18 citations · 54 across the 33 of their papers we have counts for
37 papers
It's All Training: A Fully Synthetic Single-Stage Recipe for LLMs
Pierre-Carl Langlais, Pieter Delobelle, Yannick Detrois +7
Current pre-training datasets are derived from web crawls, with all their issues, and were not designed to support mid- and post-training pipelines--for instance, they contain litt…
The Detectability Gap: Hidden Heterogeneity in Hallucination Detection Across Language Models
Pranav Darshan, Pranav A, Sravan Karthick T +2
Sampling based consistency is widely used for hallucination detection, yet aggregate performance can conceal systematic differences in which errors are detectable. This work studie…
The Temporal Tug-of-War: Visualizing and Detecting RAG Conflicts in Diffusion Models via Trajectory Variance
Sravan Karthick T, Pranav Darshan, Pranav A +2
Retrieval-Augmented Generation (RAG) introduces a specific failure mode in discrete diffusion language models: when retrieved context contradicts parametric knowledge, the iterativ…
We're Cooked! - Probing LLM Political Alignment Via Conflict-Framed Recipe Translation
Svetlana Gorovaia, Angelica Henestrosa, Ivan P. Yamshchikov
Large language models (LLMs) are increasingly deployed for translation tasks, yet their implicit political positioning in such contexts remains understudied. We ask whether a singl…
LeakageBench: Document-Level Leakage Risk for Redacting Personally Identifiable Information in Document Images
Vishnu Prasad Vijaya Kumar, Santhosh Venkatesh, Ivan P. Yamshchikov
Real-world personally identifiable information (PII) redaction often operates on document images---scans, screenshots, and PDF renderings---where OCR errors, layout structure, and…
Model in Distress: Sentiment Analysis on French Synthetic Social Media
Pierre-Carl Langlais, Pavel Chizhov, Yannick Detrois +4
Automated analysis of customer feedback on social media is hindered by three challenges: the high cost of annotated training data, the scarcity of evaluation sets, especially in mu…
From Where Words Come: Efficient Regularization of Code Tokenizers Through Source Attribution
Pavel Chizhov, Egor Bogomolov, Ivan P. Yamshchikov
Efficiency and safety of Large Language Models (LLMs), among other factors, rely on the quality of tokenization. A good tokenizer not only improves inference speed and language und…
The Company You Keep: How LLMs Respond to Dark Triad Traits
Angelica Henestrosa, Zeyi Lu, Pavel Chizhov +1
LLMs often exhibit highly agreeable conversational styles, also known as AI sycophancy. This pattern may become problematic when interacting with user prompts that reflect negative…
Teaching Old Tokenizers New Words: Efficient Tokenizer Adaptation for Pre-trained Models
Taido Purason, Pavel Chizhov, Ivan P. Yamshchikov +1
Tokenizer adaptation plays an important role in adapting pre-trained language models to new domains or languages. In this work, we address two complementary aspects of this process…
app.build: A Production Framework for Scaling Agentic Prompt-to-App Generation with Environment Scaffolding
Evgenii Kniazev, Arseny Kravchenko, Igor Rekun +5
We present app.build (https://github.com/neondatabase/appdotbuild-agent), an open-source framework that improves LLM-based application generation through systematic validation and…
ComicScene154: A Scene Dataset for Comic Analysis
Sandro Paval, Ivan P. Yamshchikov, Pascal Meißner
Comics offer a compelling yet under-explored domain for computational narrative analysis, combining text and imagery in ways distinct from purely textual or audiovisual media. We i…
Better Call Claude: Can LLMs Detect Changes of Writing Style?
Johannes Römisch, Svetlana Gorovaia, Mariia Halchynska +2
This article explores the zero-shot performance of state-of-the-art large language models (LLMs) on one of the most challenging tasks in authorship analysis: sentence-level style c…
Team "better_call_claude": Style Change Detection using a Sequential Sentence Pair Classifier
Gleb Schmidt, Johannes Römisch, Mariia Halchynska +2
Style change detection - identifying the points in a document where writing style shifts - remains one of the most important and challenging problems in computational authorship an…
Greening AI-enabled Systems with Software Engineering: A Research Agenda for Environmentally Sustainable AI Practices
Luís Cruz, João Paulo Fernandes, Maja H. Kirkeby +22
The environmental impact of Artificial Intelligence (AI)-enabled systems is increasing rapidly, and software engineering plays a critical role in developing sustainable solutions.…
Surface Fairness, Deep Bias: A Comparative Study of Bias in Language Models
Aleksandra Sorokovikova, Pavel Chizhov, Iuliia Eremenko +1
Modern language models are trained on large amounts of data. These data inevitably include controversial and stereotypical content, which contains all sorts of biases related to ge…
Common Corpus: The Largest Collection of Ethical Data for LLM Pre-Training
Pierre-Carl Langlais, Pavel Chizhov, Catherine Arnett +8
Large Language Models (LLMs) are pre-trained on large amounts of data from different sources and domains. Such datasets often contain trillions of tokens, including large portions…
Transfer of Structural Knowledge from Synthetic Languages
Mikhail Budnikov, Ivan Yamshchikov
This work explores transfer learning from several synthetic languages to English. We investigate the structure of the embeddings in the fine-tuned models, the information they cont…
Even Small Reasoners Should Quote Their Sources: Introducing the Pleias-RAG Model Family
Pierre-Carl Langlais, Pavel Chizhov, Mattia Nee +7
We introduce a new generation of small reasoning models for RAG, search, and source summarization. Pleias-RAG-350m and Pleias-RAG-1B are mid-trained on a large synthetic dataset em…
Who Benchmarks the Benchmarks? Towards Comprehensive Evaluation of Commonsense Reasoning Benchmarks
Pavel Chizhov, Mattia Nee, Anton Changalidis +4
Commonsense reasoning is a key language model capability, as it is purportedly a prerequisite for many basic tasks, unlike specific factual knowledge. It is often measured with mul…
CleanComedy: Creating Friendly Humor through Generative Techniques
Dmitry Vikhorev, Daria Galimzianova, Svetlana Gorovaia +2
Humor generation is a challenging task in natural language processing due to limited resources and the quality of existing datasets. Available humor language resources often suffer…
Toxicity of the Commons: Curating Open-Source Pre-Training Data
Catherine Arnett, Eliot Jones, Ivan P. Yamshchikov +1
Open-source large language models are becoming increasingly available and popular among researchers and practitioners. While significant progress has been made on open-weight model…
Sui Generis: Large Language Models for Authorship Attribution and Verification in Latin
Gleb Schmidt, Svetlana Gorovaia, Ivan P. Yamshchikov
This paper evaluates the performance of Large Language Models (LLMs) in authorship attribution and authorship verification tasks for Latin texts of the Patristic Era. The study sho…
BPE Gets Picky: Efficient Vocabulary Refinement During Tokenizer Training
Pavel Chizhov, Catherine Arnett, Elizaveta Korotkova +1
Language models can largely benefit from efficient tokenization. However, they still mostly utilize the classical BPE algorithm, a simple and reliable method. This has been shown t…
Echo-chambers and Idea Labs: Communication Styles on Twitter
Aleksandra Sorokovikova, Michael Becker, Ivan P. Yamshchikov
This paper investigates the communication styles and structures of Twitter (X) communities within the vaccination context. While mainstream research primarily focuses on the echo-c…