λambeq Gen II: A Quantum-Enhanced Interpretable and Scalable Text-based NLP Software Package
By Bob Coecke and Dimitri Kartsaklis
Introduction
Today we announce the next generation of λambeq , Quantinuum’s quantum natural language processing (QNLP) package.
Incorporating recent developments in both quantum NLP and quantum hardware, λambeq Gen II allows users not only to model the semantics of natural language (in terms of vectors and tensors), but to convert linguistic structures and meaning directly into quantum circuits for real quantum hardware.
Five years ago, our team reported the first realization of Quantum Natural Language Processing (QNLP). In their work, the team realized that there is a direct correspondence between the meanings of words and quantum states, and between grammatical structures and quantum entanglement. As that article put it: “Language is effectively quantum native”.
Our team realized an NLP task on quantum hardware and provided the data and code via a GitHub repository, attracting the interest of a then-nascent quantum NLP community, which has since grown around successive releases of λambeq. We released it 18 months later, supported by a research paper on the arXiv.
Λambeq: an open-source python library that turns sentences into quantum circuits, and then feeds these to quantum computers subject to VQC methodologies. Initial release in October 2021 arXiv:2110.04236
From that moment onwards, anyone could play around with QNLP on the then freely available quantum hardware. Our λambeq software has been downloaded over 50,000 times, and the user community is supported by an active Discord page, where practitioners can interact with each other and with our development team.
The QNLP Back-Story
In order to demonstrate that QNLP was possible, even on the hardware available in 2021, we focused exclusively on small noisy quantum computers. Our motivation was to produce some exploratory findings, looking for a potential quantum advantage for natural language processing using quantum hardware. We published our original scientific work in 2016, detailing a quadratic speedup over classical computers (in certain circumstances). We are strongly convinced that there is a lot more potential than indicated in that paper.
That first realization of QNLP marked a shift away from brute-force machine learning, which has now taken the world by storm in the shape of large language models (LLMs) running on algorithms called “transformers”.
Instead of the transformer approach, we decoded linguistic structure using a compositional theory of meaning. With deep roots in computational linguistics, our approach was inspired by research into compositional linguistic algorithms, and their resemblance to other quantum primitives such as quantum teleportation. As we continued our work, it became clear that our approach reduced training requirements by relying on a natural relationship between linguistic structure and quantum structure, offering near-term QNLP in practice.
Embedding recent progress in λambeq Gen II
We haven’t sat still, and neither have the teams working in the field of quantum hardware. Quantinuum’s stack now performs at a level we only dreamed of in 2020. While we look forward to continued progress on the hardware front, we are getting ahead of these future developments by shifting the focus in our algorithms and software packages, to ensure we and λambeq’s users are ready to chase far more ambitious goals!
We moved away from the compositional theory of meaning that was the focus of our early experiments, called DisCoCat, to a new mathematical foundation called DisCoCirc. This enabled us to explore the relationship between text generation and text circuits, concluding that “text circuits are generative for text”.
Formally speaking, DisCoCirc embraces substantially more compositional structure present in language than DisCoCat does, and that pays off in many ways:
- Firstly, the new theoretical backbone enables one to compose the structure of sentences into text structure, so we can now deal with large texts.
- Secondly, the compositional structure of language is represented in a compressed manner, that, in fact, makes the formalism language-neutral, as reported in this blog post.
- Thirdly, the augmented compositional linguistic structure, together with the requirement of learnability, makes a quantum model now canonical, and we now have solid theoretical evidence for genuine enhanced performance on quantum hardware, as shown in this arXiv paper.
- Fourthly, the problems associated with trainability of quantum machine learning models vanish, thanks to compositional generalization, which was the subject of this paper.
- Lastly, and surely not least, we reported on the achievement of compositional interpretability and explored the myriad ways that it supports explainable AI (XAI), which we also discussed extensively in this blog post.
Today, our users can benefit from these recent developments with the release λambeq Gen II. Our open-source tools have always benefited from the attention and feedback we receive from our users. Please give it a try, and we look forward to hearing your feedback on λambeq Gen II.
Enjoy!
About Quantinuum
Quantinuum, the world’s largest integrated quantum company, pioneers powerful quantum computers and advanced software solutions. Quantinuum’s technology drives breakthroughs in materials discovery, cybersecurity, and next-gen quantum AI. With over 500 employees, including 370+ scientists and engineers, Quantinuum leads the quantum computing revolution across continents.
