Language Modeling

Ask about this topic

Overview

Language modeling studies how systems learn from sequences, most often by predicting the next token or reconstructing masked tokens, enabling text generation and language understanding. Common approaches include autoregressive Transformers, masked language models, and alternatives such as recurrent or state-space networks, mixture-of-experts models, and diffusion methods. Researchers improve models through data and tokenizer design, continued pretraining, instruction tuning, retrieval, distillation, and compression, evaluating them with perplexity, linguistic tests, and task benchmarks. A central tension is how to make models more efficient and reliable while ensuring benchmark success reflects robust reasoning, generalization, and competence across languages and domains.

Latest papers

CardsList

525 papers73 papers in the last four weeks, up 22% on the four weeks before. 0.5% of all new papers.

Mar 15, 2026

M2^2RNN: Non-Linear RNNs with Matrix-Valued States for Scalable Language Modeling

Transformers are highly parallel but are limited to computations in the TC0^0 complexity class, excluding tasks such as entity tracking and code execution that provably require greater expressive power. Motivated by this limitation, we revisit non-linear Recurrent Neural Networks (RNNs) for language modeling and introduce Matrix-to-Matrix RNN (M2^2RNN): an architecture with matrix-valued hidden states and expressive non-linear state transitions. We demonstrate that the language modeling performance of non-linear RNNs is limited by their state size, and show how the state size expansion mechanism enables efficient use of tensor cores. Empirically, M2^2RNN achieves perfect state tracking generalization at sequence lengths not seen during training. These benefits also translate to large-scale language modeling. In hybrid settings that interleave recurrent layers with attention, Hybrid M2^2RNN outperforms equivalent Gated DeltaNet hybrids by 0.40.4-0.50.5 perplexity points on a 7B MoE model, while using 3×3\times smaller state sizes for the recurrent layers. Notably, replacing even a single recurrent layer with M2^2RNN in an existing hybrid architecture yields accuracy gains comparable to Hybrid M2^2RNN with minimal impact on training throughput. Further, the Hybrid Gated DeltaNet models with a single M2^2RNN layer also achieve superior long-context generalization, outperforming state-of-the-art hybrid linear attention architectures by up to 88 points on LongBench. Together, these results establish non-linear RNN layers as a compelling building block for efficient and scalable language models.

Mayank Mishra, Shawn Tan, Ion Stoica +2

UC Berkeley · Learning Machine · MIT-IBM Watson Lab +2

Recurrent Neural NetworksLanguage ModelingAutoregressive Language ModelingLong-Context Language Modeling
AskPDF
Mar 15, 2026

Drift and selection in LLM text ecosystems

The public text record -- the material from which both people and AI systems now learn -- is increasingly shaped by its own outputs. Generated text enters the public record, later agents learn from it, and the cycle repeats. Here we develop an exactly solvable mathematical framework for this recursive process, based on variable-order nn-gram agents, and separate two forces acting on the public corpus. The first is drift: unfiltered reuse progressively removes rare forms, and in the infinite-corpus limit we characterise the stable distributions exactly. The second is selection: publication, ranking and verification filter what enters the record, and the outcome depends on what is selected. When publication merely reflects the statistical status quo, the corpus converges to a shallow state in which further lookahead brings no benefit. When publication is normative -- rewarding quality, correctness or novelty -- deeper structure persists, and we establish an optimal upper bound on the resulting divergence from shallow equilibria. The framework therefore identifies when recursive publication compresses public text and when selective filtering sustains richer structure, with implications for the design of AI training corpora.

Søren Riis

Queen Mary University of London, London, United Kingdom

Model CollapseLLM TrainingLanguage ModelingRecursive Language Models
AskPDF
Mar 14, 2026

An Alternative Trajectory for Generative AI

The generative artificial intelligence (AI) ecosystem is undergoing rapid transformations that threaten its sustainability. As models transition from research prototypes to high-traffic products, the energetic burden has shifted from one-time training to recurring, unbounded inference. This is exacerbated by reasoning models that inflate compute costs by orders of magnitude per query. The prevailing pursuit of artificial general intelligence through scaling of monolithic models is colliding with hard physical constraints: grid failures, water consumption, and diminishing returns on data scaling. This trajectory yields models with impressive factual recall but struggles in domains requiring in-depth reasoning, possibly due to insufficient abstractions in training data. Current large language models (LLMs) exhibit genuine reasoning depth only in domains like mathematics and coding, where rigorous, pre-existing abstractions provide structural grounding. In other fields, the current approach fails to generalize well. We propose an alternative trajectory based on domain-specific superintelligence (DSS). We argue for first constructing explicit symbolic abstractions (knowledge graphs, ontologies, and formal logic) to underpin synthetic curricula enabling small language models to master domain-specific reasoning without the model collapse problem typical of LLM-based synthetic data methods. Rather than a single generalist giant model, we envision "societies of DSS models": dynamic ecosystems where orchestration agents route tasks to distinct DSS back-ends. This paradigm shift decouples capability from size, enabling intelligence to migrate from energy-intensive data centers to secure, on-device experts. By aligning algorithmic progress with physical constraints, DSS societies move generative AI from an environmental liability to a sustainable force for economic empowerment.

