Tootfinder

So. Much. This.
https://hci.social/@chrisamaphone/115391556289269102

A small number of samples can poison LLMs of any size:
https://www.anthropic.com/research/small-samples-poison
"In a joint study with the UK AI Security Institute and the Alan Turing Institute, we found that as few as 250 malicious documents can produce a "

Evaluating and Mitigating LLM-as-a-judge Bias in Communication Systems
Jiaxin Gao, Chen Chen, Yanwen Jia, Xueluan Gong, Kwok-Yan Lam, Qian Wang
https://arxiv.org/abs/2510.12462 …

Evaluating and Mitigating LLM-as-a-judge Bias in Communication Systems
Large Language Models (LLMs) are increasingly being used to autonomously evaluate the quality of content in communication systems, e.g., to assess responses in telecom customer support chatbots. However, the impartiality of these AI "judges" is not guaranteed, and any biases in their evaluation criteria could skew outcomes and undermine user trust. In this paper, we systematically investigate judgment biases in two LLM-as-a-judge models (i.e., GPT-Judge and JudgeLM) under the point-wise scoring…

And more to read in your freetime, if you are interested in AI, from Bluesky, Phillip Isola.
"Over the past year, my lab has been working on fleshing out theory applications of the Platonic Representation Hypothesis.
Today I want to share two new works on this topic:"
Eliciting higher alignment:
https://arxiv.org/…

Words That Make Language Models Perceive
Large language models (LLMs) trained purely on text ostensibly lack any direct perceptual experience, yet their internal representations are implicitly shaped by multimodal regularities encoded in language. We test the hypothesis that explicit sensory prompting can surface this latent structure, bringing a text-only LLM into closer representational alignment with specialist vision and audio encoders. When a sensory prompt tells the model to 'see' or 'hear', it cues the model to resolve its next…

Racist language
allegedly used by leaders of
Young Republican groups in leaked chats
drew widespread condemnation from both sides of the political aisle Tuesday
and prompted the national youth organization to demand the resignations of those involved.
https://www.axios.co…

Young Republicans demand leaders involved in "vile" racist chat resign
"Such behavior is disgraceful, unbecoming of any Republican," the Young Republicans said.

An Explorative Study on Distributed Computing Techniques in Training and Inference of Large Language Models
Sheikh Azizul Hakim, Saem Hasan
https://arxiv.org/abs/2510.11211 http…

An Explorative Study on Distributed Computing Techniques in Training and Inference of Large Language Models
Large language models (LLM) are advanced AI systems trained on extensive textual data, leveraging deep learning techniques to understand and generate human-like language. Today's LLMs with billions of parameters are so huge that hardly any single computing node can train, fine-tune, or infer from them. Therefore, several distributed computing techniques are being introduced in the literature to properly utilize LLMs. We have explored the application of distributed computing techniques in LLMs f…

KnowRL: Teaching Language Models to Know What They Know
Sahil Kale, Devendra Singh Dhami
https://arxiv.org/abs/2510.11407 https://arxiv.org/pdf/2510.11407

KnowRL: Teaching Language Models to Know What They Know
Truly reliable AI requires more than simply scaling up knowledge; it demands the ability to know what it knows and when it does not. Yet recent research shows that even the best LLMs misjudge their own competence in more than one in five cases, making any response born of such internal uncertainty impossible to fully trust. Inspired by self-improvement reinforcement learning techniques that require minimal data, we present a simple but powerful framework KnowRL that strengthens a model's intern…

Running on the battle-tested #Erlang virtual machine that powers planet-scale systems such as WhatsApp and Ericsson, #Gleam is ready for workloads of any size.
https://gleam…

PhysToolBench: Benchmarking Physical Tool Understanding for MLLMs
Zixin Zhang, Kanghao Chen, Xingwang Lin, Lutao Jiang, Xu Zheng, Yuanhuiyi Lyu, Litao Guo, Yinchuan Li, Ying-Cong Chen
https://arxiv.org/abs/2510.09507

