Multi-agent systems (MAS) often achieve higher reasoning accuracy than single models, but their reliance on repeated debates across agents makes them computationally expensive. We introduce SMAGDi, a distillation framework that transfers the debate dynamics of a five-agent Llama-based MAS into a compact Socratic decomposer-solver student. SMAGDi represents debate traces as directed interaction graphs, where nodes encode intermediate reasoning steps with correctness labels and edges capture continuity and cross-agent influence. The student is trained with a composite objective combining language modeling, graph-based supervision, contrastive reasoning, and embedding alignment to preserve both fluency and structured reasoning. On StrategyQA and MMLU, SMAGDi compresses a 40B multi-agent system into a 6B student while retaining 88% of its accuracy, substantially outperforming prior distillation methods such as MAGDi, standard KD, and fine-tuned baselines. These results highlight that explicitly modeling interaction graphs and Socratic decomposition enable small models to inherit the accuracy benefits of multi-agent debate while remaining efficient enough for real-world deployment.

Electricity price forecasting has become a critical tool for decision-making in energy markets, particularly as the increasing penetration of renewable energy introduces greater volatility and uncertainty. Historically, research in this field has been dominated by point forecasting methods, which provide single-value predictions but fail to quantify uncertainty. However, as power markets evolve due to renewable integration, smart grids, and regulatory changes, the need for probabilistic forecasting has become more pronounced, offering a more comprehensive approach to risk assessment and market participation. This paper presents a review of probabilistic forecasting methods, tracing their evolution from Bayesian and distribution based approaches, through quantile regression techniques, to recent developments in conformal prediction. Particular emphasis is placed on advancements in probabilistic forecasting, including validity-focused methods which address key limitations in uncertainty estimation. Additionally, this review extends beyond the Day-Ahead Market to include the Intra-Day and Balancing Markets, where forecasting challenges are intensified by higher temporal granularity and real-time operational constraints. We examine state of the art methodologies, key evaluation metrics, and ongoing challenges, such as forecast validity, model selection, and the absence of standardised benchmarks, providing researchers and practitioners with a comprehensive and timely resource for navigating the complexities of modern electricity markets.

The memorization of training data in large language models (LLMs) poses significant privacy and copyright concerns. Existing data extraction methods, particularly heuristic-based divergence attacks, often exhibit limited success and offer limited insight into the fundamental drivers of memorization leakage. This paper introduces Confusion-Inducing Attacks (CIA), a principled framework for extracting memorized data by systematically maximizing model uncertainty. We empirically demonstrate that the emission of memorized text during divergence is preceded by a sustained spike in token-level prediction entropy. CIA leverages this insight by optimizing input snippets to deliberately induce this consecutive high-entropy state. For aligned LLMs, we further propose Mismatched Supervised Fine-tuning (SFT) to simultaneously weaken their alignment and induce targeted confusion, thereby increasing susceptibility to our attacks. Experiments on various unaligned and aligned LLMs demonstrate that our proposed attacks outperform existing baselines in extracting verbatim and near-verbatim training data without requiring prior knowledge of the training data. Our findings highlight persistent memorization risks across various LLMs and offer a more systematic method for assessing these vulnerabilities.

Existing speech models suffer from competing requirements on token representations by understanding and generation tasks. This discrepancy in representation prevents speech language models from performing instruction-based free-form editing. To solve this challenge, we introduce a novel framework that unifies speech understanding, generation, and editing. The core of our unified model is a unified continuous speech tokenizer MingTok-Audio, the first continuous tokenizer to effectively integrate semantic and acoustic features, which makes it suitable for both understanding and generation tasks. Based on this unified continuous audio tokenizer, we developed the speech language model Ming-UniAudio, which achieved a balance between generation and understanding capabilities. Ming-UniAudio sets new state-of-the-art (SOTA) records on 8 out of 12 metrics on the ContextASR benchmark. Notably, for Chinese voice cloning, it achieves a highly competitive Seed-TTS-WER of 0.95. Leveraging this foundational model, we further trained a dedicated speech editing model Ming-UniAudio-Edit, the first speech language model that enables universal, free-form speech editing guided solely by natural language instructions, handling both semantic and acoustic modifications without timestamp condition. To rigorously assess the editing capability and establish a foundation for future research, we introduce Ming-Freeform-Audio-Edit, the first comprehensive benchmark tailored for instruction-based free-form speech editing, featuring diverse scenarios and evaluation dimensions spanning semantic correctness, acoustic quality, and instruction alignment. We open-sourced the continuous audio tokenizer, the unified foundational model, and the free-form instruction-based editing model to facilitate the development of unified audio understanding, generation, and manipulation.