Margarita Belova, Yuval Kansal, Yihao Liang +2

Princeton University, USA

Language ModelingNeuro-Symbolic AIEfficient Language Model Inference
AskPDF
Mar 13, 2026

BERTology of Molecular Property Prediction

Chemical language models (CLMs) have emerged as promising competitors to popular classical machine learning models for molecular property prediction (MPP) tasks. However, an increasing number of studies have reported inconsistent and contradictory results for the performance of CLMs across various MPP benchmark tasks. In this study, we conduct and analyze hundreds of meticulously controlled experiments to systematically investigate the effects of various factors, such as dataset size, model size, and standardization, on the pre-training and fine-tuning performance of CLMs for MPP. In the absence of well-established scaling laws for encoder-only masked language models, our aim is to provide comprehensive numerical evidence and a deeper understanding of the underlying mechanisms affecting the performance of CLMs for MPP tasks, some of which appear to be entirely overlooked in the literature.

Mohammad Mostafanejad, Paul Saxe, T. Daniel Crawford

Department of Chemistry, Virginia Tech, Blacksburg, Virginia 24061, USA · Molecular Sciences Software Institute, Blacksburg, Virginia 24060, USA

Chemical Language ModelsMolecular Property PredictionMolecular Representation LearningLanguage Model PretrainingLLM Fine-TuningLanguage Modeling+1
AskPDF
Mar 13, 2026

Learning from Child-Directed Speech in Two-Language Scenarios: A French-English Case Study

Research on developmentally plausible language models has largely focused on English, leaving open questions about multilingual settings. We present a systematic study of compact language models by extending BabyBERTa to English-French scenarios under strictly size-matched data conditions, covering monolingual, bilingual, and cross-lingual settings. Our design contrasts two types of training corpora: (i) child-directed speech (about 2.5M tokens), following BabyBERTa and related work, and (ii) multi-domain corpora (about 10M tokens), extending the BabyLM framework to French. To enable fair evaluation, we also introduce new resources, including French versions of QAMR and QASRL, as well as English and French multi-domain corpora. We evaluate the models on both syntactic and semantic tasks and compare them with models trained on Wikipedia-only data. The results reveal context-dependent effects: training on Wikipedia consistently benefits semantic tasks, whereas child-directed speech improves grammatical judgments in monolingual settings. Bilingual pretraining yields notable gains for textual entailment, with particularly strong improvements for French. Importantly, similar patterns emerge across BabyBERTa, RoBERTa, and LTG-BERT, suggesting consistent trends across architectures.

Liel Binyamin, Elior Sulem

Faculty of Computer and Information Science, Institute for Applied AI Research Data Science Research Center Ben-Gurion University of the Negev

Language Model PretrainingLanguage ModelingMultilingual Language ModelsMultilingual Language Model EvaluationCross-Lingual Transfer
AskPDF
Mar 13, 2026

RetroReasoner: A Reasoning LLM for Strategic Retrosynthesis Prediction

Retrosynthesis prediction aims to identify reactants that can synthesize a given product molecule. Although molecular large language models (LLMs) have recently shown promising results, most existing methods either generate reactants directly or provide only generic product-level analysis, without explicitly reasoning about bond-disconnection strategies that justify specific reactant choices. This paper proposes RetroReasoner, a retrosynthetic reasoning model that captures chemists' strategic disconnection-based thinking. RetroReasoner is trained with supervised fine-tuning and reinforcement learning. For supervised fine-tuning, SyntheticRetro generates structured disconnection rationales paired with reactant predictions. For reinforcement learning, a round-trip reward evaluates predicted reactants by passing them through a forward synthesis model and rewarding predictions that reconstruct the original product. RetroReasoner can also be applied to multi-step retrosynthetic planning by incorporating it into a parallelized Monte Carlo tree search framework, reducing search time while increasing the number and diversity of valid synthetic pathways. Experimental results show that RetroReasoner outperforms prior baselines, including not only molecular LLMs but also retrosynthesis-specific expert models, and generates a broader range of feasible reactant proposals, especially for challenging reaction instances. The code is available at https://github.com/KU-AGI/RetroReasoner.

