The quantum natural language processing team at Quantinuum, an integrated quantum computing company, has released a major update to its open-source Python library and toolkit, λambeq (pronounced 'Lambek').
λambeq converts any natural language sentence into a quantum circuit, ready to be realised on a quantum computer. The new release has been designed for a growing community of researchers, developers and users versed in quantum natural language processing (QNLP) and natural language processing (NLP). Natural language processing markets are projected to grow 27 per cent annually over the next five years. [1]
The update will support the growth of QNLP and potential future applications such as automated dialogue, text mining, language translation, text-to-speech, language generation and bioinformatics.
Quantinuum's head of applied quantum NLP research, Dr Dimitrios Kartsaklis, said: ‘Since we launched λambeq, we have received valuable feedback from a rapidly growing community of users, and many of the new features available today reflect this. The new version of λambeq now comes, for example, with a native state-of-the-art parser that has been fully integrated with the toolkit. Additionally, the toolkit is now equipped with a training package that supports popular supervised learning libraries, such as PyTorch, to help users efficiently train NLP tasks using the quantum circuits and tensor networks that λambeq generates. This update is all about accessibility – and crucially, reducing the time it takes to achieve results.’
Additionally, and importantly, λambeq's new neural-based CCG parser, Bobcat, is trained on a large human-annotated corpus of syntactic derivations. It is fully integrated with the toolkit, simplifying the installation process, and presents improved state-of-the-art parsing performance. The previous parser remains part of the toolkit, and for the benefit of the community, Bobcat will also be released as a separate stand-alone open-source tool in due course.
The new update is equipped with a command-line interface, making most of the toolkit's functionality available to users with no programming knowledge. It also contains a new supervised training module designed to simplify the process of training parameterised quantum circuits and tensor networks in a machine learning setup.
λambeq is the first quantum NLP and computational linguistics toolkit. It can convert a sentence into a quantum circuit that inherits its entanglement structure from the sentence's syntactic structure. This construction is motivated by formal mathematical correspondences between mathematical models of grammar and quantum protocols, as established by senior researchers at Quantinuum, Chief Scientist Prof. Bob Coecke and Head of AI Prof. Stephen Clark.
With this update, λambeq becomes more flexible in providing users with more options on the quantum circuits it can generate. It allows manipulation of syntax diagrams and makes it simpler to define the quantum circuits from the syntactic structure.
The visualisation of λambeq's output has also been improved, and documentation has been expanded with numerous examples to remove the barrier to entry for general users.