PhysToolBench: Benchmarking Physical Tool Understanding for MLLMs
The ability to use, understand, and create tools is a hallmark of human intelligence, enabling sophisticated interaction with the physical world. For any general-purpose intelligent agent to achieve true versatility, it must also master these fundamental skills. While modern Multimodal Large Language Models (MLLMs) leverage their extensive common knowledge for high-level planning in embodied AI and in downstream Vision-Language-Action (VLA) models, the extent of their true understanding of phys…

A study finds that as few as 250 malicious documents can produce a "backdoor" vulnerability in an LLM, regardless of model size or training data volume (Anthropic)
https://www.anthropic.com/research/small-samples-poison

A small number of samples can poison LLMs of any size
Anthropic research on data-poisoning attacks in large language models

UniFField: A Generalizable Unified Neural Feature Field for Visual, Semantic, and Spatial Uncertainties in Any Scene
Christian Maurer, Snehal Jauhri, Sophie Lueth, Georgia Chalvatzaki
https://arxiv.org/abs/2510.06754

UniFField: A Generalizable Unified Neural Feature Field for Visual, Semantic, and Spatial Uncertainties in Any Scene
Comprehensive visual, geometric, and semantic understanding of a 3D scene is crucial for successful execution of robotic tasks, especially in unstructured and complex environments. Additionally, to make robust decisions, it is necessary for the robot to evaluate the reliability of perceived information. While recent advances in 3D neural feature fields have enabled robots to leverage features from pretrained foundation models for tasks such as language-guided manipulation and navigation, existi…

Test-Time Scaling in Diffusion LLMs via Hidden Semi-Autoregressive Experts
Jihoon Lee, Hoyeon Moon, Kevin Zhai, Arun Kumar Chithanar, Anit Kumar Sahu, Soummya Kar, Chul Lee, Souradip Chakraborty, Amrit Singh Bedi
https://arxiv.org/abs/2510.05040

Test-Time Scaling in Diffusion LLMs via Hidden Semi-Autoregressive Experts
Diffusion-based large language models (dLLMs) are trained flexibly to model extreme dependence in the data distribution; however, how to best utilize this information at inference time remains an open problem. In this work, we uncover an interesting property of these models: dLLMs trained on textual data implicitly learn a mixture of semi-autoregressive experts, where different generation orders reveal different specialized behaviors. We show that committing to any single, fixed inference time …

I gotta step up my Irish learning so we can do some irish anarchism https://todon.eu/@CrimethInc/115501256616016837

CrimethInc. Ex-Workers (@CrimethInc@todon.eu)
Attached: 1 image We've published our first translation in Gallego (Galician), bringing the total number of languages represented on our website to 44. https://crimethinc.com/2025/06/18/epilogo-arredor-do-legalismo If you can help us with translation into any language, please get in touch! You can find a comprehensive list of all our work arranged by language here: https://crimethinc.com/languages

Algorithmic Temperature Induced by Adopted Regular Universal Turing Machine
Kentaro Imafuku
https://arxiv.org/abs/2510.11737 https://arxiv.org/pdf/2510.117…

Algorithmic Temperature Induced by Adopted Regular Universal Turing Machine
We prove that an effective temperature naturally emerges from the algorithmic structure of a regular universal Turing machine (UTM), without introducing any external physical parameter. In particular, the redundancy growth of the machine's wrapper language induces a Boltzmann--like exponential weighting over program lengths, yielding a canonical ensemble interpretation of algorithmic probability. This establishes a formal bridge between algorithmic information theory and statistical mechanics, …

Active Model Selection for Large Language Models
Yavuz Durmazkeser, Patrik Okanovic, Andreas Kirsch, Torsten Hoefler, Nezihe Merve G\"urel
https://arxiv.org/abs/2510.09418 …

Active Model Selection for Large Language Models
We introduce LLM SELECTOR, the first framework for active model selection of Large Language Models (LLMs). Unlike prior evaluation and benchmarking approaches that rely on fully annotated datasets, LLM SELECTOR efficiently identifies the best LLM with limited annotations. In particular, for any given task, LLM SELECTOR adaptively selects a small set of queries to annotate that are most informative about the best model for the task. To further reduce annotation cost, we leverage a judge-based or…