Hanbum Ko, Chanhui Lee, Ye Rin Kim +4

Department of Artificial Intelligence, Korea University · Materials Intelligence Lab, LG AI Research · Department of Statistics, Korea University

AI-Assisted Scientific ResearchLanguage ModelingLLM ReasoningRetrosynthesis
AskPDF
Mar 12, 2026

Interpreting Negation in GPT-2: Layer- and Head-Level Causal Analysis

Negation remains a persistent challenge for modern language models, often causing reversed meanings or factual errors. In this work, we conduct a causal analysis of how GPT-2 Small internally processes such linguistic transformations. We examine its hidden representations at both the layer and head level. Our analysis is based on a self-curated 12,000-pair dataset of matched affirmative and negated sentences, covering multiple linguistic templates and forms of negation. To quantify this behavior, we define a metric, the Negation Effect Score (NES), which measures the model's sensitivity in distinguishing between affirmative statements and their negations. We carried out two key interventions to probe causal structure. In activation patching, internal activations from affirmative sentences were inserted into their negated counterparts to see how meaning shifted. In ablation, specific attention heads were temporarily disabled to observe how logical polarity changed. Together, these steps revealed how negation signals move and evolve through GPT-2's layers. Our findings indicate that this capability is not widespread; instead, it is highly concentrated within a limited number of mid-layer attention heads, primarily within layers 4 to 6. Ablating these specific components directly disrupts the model's negation sensitivity: on our in-domain, ablation increased NES (indicating weaker negation sensitivity), and re-introducing cached affirmative activations (rescue) increased NES further, confirming that these heads carry affirmative signal rather than restoring baseline behavior. On xNot360, ablation slightly decreased NES and rescue restored performance above baseline. This pattern demonstrates that these causal patterns are consistent across various negation forms and remain detectable on the external xNot360 benchmark, though with smaller magnitude.

Abdullah Al Mofael, Lisa M. Kuhn, Ghassan Alkadi +1

∗Department of Computer Science, Southeastern Louisiana University, Hammond, LA, USA · †Department of Mathematics, Southeastern Louisiana University, Hammond, LA, USA

Language ModelingCircuits in Language ModelsTransformer InterpretabilityCausal Interventions in Language Models
AskPDF
Mar 12, 2026

Adapting Methods for Domain-Specific Japanese Small LMs: Scale, Architecture, and Quantization

This paper presents a systematic methodology for building domain-specific Japanese small language models using QLoRA fine-tuning. We address three core questions: optimal training scale, base-model selection, and architecture-aware quantization. Stage 1 (Training scale): Scale-learning experiments (1k--5k samples) identify n=4,000 as optimal, where test-set NLL reaches minimum (1.127) before overfitting at 5k samples. Stage 2 (Compare finetuned SLMs): Comparing four Japanese LLMs shows that Llama-3 models with Japanese continual pre-training (Swallow-8B, ELYZA-JP-8B) outperform multilingual models (Qwen2.5-7B). Stage 3 (Quantization): Llama-3 architectures improve under Q4_K_M quantization, while GQA architectures degrade severely (Qwen2.5: -0.280 points). Production recommendation: Swallow-8B Q4_K_M achieves 2.830/3 score, 8.9 s/question, 4.9 GB size. The methodology generalizes to low-resource technical domains and provides actionable guidance for compact Japanese specialist LMs on consumer hardware.

Takato Yasuno

Language ModelingSmall Language ModelsLLM Fine-TuningDomain Adaptation for LLMsLLM Quantization
AskPDF
Mar 12, 2026

Long-Context Encoder Models for Polish Language Understanding

While decoder-only Large Language Models (LLMs) have recently dominated the NLP landscape, encoder-only architectures remain a cost-effective and parameter-efficient standard for discriminative tasks. However, classic encoders like BERT are limited by a short context window, which is insufficient for processing long documents. In this paper, we address this limitation for the Polish by introducing a high-quality Polish model capable of processing sequences of up to 8192 tokens. The model was developed by employing a two-stage training procedure that involves positional embedding adaptation and full parameter continuous pre-training. Furthermore, we propose compressed model variants trained via knowledge distillation. The models were evaluated on 25 tasks, including the KLEJ benchmark, a newly introduced financial task suite (FinBench), and other classification and regression tasks, specifically those requiring long-document understanding. The results demonstrate that our model achieves the best average performance among Polish and multilingual models, significantly outperforming competitive solutions in long-context tasks while maintaining comparable quality on short texts.

