In this paper we explore hallucinations and related capability limitations in LLMs and LLM - based agents from the perspective of computational complexity . We show that beyond a certain complexity, LLMs are incapable of carrying out computational and agentic tasks or verifying the ir accuracy . Introduction With widespread adoption of transformer - based language models ("LLMs") in AI, there is significant interest in the limits of LLMs' capabilities, specifically so - called "hallucinations", occurrences in which LLMs provide spurious, factually incorrect or nonsensical [1, 2] information when prompted on certain subjects. Further more, there is growing interest in "agentic" uses of LLMs - that is, using LLMs to create "agents" that act autonomously or semi - autonomously to carry out various tasks, including tasks with applications in the real world. This makes it important to understand the types of tasks LLMs can and cannot perform.

Machine learning programs, such as those performing inference, fine-tuning, and training of LLMs, are commonly delegated to untrusted compute providers. To provide correctness guarantees for the client, we propose adapting the cryptographic notion of refereed delegation to the machine learning setting. This approach enables a computationally limited client to delegate a program to multiple untrusted compute providers, with a guarantee of obtaining the correct result if at least one of them is honest. Refereed delegation of ML programs poses two technical hurdles: (1) an arbitration protocol to resolve disputes when compute providers disagree on the output, and (2) the ability to bitwise reproduce ML programs across different hardware setups, For (1), we design Verde, a dispute arbitration protocol that efficiently handles the large scale and graph-based computational model of modern ML programs. For (2), we build RepOps (Reproducible Operators), a library that eliminates hardware "non-determinism" by controlling the order of floating point operations performed on all hardware. Our implementation shows that refereed delegation achieves both strong guarantees for clients and practical overheads for compute providers.

We introduce ADEPT: Adaptive Data ExPloiTation, a simple yet powerful framework to enhance the **data efficiency** and **generalization** in deep reinforcement learning (RL). Specifically, ADEPT adaptively manages the use of sampled data across different learning stages via multi-armed bandit (MAB) algorithms, optimizing data utilization while mitigating overfitting. Moreover, ADEPT can significantly reduce the computational overhead and accelerate a wide range of RL algorithms. We test ADEPT on benchmarks including Procgen, MiniGrid, and PyBullet. Extensive simulation demonstrates that ADEPT can achieve superior performance with remarkable computational efficiency, offering a practical solution to data-efficient RL. Our code is available at https://github.com/yuanmingqi/ADEPT.

Radio frequency interference (RFI) is a persistent contaminant in terrestrial radio astronomy. While new radio interferometers are becoming operational, novel sources of RFI are also emerging. In order to strengthen the mitigation of RFI in modern radio interferometers, we propose an on-line RFI mitigation scheme that can be run in the correlator of such interferometers. We combine statistics based on the energy as well as the polarization alignment of the correlated signal to develop an on-line RFI mitigation scheme that can be applied to a data stream produced by the correlator in real-time, especially targeted at low duty-cycle or transient RFI detection. In order to improve the computational efficiency, we explore the use of both single precision and half precision floating point operations in implementing the RFI mitigation algorithm. This ideally suits its deployment in accelerator computing devices such as graphics processing units (GPUs) as used by the LOFAR correlator. We provide results based on real data to demonstrate the efficacy of the proposed method.

Urban planners in Vienna, Austria, installed their first smart traffic lights specifically designed to increase pedestrian safety in 2018. After years of analysis and improvement, the Graz University of Technology (TU Graz) researchers have now rolled out a second generation of exponentially more complex, deep learning-based software to 21 lights at four crosswalks. Unlike its predecessor, however, the new system is programmed to provide greater help to pedestrians with walking aids, wheelchairs, and even baby strollers. People with disabilities are disproportionately at risk when crossing busy streets. Pedestrians using wheelchairs, for example, are 36 percent more likely to die in a car-related accident when compared to victims struck while standing.

Along with the development of technology, there are new discoveries, especially in the field of data science. One of the machine learning methods applied in data science is image processing, aka image processing. The application of image processing is closely related to everyday life. A simple example in image processing is the face detection feature on our cellphones, object detection to label a product (product detection), motor vehicle number plate detection (text extraction), and others. An example of the application of natural language processing that we usually use is machine translation, such as in Google Translate.

Graphcore has released significant new features for the PopVisionTM family of analysis tools as part of a major Poplar software stack update, Poplar SDK 2.0. We created the PopVision Graph Analyser and System Analyser to help developers maximise the performance of their applications on IPU systems. To mark this update, we are looking at how PopVision tools can be used most effectively to inform and optimise machine learning programs. With its massively parallel architecture, the IPU has been specifically built from the ground up for machine intelligence workloads and is therefore able to deliver state of the art performance across even the most complex AI models. For this reason, many of our users are usually not just looking to run standard machine intelligence models, but to exploit the highest possible performance from IPU systems, beyond what they have been able to achieve with other systems.

Computers are crunching more numbers than ever to crack the most complex problems of our time -- how to cure diseases like COVID and cancer, mitigate climate change and more. These and other grand challenges ushered computing into today's exascale era when top performance is often measured in exaflops. An exaflop is a measure of performance for a supercomputer that can calculate at least 1018 or one quintillion floating point operations per second. In exaflop, the exa- prefix means a quintillion, that's a billion billion, or one followed by 18 zeros. Similarly, an exabyte is a memory subsystem packing a quintillion bytes of data.