Knowledge catalogue
As artificial intelligence moves from experimentation into production, government agencies are becoming an unexpected proving ground for large-scale workforce transformation. Faced with mounting opera
GPT-5.5 Instant is rolling out over the next two days as the default model to all ChatGPT users, and as ‘gpt-5.5-chat-latest’ in the API. Personalization improvements are rolling out to Plus and Pro u
OpenAI is rolling out GPT-5.5 Instant to all ChatGPT users as the new default model, replacing GPT-5.3 Instant . The model produces 52.5% fewer hallucinated claims than its predecessor on high-stakes
GPT-5.5 Instant is starting to roll out in ChatGPT. It’s a big upgrade, giving you smarter, clearer, and more personalized answers in a warmer, more natural tone. And it's also more concise, which we
GPT-5.5 Instant is OpenAI's faster, more efficient variant of their GPT-5.5 model, designed to deliver improved reasoning and clarity while maintaining lower latency for real-time applications. The mo
GPT-5.5 Instant is a faster, more efficient variant of OpenAI's GPT-5.5 model designed for real-time applications and lower-latency tasks. The system card documents the model's capabilities, limitatio
arXiv:2605.01203v1 Announce Type: cross Abstract: Currently, process reward models (PRMs) have exhibited remarkable potential for test-time scaling. Since large language models (LLMs) regularly genera
arXiv:2604.03190v2 Announce Type: replace Abstract: Transformer attention computes a single softmax-weighted average over values -- a one-pass estimate that cannot correct its own errors. We introduce
arXiv:2409.14500v5 Announce Type: replace Abstract: Although data that can be naturally represented as graphs is widespread in real-world applications across diverse industries, popular graph ML bench
arXiv:2603.09625v2 Announce Type: replace Abstract: Deep learning models benefit from increasing data diversity and volume, motivating synthetic data augmentation to improve existing datasets. However
arXiv:2507.07129v3 Announce Type: replace-cross Abstract: We study a constrained training regime for decoder-only Transformers in which the token interface is fixed, previously trained dense blocks ar
arXiv:2605.02443v1 Announce Type: new Abstract: Large Language Models (LLMs) have demonstrated remarkable capabilities across diverse natural language processing tasks, yet they remain susceptible to
arXiv:2605.02596v1 Announce Type: new Abstract: With each sensing modality exhibiting inherent strengths and limitations, multi-modal approaches for wearable Human Activity Recognition (HAR) are becom
// HeavySkill // One of the cleaner takes on agentic harness design I've read. They argue that what actually drives agent harness performance is not the orchestration code. It's a single inner skill:
arXiv:2604.05134v2 Announce Type: replace Abstract: We study how reasoning evolves in a language model -- from supervised fine-tuning (SFT) to reinforcement learning (RL) -- by analyzing how a set of
arXiv:2605.00865v1 Announce Type: cross Abstract: EEG based phoneme decoding is promising for brain computer interfaces, but many prior studies rely on within subject evaluation, small cohorts, or wea
arXiv:2502.16435v4 Announce Type: replace-cross Abstract: Humans develop perception through a bottom-up hierarchy: from basic primitives and Gestalt principles to high-level semantics. In contrast, cu
This post documents a user's experiment comparing ChatGPT and Gemini's ability to create perspective portrait artwork using two reference images (a front-facing photo and a profile photo) along with a
I detected a bad Agent action, what do I do about it? this is pretty much the main question that will power the future’s Human+Agent driven improvement loops Gather data -> Mine Errors -> Find out whi
i have yet to meet a single person who feels like claude code is getting exponentially better on some kind of fast take off Anthropic pays $750K/ year per senior engineer. The creator of Claude Code j
arXiv:2601.17723v2 Announce Type: replace Abstract: Implicit neural representation (INR) has become the standard approach for arbitrary-scale image super-resolution (ASSR). To date, no empirical study
arXiv:2605.01627v1 Announce Type: new Abstract: Low-rank decomposition is a compelling approach for compressing large language models, but its effectiveness hinges on selecting which singular-vector b
“Indispensable reading” If you aren’t reading my newsletter, you probably should be! Today’s dispatch is on the Dawkins Debacle and cites @GaryMarcus, who has been indispensable reading over the last
arXiv:2505.10887v3 Announce Type: replace Abstract: This paper introduces extsc{InfantAgent-Next}, a generalist agent capable of interacting with computers in a multimodal manner, encompassing text, i
arXiv:2605.00572v1 Announce Type: new Abstract: Algorithm performance in combinatorial optimization is highly sensitive to parameter settings, while a single globally tuned configuration often fails t
arXiv:2512.21788v3 Announce Type: replace Abstract: Parameter-Efficient Fine-Tuning of Diffusion Transformers (DiTs) for diverse, multi-conditional tasks often suffers from task interference when usin
arXiv:2605.01409v1 Announce Type: cross Abstract: The growing availability of health-related instructional videos creates new opportunities for clinical training, patient rehabilitation, and health ed
arXiv:2605.01036v1 Announce Type: new Abstract: This paper tackles the problem of physics-aware human motion synthesis in a dynamic scene. Unlike existing works which mainly tend to generate physicall
arXiv:2605.00940v1 Announce Type: new Abstract: A new interpretable experiential learning model based on state history and global feedback is presented. It is capable of learning a behavioral model re
Managing agents and their actions can quickly grow in complexity and introduce security risks unique to AI. To address these challenges, at Google Cloud Next we announced Agent Gateway to provide simp