A Multimodal GUI Architecture for Interfacing with LLM-Based Conversational Assistants
Hans G. W. van Dam
https://arxiv.org/abs/2510.06223 https://arxiv.or…

A Multimodal GUI Architecture for Interfacing with LLM-Based Conversational Assistants
Advances in large language models (LLMs) and real-time speech recognition now make it possible to issue any graphical user interface (GUI) action through natural language and receive the corresponding system response directly through the GUI. Most production applications were never designed with speech in mind. This article provides a concrete architecture that enables GUIs to interface with LLM-based speech-enabled assistants. The architecture makes an application's navigation graph and sema…

@… I suspect the problem is that culture informs language, and in American English we aren't used to people taking the train many places.
I wonder what the Brits would say.
In Danish the word for "drove" would be appropriate. You could say "I drove to SFO" and it could equally mean by car, bus, or train without any further clarifi…

Is star complexity a proxy for information based complexity of graphs?
Russell K. Standish
https://arxiv.org/abs/2510.07722 https://arxiv.org/pdf/2510.0772…

Is star complexity a proxy for information based complexity of graphs?
Information-based complexity (IBC) is a well-defined complexity measure of any object given a description in a language and a classifier that identifies those descriptions with the object. Of course, the exact numerical value will vary according to the descriptive language and classifier, but under certain universality conditions (eg the classifier identifies programs of a universal Turning machine that halt and output the same value), asymptotically, the complexity measure is independent of th…

TTRV: Test-Time Reinforcement Learning for Vision Language Models
Akshit Singh, Shyam Marjit, Wei Lin, Paul Gavrikov, Serena Yeung-Levy, Hilde Kuehne, Rogerio Feris, Sivan Doveh, James Glass, M. Jehanzeb Mirza
https://arxiv.org/abs/2510.06783

TTRV: Test-Time Reinforcement Learning for Vision Language Models
Existing methods for extracting reward signals in Reinforcement Learning typically rely on labeled data and dedicated training splits, a setup that contrasts with how humans learn directly from their environment. In this work, we propose TTRV to enhance vision language understanding by adapting the model on the fly at inference time, without the need for any labeled data. Concretely, we enhance the Group Relative Policy Optimization (GRPO) framework by designing rewards based on the frequency o…

Improved Extended Regular Expression Matching
Philip Bille, Inge Li G{\o}rtz, Rikke Schjeldrup Jessen
https://arxiv.org/abs/2510.09311 https://arxiv.org/pd…

Improved Extended Regular Expression Matching
An extended regular expression $R$ specifies a set of strings formed by characters from an alphabet combined with concatenation, union, intersection, complement, and star operators. Given an extended regular expression $R$ and a string $Q$, the extended regular expression matching problem is to decide if $Q$ matches any of the strings specified by $R$. Extended regular expressions are a basic concept in formal language theory and a basic primitive for searching and processing data. Extended reg…

It’s strange to watch the world ignore that it’s not just the medium that matters, but the message does too.
And by “the message” I mean both the ideas and the precise language or visuals used to communicate them.
Subtle differences in linguistic execution of the same idea in the same format can have radically opposite effects. The same applies to an image captured from a different angle or in a different style or in a different composition.
And yet it feels like most organizations and most people within them are determined to march on ignoring any consideration of subtlety and craft.
#writing #design #art #marketing

"AI in the guise of Machine Learning, Deep Learning, GenerativeAI (GenAI), or Large Language Models (LLMs)... can be very useful in certain application areas such as recognising or generating patterns in large data sets. However, their key drawback is that any correctness arguments will be inherently probabilistic as they are usually based on unknown data distributions and are therefore susceptible to errors (sometimes termed “hallucinations”). "

RE: #fascism, especially American fascism, there's something you can do. If you're in the US, don't buy anything if you can until Dec 2nd. If you do have to buy something, buy second hand, or buy local. If you're outside, don't buy anything from the US or any US company at all.
Spread the word. Keep it on people's mind. Write your own post. Talk to people you know in person. Print out flyers and post them around town.
The system understands the language of money. If you want a response, you have to speak the language the system understands.
Trump is extremely vulnerable. Don't wait for the regime to recovery. Hit it hard right now, with an economic blockade. Who knows, it might just crumble.
#USPol