Sławomir Dadas, Rafał Poświata, Marek Kozłowski +4

National Information Processing Institute, Warsaw, Poland · PKO Bank Polski, Warsaw, Poland

Long-Context Language ModelingLong-Context ModelingLanguage Model PretrainingLanguage ModelingLong-Context Language Model Evaluation
AskPDF
Mar 12, 2026

Learning Transferable Sensor Models via Language-Informed Pretraining

Multimodal language models have demonstrated strong semantic understanding and reasoning over physiological and behavioral signals captured from diverse healthcare sensors. However, existing sensor-language models can only process a fixed number of sensor channels at a fixed sampling rate, making them difficult to adapt to new or different sensor configurations. This inflexibility prevents models from generalizing across heterogeneous health sensing ecosystems, where data spans high-frequency wearable streams to sparse, day-scale clinical signals. To bridge this gap, we introduce SLIP (Sensor Language-Informed Pretraining), an open-source framework for learning language-aligned representations that generalize across diverse sensor setups. SLIP integrates contrastive alignment with sensor-conditioned captioning, facilitating both discriminative understanding and generative reasoning. By repurposing a pretrained decoder-only language model via cross-attention and introducing a flexible patch embedder, SLIP transfers to new sensor configurations at inference without retraining, regardless of their native sampling rate or input length. Across 11 datasets, SLIP demonstrates superior performance in retrieval, signal captioning, and question answering. It achieves a 77.14% average linear-probing accuracy, a 3.56% improvement over the strongest baseline (SensorLM). Beyond classification, SLIP supports open-vocabulary sensor captioning and question answering without any architectural modifications. Our work offers broad implications for the development of sensor-language models, utilizing cross-modal supervision to ensure more robust generalizability in health sensing applications.

Yuliang Chen, Arvind Pillai, Yu Yvonne Wu +6

Dartmouth College

Multimodal PretrainingLanguage ModelingMultimodal Contrastive Learning
AskPDF
Mar 12, 2026

Bielik-Minitron-7B: Compressing Large Language Models via Structured Pruning and Knowledge Distillation for the Polish Language

This report details the creation of Bielik-Minitron-7B, a compressed 7.35B parameter version of the Bielik-11B-v3.0 model, specifically optimized for European languages. By leveraging a two-stage compression methodology inspired by the NVIDIA Minitron approach, we combined structured hybrid pruning and knowledge distillation to reduce the model's parameter count by 33.4%, from 11.04B to 7.35B. We utilized the NVIDIA Model Optimizer for structural pruning and the NVIDIA NeMo Framework for logit-based distillation for quality recovery. Following distillation, the model underwent a rigorous alignment pipeline consisting of Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO-P), and Reinforcement Learning (GRPO). Our final model successfully recovered approximately 90% of the baseline model's performance while providing up to 50% inference speedup. This approach demonstrates an efficient pathway to create language models for less-represented languages, preserving the original model quality while reducing inference deployment costs.

Remigiusz Kinas, Paweł Kiszczak, Sergio P. Perez +4

Bielik.AI, Ingenix.ai · Bielik.AI, Vstorm · NVIDIA +3

LLM CompressionLanguage ModelingLLM Inference AccelerationLLM Pruning
AskPDF
Mar 11, 2026

Making Bielik LLM Reason (Better): A Field Report

This paper presents a research program dedicated to evaluating and advancing the reasoning capabilities of Bielik, a Polish large language model. The study describes a number of stages of work: initial benchmarking and creation of evaluation methodology, analyzing of comparative results with other LLMs and outlining of future prospects that take into account the limitations of the analyses conducted so far and aims to keep Bielik in the race give the ever-changing -- and competitive -- AI landscape.

Adam Trybus, Bartosz Bartnicki, Remigiusz Kinas

Institute of Philosophy, Jagiellonian University, 52 Grodzka St., Kraków 31-044, Poland · Bielik.ai (Speakleash Foundation), 33/4 Niemczańska St., Wrocław 50-561, Poland

