The quantum natural language processing team at Quantinuum, the world’s largest, standalone quantum computing company, is excited to announce a major update to its open-source Python library and toolkit, λambeq (pronounced "Lambek"), to version 0.3.0.
This update brings several enhancements that not only improve the user experience but significantly expand the capabilities the toolkit provides to its rapidly growing user community, including a growing number of quantum developers and engineers considering natural language processing (NLP) and machine learning (ML) for the first time.
The most significant parts of today’s announcement are:
- the introduction of PennyLane integration;
- an improved model training package; and
- new tutorials to introduce quantum professionals to technical NLP and ML
Unlocking New Possibilities with PennyLane Integration
A notable feature of this update is the integration of PennyLane, a powerful quantum computing library that is widely used by engineers and developers all over the world. With PennyLane support, users of λambeq can now develop hybrid quantum-classical models using the PennyLaneModel, hooking numerically determined gradients of parameterized quantum circuits (PQCs) to modules of ML libraries like PyTorch. This integration unlocks new possibilities for researchers and developers working on quantum natural language processing.
Enhanced Training Package for a Streamlined Experience
This update also brings significant improvements to the training package, with the addition of new λambeq-native loss functions that will help users more easily to train their models using standard implementations, oriented to classification tasks and others such as regression. This streamlined experience allows users to focus on their research and development without the need for custom loss functions. It not only enables better training performance on larger models but is also part of the team’s effort to make λambeq fully ML-enabled.
Helping Quantum Computing Engineers Explore NLP and ML
Another feature in this release is new NLP-101 tutorial support. This was developed following feedback from users, who are turning their attention to quantum NLP and quantum ML with increasing frequency. Many new λambeq users do not have a deep knowledge of techniques such as text pre-processing or the best practices required to perform successful experiments.
A technical tutorial has been published in notebook form, to help developers and engineers working in QML or QNLP to explore what is possible using λambeq.
Additional Features and Enhancements
In addition to the major features highlighted above, this update includes several other enhancements and bug fixes:
- Support for Python 3.11
- Improved fail-safety in the BobcatParser model download method
- Fixed various bugs, including issues with the SPSAOptimizer and NumpyModel tests
- Enhanced exception handling and documentation requirements
QNLP in the community
The launch of λambeq 0.3.0 update is a natural step for Quantinuum, which is not only the world’s largest standalone quantum computing company, but also the pioneer and leader in QNLP, in supporting the growth of quantum natural language processing and its applications. By continually enhancing the toolkit and providing cutting-edge integrations and resources, Quantinuum is paving the way for researchers, developers, and users in the ever-growing QNLP and NLP communities. QNLP offers us a way to take full advantage of the possibilities of advancing the boundaries of AI and truly using the promise that has been exhibited in part by Large Language Models such as GPT-4 whilst continuing to work to solve some of the well documented short-comings of such classical technologies.
About Quantinuum
Quantinuum is the world’s largest, standalone quantum computing company, formed by the combination of Honeywell Quantum Solutions’ world-leading hardware and Cambridge Quantum’s class-leading middleware and applications. Science-led and enterprise driven, Quantinuum accelerates quantum computing and the development of applications across chemistry, cybersecurity, finance, and optimization. Its focus is to create scalable and commercial quantum solutions to solve the world’s most pressing problems, in fields such as energy, logistics, climate change, and health. The company employs over 480 people including 350 scientists, at nine sites in the US, Europe, and Japan.
About Xanadu
Xanadu is a Canadian quantum computing company with the mission to build quantum computers that are useful and available to people everywhere. Founded in 2016, Xanadu has become one of the world's leading quantum hardware and software companies. The company also leads the development of PennyLane, an open-source software library for quantum computing and application development. Visit www.xanadu.ai or follow us on Twitter @XanaduAI.
About PennyLane
PennyLane is an open-source software framework for quantum machine learning, quantum chemistry, and quantum computing with the ability to run on all hardware. To find out more, visit the PennyLane website, or check out the PennyLane demos: a gallery of hands-on quantum computing content (https://pennylane.ai/qml/demonstrations.html).