I've been thinking about that stupid "Unified Executive" rubbish that the maga-klan likes to spout.
They argue that Art II Sect 1 sez "The executive Power shall be vested in a President of the United States of America." meaning that the entire executive, all of it, every bit, is carried within the human frame of the President.
OK.
That language does not admit of any possibility of delegation of any part of that executive authority to another. So if …

Improving Discrete Diffusion Unmasking Policies Beyond Explicit Reference Policies
Chunsan Hong, Seonho An, Min-Soo Kim, Jong Chul Ye
https://arxiv.org/abs/2510.05725 https://…

Improving Discrete Diffusion Unmasking Policies Beyond Explicit Reference Policies
Masked diffusion models (MDMs) have recently emerged as a novel framework for language modeling. MDMs generate sentences by iteratively denoising masked sequences, filling in [MASK] tokens step by step. Although MDMs support any-order sampling, performance is highly sensitive to the choice of which position to unmask next. Prior work typically relies on rule-based schedules (e.g., max-confidence, max-margin), which provide ad hoc improvements. In contrast, we replace these heuristics with a lea…

Observing Without Doing: Pseudo-Apprenticeship Patterns in Student LLM Use
Jade Hak, Nathaniel Lam Johnson, Matin Amoozadeh, Amin Alipour, Souti Chattopadhyay
https://arxiv.org/abs/2510.04986

Observing Without Doing: Pseudo-Apprenticeship Patterns in Student LLM Use
Large Language Models (LLMs) such as ChatGPT have quickly become part of student programmers' toolkits, whether allowed by instructors or not. This paper examines how introductory programming (CS1) students integrate LLMs into their problem-solving processes. We conducted a mixed-methods study with 14 undergraduates completing three programming tasks while thinking aloud and permitted to access any resources they choose. The tasks varied in open-endedness and familiarity to the participants and…

See, Point, Fly: A Learning-Free VLM Framework for Universal Unmanned Aerial Navigation
Chih Yao Hu, Yang-Sen Lin, Yuna Lee, Chih-Hai Su, Jie-Ying Lee, Shr-Ruei Tsai, Chin-Yang Lin, Kuan-Wen Chen, Tsung-Wei Ke, Yu-Lun Liu
https://arxiv.org/abs/2509.22653

See, Point, Fly: A Learning-Free VLM Framework for Universal Unmanned Aerial Navigation
We present See, Point, Fly (SPF), a training-free aerial vision-and-language navigation (AVLN) framework built atop vision-language models (VLMs). SPF is capable of navigating to any goal based on any type of free-form instructions in any kind of environment. In contrast to existing VLM-based approaches that treat action prediction as a text generation task, our key insight is to consider action prediction for AVLN as a 2D spatial grounding task. SPF harnesses VLMs to decompose vague language i…

Satellite: Detecting and Analyzing Smart Contract Vulnerabilities caused by Subcontract Misuse
Zeqin Liao, Yuhong Nan, Zixu Gao, Henglong Liang, Sicheng Hao, Jiajing Wu, Zibin Zheng
https://arxiv.org/abs/2509.23679

Satellite: Detecting and Analyzing Smart Contract Vulnerabilities caused by Subcontract Misuse
Developers of smart contracts pervasively reuse subcontracts to improve development efficiency. Like any program language, such subcontract reuse may unexpectedly include, or introduce vulnerabilities to the end-point smart contract. Unfortunately, automatically detecting such issues poses several unique challenges. Particularly, in most cases, smart contracts are compiled as bytecode, whose class-level information (e.g., inheritance, virtual function table), and even semantics (e.g., control f…

Stochastic Self-Organization in Multi-Agent Systems
Nurbek Tastan, Samuel Horvath, Karthik Nandakumar
https://arxiv.org/abs/2510.00685 https://arxiv.org/pd…

