Nuclear fusion using magnetic confinement, in particular in the tokamak configuration, is a promising path towards sustainable energy. A core challenge is to shape and maintain a high-temperature plasma within the tokamak vessel. This requires high-dimensional, high-frequency, closed-loop control using magnetic actuator coils, further complicated by the diverse requirements across a wide range of plasma configurations. In this work, we introduce a previously undescribed architecture for tokamak magnetic controller design that autonomously learns to command the full set of control coils. This architecture meets control objectives specified at a high level, at the same time satisfying physical and operational constraints. This approach has unprecedented flexibility and generality in problem specification and yields a notable reduction in design effort to produce new plasma configurations. We successfully produce and control a diverse set of plasma configurations on the Tokamak à Configuration Variable1,2, including elongated, conventional shapes, as well as advanced configurations, such as negative triangularity and ‘snowflake’ configurations. Our approach achieves accurate tracking of the location, current and shape for these configurations. We also demonstrate sustained ‘droplets’ on TCV, in which two separate plasmas are maintained simultaneously within the vessel. This represents a notable advance for tokamak feedback control, showing the potential of reinforcement learning to accelerate research in the fusion domain, and is one of the most challenging real-world systems to which reinforcement learning has been applied.

Language modelling provides a step towards intelligent communication systems by harnessing large repositories of written human knowledge to better predict and understand the world. In this paper, we present an analysis of Transformer-based language model performance across a wide range of model scales -from models with tens of millions of parameters up to a 280 billion parameter model called Gopher. These models are evaluated on 152 diverse tasks, achieving state-of-the-art performance across the majority. Gains from scale are largest in areas such as reading comprehension, fact-checking, and the identification of toxic language, but logical and mathematical reasoning see less benefit. We provide a holistic analysis of the training dataset and model's behaviour, covering the intersection of model scale with bias and toxicity. Finally we discuss the application of language models to AI safety and the mitigation of downstream harms.

When trained at sufficient scale, auto-regressive language models exhibit the notable ability to learn a new language task after being prompted with just a few examples. Here, we present a simple, yet effective, approach for transferring this few-shot learning ability to a multimodal setting (vision and language). Using aligned image and caption data, we train a vision encoder to represent each image as a sequence of continuous embeddings, such that a pre-trained, frozen language model prompted with this prefix generates the appropriate caption. The resulting system is a multimodal few-shot learner, with the surprising ability to learn a variety of new tasks when conditioned on examples, represented as a sequence of multiple interleaved image and text embeddings. We demonstrate that it can rapidly learn words for new objects and novel visual categories, do visual question-answering with only a handful of examples, and make use of outside knowledge, by measuring a single model on a variety of established and new benchmarks. IntroductionAuto-regressive transformers have been shown to be very impressive models of natural language [40]. Large-scale language transformers exhibit several surprising abilities beyond that of standard text generation [4,30]. Perhaps most notably, they are few-shot learners; they can learn to perform a new task from a few examples without any further gradient updates. Equipped with this ability, these models have been shown to rapidly adapt to new tasks and styles of generation via prompting (e.g. switching from formal to informal language) [4], to quickly retrieve relevant encyclopedic or general knowledge when primed with a relevant context (e.g. answering questions such as 'When did the French Revolution begin?') [33,1,27] and to use new words in appropriate ways straight after being taught what those words mean (sometimes referred to as 'fast binding') [12,4]. Despite these impressive capabilities, such large scale language models are 'blind' to modalities other than text, preventing us from communicating visual tasks, questions or concepts to them. Indeed, philosophers and linguists have questioned whether an un-grounded language model can ever achieve true understanding of the language it processes [5,2]. Here, we present Frozen, a method for giving a pre-trained language model access to visual information in a way that extends its few-shot learning capabilities to a multimodal setting, without changing its weights. Frozen consists of a neural network trained to encode images into the word embedding space of a large pre-trained language model such that the language model generates captions for those images. The weights of the language model are kept frozen, but gradients are back-propagated through it to train the image encoder from Preprint. Under review.

PurposeThis study aimed to assess the accuracy and dosimetric impact of the Acuros XB (AXB) algorithm compared to the Anisotropic Analytical Algorithm (AAA) in two situations. First, simple phantom geometries were set and analyzed; moreover, volumetric modulated arc therapy (VMAT) clinical plans for Head & Neck and lung cases were calculated and compared.MethodsFirst, a phantom study was performed to compare the algorithms with radiochromic EBT3 film doses using one PMMA slab phantom and two others containing foam or air gap. Subsequently, a clinical study was conducted, including 20 Head & Neck and 15 lung cases irradiated with the VMAT technique. The treatment plans calculated by AXB and AAA were evaluated in terms of planning target volume (PTV) coverage (V95%), dose received by relevant organs at risk (OARs), and the impact of using AXB with a grid size of 1 mm. Finally, patient‐specific quality assurance (PSQA) was performed and compared for 17 treatment plans.ResultsPhantom dose calculations showed a better agreement of AXB with the film measurements. In the clinical study, AXB plans exhibited lower Conformity Index and PTV V95%, higher maximum PTV dose, and lower mean and minimum PTV doses for all anatomical sites. The most notable differences were detected in regions of intense heterogeneity. AXB predicted lower doses for the OARs, while the calculation time with a grid size of 1 mm was remarkably higher. Regarding PSQA, although AAA was found to exhibit slightly higher gamma passing rates, the difference did not affect the AXB treatment plan quality.ConclusionsAXB demonstrated higher accuracy than AAA in dose calculations of both phantom and clinical conditions, specifically in interface regions, making it suitable for sites with large heterogeneities. Hence, such dosimetric differences between the two algorithms should always be considered in clinical practice.