LLM ReasoningLanguage ModelingLLM Evaluation
AskPDF
Mar 10, 2026

HoloByte: Continuous Hyperspherical Distillation for Tokenizer-Free Modeling

Sequence modeling universally relies on discrete subword tokenization to circumvent the O(N2)\mathcal{O}(N^2) computational intractability of native byte-level attention. However, this heuristic quantization imposes artificial morphological boundaries, enforces vocabulary dependence, and fractures the continuity of the optimization landscape. To resolve this dichotomy, we introduce HoloByte: a strictly tokenizer-free framework utilizing Continuous Hyperspherical Distillation. HoloByte partitions discrete byte sequences into fixed-capacity chunks and projects them into a continuous, strictly bounded hyperspherical manifold via an invertible, dimension-preserving orthogonal rotation operator. This spatial superposition allows a macroscopic transformer to operate exclusively on compressed continuous representations, formally reducing the exact attention time complexity from O(N2D)\mathcal{O}(N^2D) to O(N2W2D+ND2)\mathcal{O}\left( \frac{N^2}{W^2}D + ND^2 \right). A localized causal micro-decoder subsequently unbinds these representations to compute exact byte-level distributions. To govern this continuous trajectory, we propose a dual-objective formulation incorporating a mathematically precise Holographic Latent Mean Squared Error, which strictly bounds the gradient and guarantees asymptotic stability. Theoretically, we derive the minimal embedding dimension D=Ω(Wln⁡∣V∣)D = Ω(W \ln |\mathcal{V}|) required to ensure error-free discrete recovery from the continuous manifold. Empirically, under strictly matched parameter constraints, HoloByte is systematically outperforming a comparable discrete Byte-Pair Encoding (BPE) baseline. These results establish Continuous Hyperspherical Distillation as a mathematically rigorous and computationally tractable foundation for vocabulary-invariant sequence modeling. The code is available at https://github.com/VladimerKhasia/HoloByte

Vladimer Khasia

Independent Researcher

Byte-Level Language ModelLanguage ModelingEfficient AttentionLanguage Model Distillation
AskPDF
Mar 10, 2026

Sabiá-4 Technical Report

This technical report presents Sabiá-4 and Sabiazinho-4, a new generation of Portuguese language models with a focus on Brazilian Portuguese language. The models were developed through a four-stage training pipeline: continued pre-training on Portuguese and Brazilian legal corpora, long-context extension to 128K tokens, supervised fine-tuning on instruction data spanning chat, code, legal tasks, and function calling, and preference alignment. We evaluate the models on six benchmark categories: conversational capabilities in Brazilian Portuguese, knowledge of Brazilian legislation, long-context understanding, instruction following, standardized exams, and agentic capabilities including tool use and web navigation. Results show that Sabiá-4 and Sabiazinho-4 achieve a favorable cost-performance trade-off compared to other models, positioning them in the upper-left region of the pricing-accuracy chart. The models show improvements over previous generations in legal document drafting, multi-turn dialogue quality, and agentic task completion.

Thiago Laitz, Thales Sales Almeida, Hugo Abonizio +6

Maritaca AI · Jusbrasil

Portuguese Language ModelsLanguage ModelingLanguage Model Post-TrainingLanguage Model PretrainingLLM Tool Use
AskPDF
Mar 10, 2026

N-gram-like Language Models Predict Naturalistic Reading Time Best

Recent work has found that contemporary language models such as transformers can become so good at next-word prediction that the probabilities they calculate become worse for predicting naturalistic reading time. In this paper, we propose that this can be explained by reading time being shaped by simple n-gram statistics rather than the more complex statistics learned by state-of-the-art transformer language models. We demonstrate that the neural language models whose predictions are most correlated with n-gram probability are also those that calculate probabilities that are the most correlated with eye-tracking-based metrics of reading time on naturalistic text.

James A. Michaelov, Roger P. Levy

Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology · MIT Libraries CREOS, Massachusetts Institute of Technology

Language ModelingAutoregressive Language Modeling
AskPDF
Mar 10, 2026

Build, Borrow, or Just Fine-Tune? A Political Scientist's Guide to Choosing NLP Models

Political scientists increasingly face a consequential choice when adopting natural language processing tools: build a domain-specific model from scratch, borrow and adapt an existing one, or simply fine-tune a general-purpose model on task data? Each approach occupies a different point on the spectrum of performance, cost, and required expertise, yet the discipline has offered little empirical guidance on how to navigate this trade-off. This paper provides such guidance. Using conflict event classification as a test case, I fine-tune ModernBERT on the Global Terrorism Database (GTD) to create Confli-mBERT and systematically compare it against ConfliBERT, a domain-specific pretrained model that represents the current gold standard. Confli-mBERT achieves 75.46% accuracy compared to ConfliBERT's 79.34%. Critically, the four-percentage-point gap is not uniform: on high-frequency attack types such as Bombing/Explosion (F1 = 0.95 vs. 0.96) and Kidnapping (F1 = 0.92 vs. 0.91), the models are nearly indistinguishable. Performance differences concentrate in rare event categories comprising fewer than 2% of all incidents. I use these findings to develop a practical decision framework for political scientists considering any NLP-assisted research task: when does the research question demand a specialized model, and when does an accessible fine-tuned alternative suffice? The answer, I argue, depends not on which model is "better" in the abstract, but on the specific intersection of class prevalence, error tolerance, and available resources. The model, training code, and data are publicly available on Hugging Face.

