In this work, we survey the way in which classification is used as a sensemaking practice in cultural analytics, and assess where large language models can fit into this landscape. We identify ten tasks supported by publicly available datasets on which we empirically assess the performance of LLMs compared to traditional supervised methods, and explore the ways in which LLMs can be employed for sensemaking goals beyond mere accuracy. We find that prompt-based LLMs are competitive with traditional supervised models for established tasks, but perform less well on de novo tasks. In addition, LLMs can assist sensemaking by acting as an intermediary input to formal theory testing.

Bias in machine learning models has been a chronic problem, especially as these models influence decision-making in human society. In generative AI, such as Large Language Models, the impact of bias is even more profound compared to the classification models. LLMs produce realistic and human-like content that users may unconsciously trust, which could perpetuate harmful stereotypes to the uncontrolled public. It becomes particularly concerning when utilized in journalism or education. While prior studies have explored and quantified bias in individual AI models, no work has yet compared bias similarity across different LLMs. To fill this gap, we take a comprehensive look at ten open- and closed-source LLMs from four model families, assessing the extent of biases through output distribution. Using two datasets-one containing 4k questions and another with one million questions for each of the four bias dimensions -- we measure functional similarity to understand how biases manifest across models. Our findings reveal that 1) fine-tuning does not significantly alter output distributions, which would limit its ability to mitigate bias, 2) LLMs within the same family tree do not produce similar output distributions, implying that addressing bias in one model could have limited implications for others in the same family, and 3) there is a possible risk of training data information leakage, raising concerns about privacy and data security. Our analysis provides insight into LLM behavior and highlights potential risks in real-world deployment.

Recent advances in knowledge distillation (KD) have enabled smaller student models to approach the performance of larger teacher models. However, popular methods such as supervised KD and on-policy KD, are adversely impacted by the knowledge gaps between teacher-student in practical scenarios. Supervised KD suffers from a distribution mismatch between training with a static dataset and inference over final student-generated outputs. Conversely, on-policy KD, which uses student-generated samples for training, can suffer from low-quality training examples with which teacher models are not familiar, resulting in inaccurate teacher feedback. To address these limitations, we introduce Speculative Knowledge Distillation (SKD), a novel approach that leverages cooperation between student and teacher models to generate high-quality training data on-the-fly while aligning with the student's inference-time distribution. In SKD, the student proposes tokens, and the teacher replaces poorly ranked ones based on its own distribution, transferring high-quality knowledge adaptively. We evaluate SKD on various text generation tasks, including translation, summarization, math, and instruction following, and show that SKD consistently outperforms existing KD methods across different domains, data sizes, and model initialization strategies. Figure 1: SKD outperforms supervised and on-policy KD for our tested tasks: Assamese-to-English translation, dialogue summarization, and arithmetic reasoning. Supervised KD is trained on ground-truth outputs, while on-policy KD uses self-generated data. All models use greedy decoding for evaluation. Work done as a student researcher at Google Cloud AI Research. Left: SKD addresses the limitations of on-policy knowledge distillation (KD) by filtering out low-quality student samples and replacing them with teacher generated tokens. However, the substantial inference-time costs and memory footprint associated with LLMs present significant challenges for practical deployment (Agarwal et al., 2024). Therefore, compressing LLMs while maintaining their performance is crucial for real-time practical applications. Knowledge Distillation (KD) (Hinton et al., 2015) is a widely used method to compress LLMs by transferring knowledge from a larger teacher model to a smaller student model. Traditional KD approaches, such as supervised KD (Sanh et al., 2020) and SeqKD (Kim & Rush, 2016b), rely on a static dataset of outputs to train the student model. However, this fixed dataset can lead to a distribution mismatch between the training data and the student's generated samples at inference time, hindering the student's learning.

Characterizing a quantum system by learning its state or evolution is a fundamental problem in quantum physics and learning theory with a myriad of applications. Recently, as a new approach to this problem, the task of agnostic state tomography was defined, in which one aims to approximate an arbitrary quantum state by a simpler one in a given class. Generalizing this notion to quantum processes, we initiate the study of agnostic process tomography: given query access to an unknown quantum channel $\Phi$ and a known concept class $\mathcal{C}$ of channels, output a quantum channel that approximates $\Phi$ as well as any channel in the concept class $\mathcal{C}$, up to some error. In this work, we propose several natural applications for this new task in quantum machine learning, quantum metrology, classical simulation, and error mitigation. In addition, we give efficient agnostic process tomography algorithms for a wide variety of concept classes, including Pauli strings, Pauli channels, quantum junta channels, low-degree channels, and a class of channels produced by $\mathsf{QAC}^0$ circuits. The main technical tool we use is Pauli spectrum analysis of operators and superoperators. We also prove that, using ancilla qubits, any agnostic state tomography algorithm can be extended to one solving agnostic process tomography for a compatible concept class of unitaries, immediately giving us efficient agnostic learning algorithms for Clifford circuits, Clifford circuits with few T gates, and circuits consisting of a tensor product of single-qubit gates. Together, our results provide insight into the conditions and new algorithms necessary to extend the learnability of a concept class from the standard tomographic setting to the agnostic one.

