转文字和文本预测功能的离线、本地优先策略表示赞赏。 * **优势：** 用户称赞其响应速度、可定制的布局，以及与离线语音模型集成的能力。这被视为减少在基础移动输入方面对大型科技公司依赖的重要一步。 * **批评与挑战：** 一些用户指出仍存在一些问题，例如缺乏多语言支持、偶尔出现的界面小瑕疵，以及相比成熟应用较高的学习成本。 * **许可争议：** 该项目采用了非标准的限制性许可证，这引起了自由开源软件（FOSS）社区部分人士的批评，他们认为这阻碍了真正的开源分发。尽管尚不完美，但社区认为 FUTO 是一款极具潜力的工具，对于那些希望在不牺牲滑动输入速度和便利性的前提下实现设备“去谷歌化”的用户来说，它正在迅速走向成熟。

原文

FUTO Swipe

Fast, accurate swipe typing system. Use it today in FUTO Keyboard, our fully offline Android keyboard app. Or download the models and build with it.

This is a serverside demo to keep this webpage small. In production, it runs on-device, with much lower latency.

For a long time, good mobile swipe typing was locked behind privacy-invasive keyboard apps or unlicensed private libraries.

FUTO Swipe is our family of open models and algorithms that aims to solve this problem. We developed this primarily for FUTO Keyboard, but we also welcome the broader community to make use of the FUTO Swipe models. As this has been a long-term investment for us, we ask that an attribution is made visible to end-users. Read license

Dataset

In August 2024, we launched a dataset collection effort on the swipe.futo.org domain to collect QWERTY English swipes. Users would voluntarily visit the webpage on their mobile phone and be given instructions and information about the dataset. After consenting, they would be given sentences, primarily from Wikipedia, and would be asked to swipe them word-by-word.

In the end, this produced over 1 million swipes. We filtered out a small set of low-quality swipes. In March 2025, we released a dataset of 1 million swipes under the MIT license, and it is available today on HuggingFace.

We made heavy use of this data to train our models and to evaluate different swipe typing systems.

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%.

Note: These numbers heavily depend on the benchmark, so real-world use may vary, but we believe we match big tech's keyboards.

Footprint

The encoder model is just 635,140 parameters, and the decoder is 304,155 extra. The biggest one is the ContextLM at 1.5 million, but 1.1 million of that is just embeddings. This brings us to 1,364,271 active parameters, or 2,494,767 total parameters.

This means the footprint of the models are very small, and the model can run on low-end devices in milliseconds. In addition, the environmental costs involved in training the models were also very low, because we never needed more than 1 workstation GPU!

C++ Library

The models themselves are only half of the story when going from a swipe to word predictions. The model predictions are not very useful on their own and it's necessary to perform a dictionary-constrained beam search to score a set of words and find the most likely candidates.

For this, we release swipe-library, a library written in C++ that handles the entire inference, decoding, and beam search part so you can easily go from swipe paths to word predictions.

Make something cool!

Want to build with FUTO Swipe?

The FUTO Swipe models are available under the FUTO Model License, and the inference library is under GPL. We are working on a paper that will detail more on the training and architecture.

FUTO Swipe – 一款全新的滑行输入法 FUTO Swipe – A new swipe typing model

Want to build with FUTO Swipe?

FUTO Swipe – 一款全新的滑行输入法
FUTO Swipe – A new swipe typing model