Shreyas Meher

School of Social and Behavioural Sciences, Erasmus University Rotterdam

LLM Fine-TuningLanguage ModelingText Classification
AskPDF
Mar 10, 2026

Exclusive Self Attention

We introduce exclusive self attention (XSA), a simple modification of self attention (SA) that improves Transformer's sequence modeling performance. The key idea is to constrain attention to capture only information orthogonal to the token's own value vector (thus excluding information of self position), encouraging better context modeling. Evaluated on the standard language modeling task, XSA consistently outperforms SA across model sizes up to 2.7B parameters and shows increasingly larger gains as sequence length grows.

Shuangfei Zhai

Apple

Self-AttentionTransformerLanguage Modeling
AskPDF
Mar 9, 2026

Quantifying Memorization and Privacy Risks in Genomic Language Models

Genomic language models (GLMs) have emerged as powerful tools for learning representations of DNA sequences, enabling advances in variant prediction, regulatory element identification, and cross-task transfer learning. However, as these models are increasingly trained or fine-tuned on sensitive genomic cohorts, they risk memorizing specific sequences from their training data, raising serious concerns around privacy, data leakage, and regulatory compliance. Despite growing awareness of memorization risks in general-purpose language models, little systematic evaluation exists for these risks in the genomic domain, where data exhibit unique properties such as a fixed nucleotide alphabet, strong biological structure, and individual identifiability. We present a comprehensive, multi-vector privacy evaluation framework designed to quantify memorization risks in GLMs. Our approach integrates three complementary risk assessment methodologies: perplexity-based detection, canary sequence extraction, and membership inference. These are combined into a unified evaluation pipeline that produces a worst-case memorization risk score. To enable controlled evaluation, we plant canary sequences at varying repetition rates into both synthetic and real genomic datasets, allowing precise quantification of how repetition and training dynamics influence memorization. We evaluate our framework across multiple GLM architectures, examining the relationship between sequence repetition, model capacity, and memorization risk. Our results establish that GLMs exhibit measurable memorization and that the degree of memorization varies across architectures and training regimes. These findings reveal that no single attack vector captures the full scope of memorization risk, underscoring the need for multi-vector privacy auditing as a standard practice for genomic AI systems.

Alexander Nemecek, Wenbiao Li, Xiaoqian Jiang +2

Case Western Reserve University Cleveland, Ohio, USA · The University of Texas Health Science Center at Houston Houston, Texas, USA · Rutgers University Newark, New Jersey, USA

Memorization in Language ModelsPrivacy Leakage in Language ModelsLanguage ModelingMembership Inference AttacksData Repetition in Language Model Training
AskPDF
Mar 9, 2026

Eyla: Toward an Identity-Anchored LLM Architecture with Integrated Biological Priors -- Vision, Implementation Attempt, and Lessons from AI-Assisted Development

We present the design rationale, implementation attempt, and failure analysis of Eyla, a proposed identity-anchored LLM architecture that integrates biologically-inspired subsystems -- including HiPPO-initialized state-space models, zero-initialized adapters, episodic memory retrieval, and calibrated uncertainty training -- into a unified agent operating system running on consumer hardware. Unlike existing approaches that optimize models for generic helpfulness, Eyla targets identity consistency: the ability to maintain a coherent self-model under adversarial pressure, admit uncertainty, and resist manipulation. We propose the Identity Consistency Score (ICS), a novel benchmark for evaluating this property across LLMs. We then present an honest account of attempting to implement this architecture using AI coding assistants (Claude Code, Cursor) as a non-programmer, documenting a $1,000+ failure that produced a 1.27B parameter model with 86 brain subsystems contributing less than 2% to output. Our analysis identifies five systematic failure modes of AI-assisted development for novel architectures and offers concrete recommendations. To our knowledge, this is the first paper to combine an architectural vision with a documented first-person failure analysis of AI-assisted LLM development, providing lessons for both the AI systems and AI-assisted software engineering communities.