This chapter bridges technical analysis and organizational preparedness by tracing the path from layered failure modes to reliability awareness in generative and agentic AI systems. We first introduce an 11-layer failure stack, a structured framework for identifying vulnerabilities ranging from hardware and power foundations to adaptive learning and agentic reasoning. Building on this, the chapter demonstrates how failures rarely occur in isolation but propagate across layers, creating cascading effects with systemic consequences. To complement this diagnostic lens, we develop the concept of awareness mapping: a maturity-oriented framework that quantifies how well individuals and organizations recognize reliability risks across the AI stack. Awareness is treated not only as a diagnostic score but also as a strategic input for AI governance, guiding improvement and resilience planning. By linking layered failures to awareness levels and further integrating this into Dependability-Centred Asset Management (DCAM), the chapter positions awareness mapping as both a measurement tool and a roadmap for trustworthy and sustainable AI deployment across mission-critical domains.

Vision Transformers (ViTs), such as DINOv2, achieve strong performance across domains but often repurpose low-informative patch tokens in ways that reduce the interpretability of attention and feature maps. This challenge is especially evident in medical imaging, where domain shifts can degrade both performance and transparency. In this paper, we introduce Randomized-MLP (RMLP) regularization, a contrastive learning-based method that encourages more semantically aligned representations. We use RMLPs when fine-tuning DINOv2 to both medical and natural image modalities, showing that it improves or maintains downstream performance while producing more interpretable attention maps. We also provide a mathematical analysis of RMLPs, offering insights into its role in enhancing ViT-based models and advancing our understanding of contrastive learning.

We present a systematic, empirical evaluation of five local large language model (LLM) runtimes on Apple Silicon: MLX, MLC-LLM, llama.cpp, Ollama, and PyTorch MPS. Experiments were conducted on a Mac Studio equipped with an M2 Ultra processor and 192 GB of unified memory. Using the Qwen-2.5 model family across prompts ranging from a few hundred to 100,000 tokens, we measure time-to-first-token (TTFT), steady-state throughput, latency percentiles, long-context behavior (key-value and prompt caching), quantization support, streaming performance, batching and concurrency behavior, and deployment complexity. Under our settings, MLX achieves the highest sustained generation throughput, while MLC-LLM delivers consistently lower TTFT for moderate prompt sizes and offers stronger out-of-the-box inference features. llama.cpp is highly efficient for lightweight single-stream use, Ollama emphasizes developer ergonomics but lags in throughput and TTFT, and PyTorch MPS remains limited by memory constraints on large models and long contexts. All frameworks execute fully on-device with no telemetry, ensuring strong privacy guarantees. We release scripts, logs, and plots to reproduce all results. Our analysis clarifies the design trade-offs in Apple-centric LLM deployments and provides evidence-based recommendations for interactive and long-context processing. Although Apple Silicon inference frameworks still trail NVIDIA GPU-based systems such as vLLM in absolute performance, they are rapidly maturing into viable, production-grade solutions for private, on-device LLM inference.

Benchmarks play a significant role in how researchers and the public understand generative AI systems. However, the widespread use of benchmark scores to communicate about model capabilities has led to criticisms of validity, especially whether benchmarks test what they claim to test (i.e. construct validity) and whether benchmark evaluations are representative of how models are used in the wild (i.e. ecological validity). In this work we explore how to create an LLM benchmark that addresses these issues by taking a human-centered approach. We focus on designing a domain-oriented benchmark for journalism practitioners, drawing on insights from a workshop of 23 journalism professionals. Our workshop findings surface specific challenges that inform benchmark design opportunities, which we instantiate in a case study that addresses underlying criticisms and specific domain concerns. Through our findings and design case study, this work provides design guidance for developing benchmarks that are better tuned to specific domains.

Oscar nominations are an important factor in the movie industry because they can boost both the visibility and the commercial success. This work explores whether it is possible to predict Oscar nominations for screenplays using modern language models. Since no suitable dataset was available, a new one called Movie-O-Label was created by combining the MovieSum collection of movie scripts with curated Oscar records. Each screenplay was represented by its title, Wikipedia summary, and full script. Long scripts were split into overlapping text chunks and encoded with the E5 sentence em bedding model. Then, the screenplay embed dings were classified using a logistic regression model. The best results were achieved when three feature inputs related to screenplays (script, summary, and title) were combined. The best-performing model reached a macro F1 score of 0.66, a precision recall AP of 0.445 with baseline 0.19 and a ROC-AUC of 0.79. The results suggest that even simple models based on modern text embeddings demonstrate good prediction performance and might be a starting point for future research.

Given that conventional recommenders, while deeply effective, rely on large distributed systems pre-trained on aggregate user data, incorporating new data necessitates large training cycles, making them slow to adapt to real-time user feedback and often lacking transparency in recommendation rationale. We explore the performance of smaller personal models trained on per-user data using weightless neural networks (WNNs), an alternative to neural backpropagation that enable continuous learning by using neural networks as a state machine rather than a system with pretrained weights. We contrast our approach against a classic weighted system, also on a per-user level, and standard collaborative filtering, achieving competitive levels of accuracy on a subset of the MovieLens dataset. We close with a discussion of how weightless systems can be developed to augment centralized systems to achieve higher subjective accuracy through recommenders more directly tunable by end-users.