Stochastic Self-Organization in Multi-Agent Systems
Multi-agent systems (MAS) based on Large Language Models (LLMs) have the potential to solve tasks that are beyond the reach of any single LLM. However, this potential can only be realized when the collaboration mechanism between agents is optimized. Specifically, optimizing the communication structure between agents is critical for fruitful collaboration. Most existing approaches rely on fixed topologies, pretrained graph generators, optimization over edges, or employ external LLM judges, there…

Where MLLMs Attend and What They Rely On: Explaining Autoregressive Token Generation
Ruoyu Chen, Xiaoqing Guo, Kangwei Liu, Siyuan Liang, Shiming Liu, Qunli Zhang, Hua Zhang, Xiaochun Cao
https://arxiv.org/abs/2509.22496

Where MLLMs Attend and What They Rely On: Explaining Autoregressive Token Generation
Multimodal large language models (MLLMs) have demonstrated remarkable capabilities in aligning visual inputs with natural language outputs. Yet, the extent to which generated tokens depend on visual modalities remains poorly understood, limiting interpretability and reliability. In this work, we present EAGLE, a lightweight black-box framework for explaining autoregressive token generation in MLLMs. EAGLE attributes any selected tokens to compact perceptual regions while quantifying the relativ…

FURINA: A Fully Customizable Role-Playing Benchmark via Scalable Multi-Agent Collaboration Pipeline
Haotian Wu, Shufan Jiang, Chios Chen, Yiyang Feng, Hehai Lin, Heqing Zou, Yao Shu, Yanran Li, Chengwei Qin
https://arxiv.org/abs/2510.06800

FURINA: A Fully Customizable Role-Playing Benchmark via Scalable Multi-Agent Collaboration Pipeline
As large language models (LLMs) advance in role-playing (RP) tasks, existing benchmarks quickly become obsolete due to their narrow scope, outdated interaction paradigms, and limited adaptability across diverse application scenarios. To address this gap, we introduce FURINA-Builder, a novel multi-agent collaboration pipeline that automatically constructs fully customizable RP benchmarks at any scale. It enables evaluation of arbitrary characters across diverse scenarios and prompt formats, as t…

Go With The Flow: Churn-Tolerant Decentralized Training of Large Language Models
Nikolay Blagoev, Bart Cox, J\'er\'emie Decouchant, Lydia Y. Chen
https://arxiv.org/abs/2509.21221

Go With The Flow: Churn-Tolerant Decentralized Training of Large Language Models
Motivated by the emergence of large language models (LLMs) and the importance of democratizing their training, we propose GWTF, the first crash tolerant practical decentralized training framework for LLMs. Differently from existing distributed and federated training frameworks, GWTF enables the efficient collaborative training of a LLM on heterogeneous clients that volunteer their resources. In addition, GWTF addresses node churn, i.e., clients joining or leaving the system at any time, and net…

Unsupervised Learning and Representation of Mandarin Tonal Categories by a Generative CNN
Kai Schenck, Ga\v{s}per Begu\v{s}
https://arxiv.org/abs/2509.17859 https://

Unsupervised Learning and Representation of Mandarin Tonal Categories by a Generative CNN
This paper outlines the methodology for modeling tonal learning in fully unsupervised models of human language acquisition. Tonal patterns are among the computationally most complex learning objectives in language. We argue that a realistic generative model of human language (ciwGAN) can learn to associate its categorical variables with Mandarin Chinese tonal categories without any labeled data. All three trained models showed statistically significant differences in F0 across categorical varia…

POVQA: Preference-Optimized Video Question Answering with Rationales for Data Efficiency
Ashim Dahal, Ankit Ghimire, Saydul Akbar Murad, Nick Rahimi
https://arxiv.org/abs/2510.01009

POVQA: Preference-Optimized Video Question Answering with Rationales for Data Efficiency
Video Question Answering (VQA) with Large Vision Language Models (LVLMs) has gained significant traction in research ever since the Flamingo was introduced by Deepmind. Recent advancements in large context/long video question answering have allowed VQA tasks to have context window of 1500+ frames. However, this only leads to 50 seconds of video footage without losing any significant information. We introduce POVQA, a data-efficient pipeline that compresses each second of video into a single tem…