Quantinuum is focusing on redefining what’s possible in hybrid quantum–classical computing by integrating Quantinuum’s best-in-class systems with high-performance NVIDIA accelerated computing to create powerful new architectures that can solve the world’s most pressing challenges.
The launch of Helios, Powered by Honeywell, the world’s most accurate quantum computer, marks a major milestone in quantum computing. Helios is now available to all customers through the cloud or on-premise deployment, launched with a go-to-market offering that seamlessly pairs Helios with the NVIDIA Grace Blackwell platform, targeting specific end markets such as drug discovery, finance, materials science, and advanced AI research.
We are also working with NVIDIA to adopt NVIDIA NVQLink, an open system architecture, as a standard for advancing hybrid quantum-classical supercomputing. Using this technology with Quantinuum Guppy and the NVIDIA CUDA-Q platform, Quantinuum has implemented NVIDIA accelerated computing across Helios and future systems to perform real-time decoding for quantum error correction.
In an industry-first demonstration, an NVIDIA GPU-based decoder integrated in the Helios control engine improved the logical fidelity of quantum operations by more than 3% — a notable gain given Helios’ already exceptionally low error rate. These results demonstrate how integration with NVIDIA accelerated computing through NVQLink can directly enhance the accuracy and scalability of quantum computation.
This unique collaboration spans the full Quantinuum technology stack. Quantinuum’s next-generation software development environment allows users to interleave quantum and GPU-accelerated classical computations in a single workflow. Developers can build hybrid applications using tools such as NVIDIA CUDA-Q, NVIDIA CUDA-QX, and Quantinuum’s Guppy, to make advanced quantum programming accessible to a broad community of innovators.
The collaboration also reaches into applied research through the NVIDIA Accelerated Quantum Computing Research Center (NVAQC), where an NVIDIA GB200 NVL72 supercomputer can be paired with Quantinuum’s Helios to further drive hybrid quantum-GPU research, including the development of breakthrough quantum-enhanced AI applications.
A recent achievement illustrates this potential: The ADAPT-GQE framework, a transformer-based Generative Quantum AI (GenQAI) approach, uses a Generative AI model to efficiently synthesize circuits to prepare the ground state of a chemical system on a quantum computer. Developed by Quantinuum, NVIDIA, and a pharmaceutical industry leader—and leveraging NVIDIA CUDA-Q with GPU-accelerated methods—ADAPT-GQE achieved a 234x speed-up in generating training data for complex molecules. The team used the framework to explore imipramine, a molecule crucial to pharmaceutical development. The transformer was trained on imipramine conformers to synthesize ground state circuits at orders of magnitude faster than ADAPT-VQE, and the circuit produced by the transformer was run on Helios to prepare the ground state using InQuanto, Quantinuum's computational chemistry platform.
From collaborating on hardware and software integrations to GenQAI applications, the collaboration between Quantinuum and NVIDIA is building the bridge between classical and quantum computing and creating a future where AI becomes more expansive through quantum computing, and quantum computing becomes more powerful through AI.
By Dr. Noah Berthusen
The earliest works on quantum error correction showed that by combining many noisy physical qubits into a complex entangled state called a "logical qubit," this state could survive for arbitrarily long times. QEC researchers devote much effort to hunt for codes that function well as "quantum memories," as they are called. Many promising code families have been found, but this is only half of the story.
Being able to keep a qubit around for a long time is one thing, but to realize the theoretical advantages of quantum computing we need to run quantum circuits. And to make sure noise doesn't ruin our computation, these circuits need to be run on the logical qubits of our code. This is often much more challenging than performing gates on the physical qubits of our device, as these "logical gates" often require many physical operations in their implementation. What's more, it often is not immediately obvious which logical gates a code has, and so converting a physical circuit into a logical circuit can be rather difficult.
Some codes, like the famous surface code, are good quantum memories and also have easy logical gates. The drawback is that the ratio of physical qubits to logical qubits (the "encoding rate") is low, and so many physical qubits are required to implement large logical algorithms. High-rate codes that are good quantum memories have also been found, but computing on them is much more difficult. The holy grail of QEC, so to speak, would be a high-rate code that is a good quantum memory and also has easy logical gates. Here, we make progress on that front by developing a new code with those properties.