Arif Aditto

Independent Researcher, Dhaka, Bangladesh

Language ModelingAI-Assisted Software DevelopmentMemory-Augmented Language Model Agents
AskPDF
Mar 9, 2026

Towards a more efficient bias detection in financial language models

Bias in financial language models constitutes a major obstacle to their adoption in real-world applications. Detecting such bias is challenging, as it requires identifying inputs whose predictions change when varying properties unrelated to the decision, such as demographic attributes. Existing approaches typically rely on exhaustive mutation and pairwise prediction analysis over large corpora, which is effective but computationally expensive-particularly for large language models and can become impractical in continuous retraining and releasing processes. Aiming at reducing this cost, we conduct a large-scale study of bias in five financial language models, examining similarities in their bias tendencies across protected attributes and exploring cross-model-guided bias detection to identify bias-revealing inputs earlier. Our study uses approximately 17k real financial news sentences, mutated to construct over 125k original-mutant pairs. Results show that all models exhibit bias under both atomic (0.58%-6.05%) and intersectional (0.75%-5.97%) settings. Moreover, we observe consistent patterns in bias-revealing inputs across models, enabling substantial reuse and cost reduction in bias detection. For example, up to 73% of FinMA's biased behaviours can be uncovered using only 20% of the input pairs when guided by properties derived from DistilRoBERTa outputs.

Firas Hadj Kacem, Ahmed Khanfir, Mike Papadakis

SnT, University of Luxembourg, Luxembourg · RIADI, ENSI, University of Manouba, Tunisia

Financial Language Model EvaluationIntersectional Bias in Language ModelsLanguage Model Bias EvaluationSocial Bias in Language ModelsDemographic Bias in Language ModelsLanguage Modeling
AskPDF
Mar 8, 2026

The Dual-Stream Transformer: Channelized Architecture for Interpretable Language Modeling

Standard transformers entangle all computation in a single residual stream, obscuring which components perform which functions. We introduce the Dual-Stream Transformer, which decomposes the residual stream into two functionally distinct components: a token stream updated by attention and a context stream updated by feed-forward networks. Information flow between attention heads is controlled through a hierarchy of mixing strategies, from fully independent (maximum interpretability) to dense (standard transformer behavior). This design exposes a tunable tradeoff between interpretability and performance. We measure this tradeoff on language modeling tasks at 29M parameters. Fully independent head mixing increases validation loss by 8% relative to dense baselines. The recommended Kronecker mixing strategy, which permits scalar communication between heads while preserving within-head structure, costs only 2.5%. All configurations maintain functional generation under attention amplification (scaling logits by factors up to 16 at inference time), with degradation ranging from 16% to 27%. This robustness suggests the architectures learn discrete algorithms that operate independently of soft probabilistic mixing. The architecture provides a foundation for interpretable language models where internal structure is exposed by design. \footnote{This work was partially supported by DARPA Contract HR001125C0302.}

J. Clayton Kerce, Alexis Fox

Georgia Tech Research Institute

TransformerTransformer InterpretabilityTransformer AttentionLanguage Modeling
AskPDF
Mar 7, 2026

Unlocking Data Value in Finance: A Study on Distillation and Difficulty-Aware Training

Large Language Models (LLMs) have demonstrated strong general capabilities, yet their deployment in finance remains challenging due to dense domain-specific terminology, stringent numerical reasoning requirements, and low tolerance for factual errors. We conduct a controlled empirical study showing that in specialized vertical domains, performance is largely determined by the quality and difficulty/verifiability profile of post-training data. We introduce ODA-Fin-SFT-318k, constructed via multi-stage distillation and verification to produce high-quality Chain-of-Thought supervision, and ODA-Fin-RL-12k, curated for hard-but-verifiable tasks that balance reward precision and task diversity. Using standard SFT and RL pipelines, we show that high-quality CoT distillation establishes a robust foundation during SFT, while difficulty- and verifiability-aware sampling improves RL generalization. Evaluated on nine benchmarks spanning general financial tasks, sentiment analysis, and numerical reasoning, our ODA-Fin-RL-8B consistently surpasses open-source state-of-the-art (SOTA) financial LLMs of comparable size. We release our ODA-Fin-SFT-318k and ODA-Fin-RL-12k datasets, along with trained models to advance data-centric financial AI research.