Erase to Improve: Erasable Reinforcement Learning for Search-Augmented LLMs
Ziliang Wang, Kang An, Xuhui Zheng, Faqiang Qian, Weikun Zhang, Cijun Ouyang, Jialu Cai, Yuhang Wang, Yichao Wu
https://arxiv.org/abs/2510.00861

Erase to Improve: Erasable Reinforcement Learning for Search-Augmented LLMs
While search-augmented large language models (LLMs) exhibit impressive capabilities, their reliability in complex multi-hop reasoning remains limited. This limitation arises from three fundamental challenges: decomposition errors, where tasks are incorrectly broken down; retrieval missing, where key evidence fails to be retrieved; and reasoning errors, where flawed logic propagates through the reasoning chain. A single failure in any of these stages can derail the final answer. We propose Erasa…

Robust Vision-Language Models via Tensor Decomposition: A Defense Against Adversarial Attacks
Het Patel, Muzammil Allie, Qian Zhang, Jia Chen, Evangelos E. Papalexakis
https://arxiv.org/abs/2509.16163 …

Robust Vision-Language Models via Tensor Decomposition: A Defense Against Adversarial Attacks
Vision language models (VLMs) excel in multimodal understanding but are prone to adversarial attacks. Existing defenses often demand costly retraining or significant architecture changes. We introduce a lightweight defense using tensor decomposition suitable for any pre-trained VLM, requiring no retraining. By decomposing and reconstructing vision encoder representations, it filters adversarial noise while preserving meaning. Experiments with CLIP on COCO and Flickr30K show improved robustness.…

Synthetic Dialogue Generation for Interactive Conversational Elicitation & Recommendation (ICER)
Moonkyung Ryu, Chih-Wei Hsu, Yinlam Chow, Mohammad Ghavamzadeh, Craig Boutilier
https://arxiv.org/abs/2510.02331

Synthetic Dialogue Generation for Interactive Conversational Elicitation & Recommendation (ICER)
While language models (LMs) offer great potential for conversational recommender systems (CRSs), the paucity of public CRS data makes fine-tuning LMs for CRSs challenging. In response, LMs as user simulators qua data generators can be used to train LM-based CRSs, but often lack behavioral consistency, generating utterance sequences inconsistent with those of any real user. To address this, we develop a methodology for generating natural dialogues that are consistent with a user's underlying sta…

WISE: Weak-Supervision-Guided Step-by-Step Explanations for Multimodal LLMs in Image Classification
Yiwen Jiang, Deval Mehta, Siyuan Yan, Yaling Shen, Zimu Wang, Zongyuan Ge
https://arxiv.org/abs/2509.17740

WISE: Weak-Supervision-Guided Step-by-Step Explanations for Multimodal LLMs in Image Classification
Multimodal Large Language Models (MLLMs) have shown promise in visual-textual reasoning, with Multimodal Chain-of-Thought (MCoT) prompting significantly enhancing interpretability. However, existing MCoT methods rely on rationale-rich datasets and largely focus on inter-object reasoning, overlooking the intra-object understanding crucial for image classification. To address this gap, we propose WISE, a Weak-supervision-guided Step-by-step Explanation method that augments any image classificatio…

Beyond the Score: Uncertainty-Calibrated LLMs for Automated Essay Assessment
Ahmed Karim (Judy), Qiao Wang (Judy), Zheng Yuan
https://arxiv.org/abs/2509.15926 https://

Beyond the Score: Uncertainty-Calibrated LLMs for Automated Essay Assessment
Automated Essay Scoring (AES) systems now reach near human agreement on some public benchmarks, yet real-world adoption, especially in high-stakes examinations, remains limited. A principal obstacle is that most models output a single score without any accompanying measure of confidence or explanation. We address this gap with conformal prediction, a distribution-free wrapper that equips any classifier with set-valued outputs and formal coverage guarantees. Two open-source large language models…