Building on prior error correcting codes
A recent work from Quantinuum QEC researchers introduced genon codes. The underlying construction method for these codes, called the "symplectic double cover," also provided a way to obtain logical gates that are well suited for Quantinuum's QCCD architecture. Namely, these "SWAP-transversal" gates are performed by applying single qubit operations and relabeling the physical qubits of the device. Thanks to the all-to-all connectivity facilitated through qubit movement on the QCCD architecture, this relabeling can be done in software essentially for free. Combined with extremely high fidelity (~1.2 x10-5) single-qubit operations, the resulting logical gates are similarly high fidelity.
Given the promise of these codes, we take them a step further in our new paper. We combine the symplectic double codes with the [[4,2,2]] Iceberg code using a procedure called "code concatenation". A concatenated code is a bit like nesting dolls, with an outer code containing codes within it---with these too potentially containing codes. More technically, in a concatenated code the logical qubits of one code act as the physical qubits of another code.
The new codes, which we call "concatenated symplectic double codes", were designed in such a way that they have many of these easily-implementable SWAP-transversal gates. Central to its construction, we show how the concatenation method allows us to "upgrade" logical gates in terms of their ease of implementation; this procedure may provide insights for constructing other codes with convenient logical gates. Notably, the SWAP-transversal gate set on this code is so powerful that only two additional operations (logical T and S) are necessary for universal computation. Furthermore, these codes have many logical qubits, and we also present numerical evidence to suggest that they are good quantum memories.
Concatenated symplectic double codes have one of the easiest logical computation schemes, and we didn’t have to sacrifice rate to achieve it. Looking forward in our roadmap, we are targeting hundreds of logical qubits at ~ 1x 10-8 logical error rate by 2029. These codes put us in a prime position to leverage the best characteristics of our hardware and create a device that can achieve real commercial advantage.
Every year, the International Conference for High Performance Computing, Networking, Storage, and Analysis (SC) brings together the global supercomputing community to explore the technologies driving the future of computing.
Join Quantinuum at this year’s conference, taking place November 16th – 21st in St. Louis, Missouri, where we will showcase how our quantum hardware, software, and partnerships are helping define the next era of high-performance and quantum computing.
Visit Quantinuum in the Expo Hall
The Quantinuum team will be on-site at booth #4432 to showcase how we’re building the bridge between HPC and quantum.
- Live demo unit of our quantum hardware
- Our new Helios replica, providing an up-close look at the design behind our next-generation system
- The Helios chip, highlighting the innovation driving the world’s most advanced trapped-ion quantum computers
On Tuesday and Wednesday, our quantum computing experts will host daily tutorials at our booth on Helios, our next-generation hardware platform, Nexus, our all-in-one quantum computing platform, and Hybrid Workflows, featuring the integration of NVIDIA CUDA-Q with Quantinuum Systems.
Speaking Sessions at SC25
Join our team as they share insights on the opportunities and challenges of quantum integration within the HPC ecosystem:
Panel Session: The Quantum Era of HPC: Roadmaps, Challenges and Opportunities in Navigating the Integration Frontier
November 19th | 10:30 – 12:00pm CST
During this panel session, Kentaro Yamamoto from Quantinuum, will join experts from Lawrence Berkeley National Laboratory, IBM, QuEra, RIKEN, and Pawsey Supercomputing Research Centre to explore how quantum and classical systems are being brought together to accelerate scientific discovery and industrial innovation.
BoF Session: Bridging the Gap: Making Quantum-Classical Hybridization Work in HPC
November 19th | 5:15 – 6:45pm CST
Quantum-classical hybrid computing is moving from theory to reality, yet no clear roadmap exists for how best to integrate quantum processing units (QPUs) into established HPC environments. In this Birds of a Feather discussion, co-led by Quantinuum’s Grahame Vittorini and representatives from BCS, DOE, EPCC, Inria, ORNL NVIDIA, and RIKEN we hope to bring together a global community of HPC practitioners, system architects, quantum computing specialists and workflow researchers, including participants in the Workflow Community Initiative, to assess the state of hybrid integration and identify practical steps toward scalable, impactful deployment.