arXiv:2602.07238v2 Announce Type: replace-cross Abstract: Do leading LLM developers possess a proprietary ``secret sauce'', or is LLM performance driven by scaling up compute? Using training and bench
arXiv:2512.04032v3 Announce Type: replace Abstract: We present jina-vlm, a token-efficient 2.4B parameter vision-language model that achieves state-of-the-art multilingual VQA performance among open 2
arXiv:2605.01742v1 Announce Type: new Abstract: Vision Transformers (ViTs) have achieved strong performance in visual recognition, yet their deployment in resource-constrained industrial environments
arXiv:2605.02288v1 Announce Type: new Abstract: Automated laboratories hold the promise of accelerating scientific discovery, yet their deployment is bottlenecked by the difficulty of designing safe a
arXiv:2604.17465v2 Announce Type: replace Abstract: We provide evidence that language models can detect, localize and, to a certain degree, verbalize the difference between perturbations applied to th
This newsletter episode covers recent AI industry developments including a legal dispute between OpenAI and Elon Musk, Microsoft's involvement in AI developments, the release of DeepSeek's v4 model, a
arXiv:2505.20340v3 Announce Type: replace Abstract: Understanding how latent representations evolve during generation is a central open problem in large language model interpretability. We introduce e
arXiv:2605.00899v1 Announce Type: new Abstract: We present LatentDiff, a scalable framework for semantic dataset comparison that operates directly in the latent space of pretrained vision encoders. By
arXiv:2605.01046v1 Announce Type: new Abstract: LoRA adapts large language models (LLMs) by restricting updates to low-rank subspaces of pre-trained weights. While this substantially reduces training
arXiv:2605.00913v1 Announce Type: new Abstract: Sensor-based Human Activity Recognition (HAR) has attracted increasing attention in medical and healthcare monitoring, particularly with the growth of I
arXiv:2605.01711v1 Announce Type: new Abstract: Existing research largely attributes the global sequence modeling capability of Transformers to the explicit computation of attention weights, a process
arXiv:2605.00884v1 Announce Type: new Abstract: Vision-language-action (VLA) models have shown strong semantic grounding and task generalization in manipulation, but aerial deployment remains difficul
arXiv:2603.00042v2 Announce Type: replace Abstract: We identify the Spectral Energy Gain in extreme model compression, where low-rank binary approximations outperform tiny-rank floating-point baseline
arXiv:2605.01461v1 Announce Type: new Abstract: Swarm foraging algorithms, such as the central-place foraging algorithm (CPFA), typically rely on offline parameter optimization using genetic algorithm
arXiv:2605.02133v1 Announce Type: new Abstract: AC optimal power flow (ACOPF) is foundational yet computationally expensive in power grid operations, driving learning-based surrogates for large-scale
arXiv:2605.02641v1 Announce Type: new Abstract: We present Mamoda2.5, a unified AR-Diffusion framework that seamlessly integrates multimodal understanding and generation within a single architecture.
arXiv:2605.02695v1 Announce Type: new Abstract: SemEval-2026 Task 9 is focused on multilingual polarization detection. Specifically, it covers the identification of multilingual, multicultural and mul
arXiv:2605.00007v1 Announce Type: cross Abstract: Independent sample generation is the prevailing paradigm in modern diffusion-based generative models of AI. We ask a different question: can samples c
arXiv:2605.01417v1 Announce Type: new Abstract: Evaluating large language models (LLMs) for medical applications remains challenging due to benchmark saturation, limited data accessibility, and insuff
arXiv:2605.00969v1 Announce Type: cross Abstract: We present MedMosaic, a medical audio question-answering dataset designed to benchmark language and audio reasoning models under realistic clinical co
arXiv:2508.11205v2 Announce Type: replace Abstract: Structure-preserving approaches to dynamics discovery have demonstrated great potential for modeling physical systems due to their use of strong ind
arXiv:2605.02206v1 Announce Type: new Abstract: Machine unlearning in Vision-Language Models (VLMs) is required for compliance with the General Data Protection Regulation (GDPR), yet current evaluatio
arXiv:2605.02444v1 Announce Type: new Abstract: Encoder-decoder imbalance and the reliance on large input volumes make many 3D brain tumor segmentation models both compute-heavy and brittle. We presen
arXiv:2605.00123v1 Announce Type: new Abstract: Safety trained large language models (LLMs) can often be induced to answer harmful requests through jailbreak prompts. Because we lack a robust understa
arXiv:2605.01579v1 Announce Type: cross Abstract: Empirical causal claims depend on many analyst decisions, from selecting covariates to choosing estimators. Existing robustness tools summarize how re
arXiv:2602.04509v4 Announce Type: replace Abstract: Fine-tuning Multimodal Large Language Models (MLLMs) on task-specific data is an effective way to improve performance on downstream applications. Ho
arXiv:2605.01580v1 Announce Type: new Abstract: Modern deep learning usually treats models as separate artifacts: trained independently, specialized for particular purposes, and replaced when improved
arXiv:2506.05952v4 Announce Type: replace Abstract: Recent advances in transformer-based text-to-motion generation have led to impressive progress in synthesizing high-quality human motion. Neverthele
arXiv:2605.01822v1 Announce Type: new Abstract: Large Language Models (LLMs) are increasingly being used to support scientific discovery. In chemistry, tasks such as reaction prediction and structure
arXiv:2605.02881v1 Announce Type: new Abstract: Vision-Language-Action (VLA) models aim to provide a single generalist controller for robots, but today's systems fall short on the criteria that matter