Robert McHardy

Julien Abadji, Marah Abdin, Connor Adams et al.

We present Laguna M.1 and Laguna XS.2, two Mixture-of-Experts foundation models built for long-horizon, agentic coding: M.1 has $225.8$B total parameters ($23.4$B activated per token) and XS.2 has $33.4$B total ($3$B activated). Both models were trained from scratch end-to-end inside the same internal system that we refer to as our Model Factory: a tightly-integrated stack of versioned data, training, evaluation, and inference components that turn model development into an industrial process. We describe the principles and design choices of the Model Factory and also detail the end-to-end training process of our models, throughout pre-training data and architecture, post-training stages, evaluation, and quantization. On agentic software engineering and terminal benchmarks (SWE-bench Verified, SWE-bench Multilingual, SWE-Bench Pro, and Terminal-Bench 2.0) M.1 and XS.2 are competitive with state-of-the-art open models in their respective weight classes. Laguna XS.2 weights are released under Apache~2.0 at https://huggingface.co/collections/poolside/laguna-xs2.

Jean Kaddour, Joshua Harris, Maximilian Mozes et al. · cambridge

Large Language Models (LLMs) went from non-existent to ubiquitous in the machine learning discourse within a few years. Due to the fast pace of the field, it is difficult to identify the remaining challenges and already fruitful application areas. In this paper, we aim to establish a systematic set of open problems and application successes so that ML researchers can comprehend the field's current state more quickly and become productive.

Aryo Pradipta Gema, Joshua Ong Jun Leang, Giwon Hong et al.

Maybe not. We identify and analyse errors in the popular Massive Multitask Language Understanding (MMLU) benchmark. Even though MMLU is widely adopted, our analysis demonstrates numerous ground truth errors that obscure the true capabilities of LLMs. For example, we find that 57% of the analysed questions in the Virology subset contain errors. To address this issue, we introduce a comprehensive framework for identifying dataset errors using a novel error annotation protocol. Then, we create MMLU-Redux, which is a subset of 5,700 manually re-annotated questions across all 57 MMLU subjects. We estimate that 6.49% of MMLU questions contain errors. Using MMLU-Redux, we demonstrate significant discrepancies with the model performance metrics that were originally reported. Our results strongly advocate for revising MMLU's error-ridden questions to enhance its future utility and reliability as a benchmark. https://huggingface.co/datasets/edinburgh-dawg/mmlu-redux-2.0.

Francis McCann Ramirez, Luka Chkhetiani, Andrew Ehrenberg et al. · deepmind

This paper describes AssemblyAI's industrial-scale automatic speech recognition (ASR) system, designed to meet the requirements of large-scale, multilingual ASR serving various application needs. Our system leverages a diverse training dataset comprising unsupervised (12.5M hours), supervised (188k hours), and pseudo-labeled (1.6M hours) data across four languages. We provide a detailed description of our model architecture, consisting of a full-context 600M-parameter Conformer encoder pre-trained with BEST-RQ and an RNN-T decoder fine-tuned jointly with the encoder. Our extensive evaluation demonstrates competitive word error rates (WERs) against larger and more computationally expensive models, such as Whisper large and Canary-1B. Furthermore, our architectural choices yield several key advantages, including an improved code-switching capability, a 5x inference speedup compared to an optimized Whisper baseline, a 30% reduction in hallucination rate on speech data, and a 90% reduction in ambient noise compared to Whisper, along with significantly improved time-stamp accuracy. Throughout this work, we adopt a system-centric approach to analyzing various aspects of fully-fledged ASR models to gain practically relevant insights useful for real-world services operating at scale.

Robert McHardy, Heike Adel, Roman Klinger

The automatic detection of satire vs. regular news is relevant for downstream applications (for instance, knowledge base population) and to improve the understanding of linguistic characteristics of satire. Recent approaches build upon corpora which have been labeled automatically based on article sources. We hypothesize that this encourages the models to learn characteristics for different publication sources (e.g., "The Onion" vs. "The Guardian") rather than characteristics of satire, leading to poor generalization performance to unseen publication sources. We therefore propose a novel model for satire detection with an adversarial component to control for the confounding variable of publication source. On a large novel data set collected from German news (which we make available to the research community), we observe comparable satire classification performance and, as desired, a considerable drop in publication classification performance with adversarial training. Our analysis shows that the adversarial component is crucial for the model to learn to pay attention to linguistic properties of satire.