Scaling laws predict the loss of a target machine learning model by extrapolating from easier-to-train models with fewer parameters or smaller training sets. This provides an efficient way for practitioners and researchers alike to compare pretraining decisions involving optimizers, datasets, and model architectures. Despite the widespread use of scaling laws to model the dynamics of language model training, there has been little work on understanding how to best estimate and interpret them. We collect (and release) a large-scale dataset containing losses and downstream evaluations for 485 previously published pretrained models. We use these to estimate more than 1000 scaling laws, then derive a set of best practices for estimating scaling laws in new model families. We find that fitting scaling laws to intermediate checkpoints of training runs (and not just their final losses) substantially improves accuracy, and that -- all else equal -- estimates of performance are generally most accurate when derived from other models of similar sizes. However, because there is a significant degree of variability across model seeds, training multiple small models is sometimes more useful than training a single large one. Moreover, while different model families differ scaling behavior, they are often similar enough that a target model's behavior can be predicted from a single model with the same architecture, along with scaling parameter estimates derived from other model families.

Humanitarian organizations can enhance their effectiveness by analyzing data to discover trends, gather aggregated insights, manage their security risks, support decision-making, and inform advocacy and funding proposals. However, data about violent incidents with direct impact and relevance for humanitarian aid operations is not readily available. An automatic data collection and NLP-backed classification framework aligned with humanitarian perspectives can help bridge this gap. In this paper, we present HumVI - a dataset comprising news articles in three languages (English, French, Arabic) containing instances of different types of violent incidents categorized by the humanitarian sector they impact, e.g., aid security, education, food security, health, and protection. Reliable labels were obtained for the dataset by partnering with a data-backed humanitarian organization, Insecurity Insight. We provide multiple benchmarks for the dataset, employing various deep learning architectures and techniques, including data augmentation and mask loss, to address different task-related challenges, e.g., domain expansion. The dataset is publicly available at https://github.com/dataminr-ai/humvi-dataset.

Pokémon maker confirms it was victim of hack The Pokémon CompanyPokémon is one of the world's best-known entertainment brands Pokémon maker Game Freak has confirmed it was the victim of a data leak after information appeared online over the weekend. The company, which has developed the Nintendo-exclusive video game series since 1996, said its servers were hacked in August this year. A statement said 2,606 items containing the names and email addresses of current, former and contract employees were accessed. The company did not comment on other information shared online claiming to show details of unreleased and upcoming projects. Game Freak said it would individually contact those affected where possible, and strengthen security measures to prevent similar hacks in future.

Two years ago, Jon Fath moved with his family to Portugal from the Netherlands with the sole purpose of launching a fintech startup there. "This country is brimming with talent and ambition," Fath says. "I thank Lisbon for welcoming me, along with so many other expats and entrepreneurs, so warmly." Indeed, it's no surprise that the European Commission named Lisbon as 2023's European Capital of Innovation, while the Financial Times, in partnership with Statista, ranked two Portuguese startup hubs in Europe's top ten startup hubs--including the Unicorn Factory Lisboa, which launched in 2022 and has already supported more than 820 startups and helped raise more than 1 billion ( 1.1 billion) . "Portugal offers unique advantages, such as its climate, safety, and cost of living, which make it an attractive choice over countries in central or northern Europe," says Nuno Pereira, CEO of Paynest.

In the "Startup-up, Scale-up" review report published last year, chancellor Rachel Reeves promised to make Britain the "high growth, start-up hub of the world". Now, almost six months into the new government, entrepreneurs remain encouraged by the promises made in the Labour manifesto. "The ambition embodied in Great British Energy and the 2030 decarbonization targets is precisely what we need and deserve," says Shilpika Gautam, CEO of greentech startup Opna, about Labour's energy policies. "It's high time the UK caught up with the policy and financing innovations in other countries, such as the Inflation Reduction Act in the US." Amit Gudka, founder of Field, agrees: "We welcome Labour's plans to double onshore wind, triple solar and quadruple offshore wind by 2030. These plans are ambitious, but not unrealistic, provided the Government continues to make clear policy decisions and create a stable policy and regulatory environment."

Having spent many years as second fiddle to Barcelona, Madrid surpassed its Catalan cousin in 2023 with startups securing 605 million ( 672 million) investment above Barcelona's 457 million ( 507 million). "Lots of Latin American talent is arriving thanks to the recent entrepreneur visa and talent programs run by Telefonica to bring promising startup founders from Mexico, Argentina, Columbia and Venezuela," explains Bu Haces, innovation consultant at Madrid's Impact Hub. The city has seen solid growth in transportation, mobility and fintech startups during the last three years with AI and deep tech supercharged by an astonishing 56 universities. "The business schools in particular are providing lots of startup networking opportunities, and are keen on developing an entrepreneurial ecosystem," says Miguel Arias, general partner at VC K Fund. With Meta, IBM, Google and Amazon all expanding in the city, the main worry is lack of housing stock for the flood of students, engineers and entrepreneurs.