Chuxue Cao, Honglin Lin, Zhanping Zhong +5

Shanghai Artificial Intelligence Laboratory, OpenDataLab, OpenDataArena · Hong Kong University of Science and Technology

Financial Language Model EvaluationLLM TrainingLanguage ModelingLanguage Model DistillationSynthetic Data Generation for Language ModelsRL for Language Models
AskPDF
Mar 6, 2026

Experiences Build Characters: The Linguistic Origins and Functional Impact of LLM Personality

Human problem-solving is enriched by a diversity of styles and personality traits, yet the development of Large Language Models (LLMs) has largely prioritized uniform performance benchmarks that favour specific behavioural tendencies such as assertiveness. To investigate how diverse experiences shape machine personality and influence problem-solving, this study employs continued pre-training to expose models to domain-specific texts in an unsupervised manner, simulating the accumulation of experience. By adapting the Big Five framework via the Machine Personality Inventory (MPI), we quantify the personality traits of these model variants and analyse their relationship to linguistic style and reasoning behaviour. The findings reveal that model competence is bimodal, peaking at "Expressive Generalists" and "Suppressed Specialists," while identifying a "Suppression Advantage" where reduced social traits enhance complex reasoning performance. This study further establishes a causal link between training data linguistics, such as imperative frequency, and lexical diversity, providing a roadmap for "Personality Engineering".

Xi Wang, Mengdie Zhuang, Jiqun Liu

University of Sheffield, Sheffield, UK · University of Oklahoma, Oklahoma, USA

Personality Modeling in Language ModelsLanguage Modeling
AskPDF
Mar 6, 2026

Addressing the Ecological Fallacy in Larger LMs with Human Context

Language model training and inference ignore a fundamental linguistic fact -- there is a dependence between multiple sequences of text written by the same person. Prior work has shown that addressing this form of ecological fallacy can greatly improve the performance of multiple smaller (~124M) GPT-based models. In this work, we ask if addressing the ecological fallacy by modeling the author's language context with a specific LM task (called HuLM) can provide similar benefits for a larger-scale model, an 8B Llama model. To this end, we explore variants that process an author's language in the context of their other temporally ordered texts. We study the effect of pre-training with this author context using the HuLM objective, as well as using it during fine-tuning with author context (HuFT:Human-aware Fine-Tuning). Empirical comparisons show that addressing the ecological fallacy during fine-tuning alone using QLoRA improves the performance of the larger 8B model over standard fine-tuning. Additionally, QLoRA-based continued HuLM pre-training results in a human-aware model generalizable for improved performance over eight downstream tasks with linear task classifier training alone. These results indicate the utility and importance of modeling language in the context of its original generators, the authors.

Nikita Soni, Dhruv Vijay Kunjadiya, Pratham Piyush Shah +3

Department of Computer Science, Stony Brook University · College of Connected Computing, Vanderbilt University, USA

Language ModelingLLM TrainingLLM Fine-TuningFine-Tuning Language ModelsPersonalized Language Models
AskPDF
Mar 5, 2026

FireBench: Evaluating Instruction Following in Enterprise and API-Driven LLM Applications

Instruction following is critical for LLMs deployed in enterprise and API-driven settings, where strict adherence to output formats, content constraints, and procedural requirements is essential for enabling reliable LLM-assisted workflows. However, existing instruction following benchmarks predominantly evaluate natural language generation constraints that reflect the needs of chat assistants rather than enterprise users. To bridge this gap, we introduce FireBench, an LLM instruction following benchmark grounded in real-world enterprise and API usage patterns. FireBench evaluates six core capability dimensions across diverse applications including information extraction, customer support, and coding agents, comprising over 2,400 samples. We evaluate 11 LLMs and present key findings on their instruction following behavior in enterprise scenarios. We open-source FireBench at fire-bench.com to help users assess model suitability, support model developers in diagnosing performance, and invite community contributions.

Yunfan Zhang, Yijie Bei, Jetashree Ravi +1

Columbia University · Fireworks AI

LLM EvaluationInstruction FollowingGenerative AI EvaluationLanguage Modeling
AskPDF
Mar 4, 2026

Bielik-Q2-Sharp: A Comparative Study of Extreme 2-bit Quantization Methods for a Polish 11B Language Model

We present Bielik-Q2-Sharp, the first systematic academic evaluation of extreme 2-bit quantization applied to a Polish large language model. Using Bielik-11B-v2.3-Instruct (11B parameters, Mistral architecture) as our base model, we compare six state-of-the-art post-training quantization methods -- QuIP#, SpinQuant+GPTQ, ButterflyQuant, QTIP, VPTQ, and AQLM -- all calibrated on a Polish-language corpus (CulturaX-PL) with shared Hessian matrices. Our best variant (QuIP# E8P12) achieves 71.92% across 22 Polish benchmarks versus 72.07% for the IQ2_XXS baseline -- within statistical noise, at a modest size premium (3.26 GB vs. ~2.6 GB). On eq_bench, our method scores 47.14 versus 43.53 (+3.6pp), suggesting superior preservation of higher-order reasoning. QTIP achieves the best per-bit efficiency (79.4% MC acc_norm at ~2.4 bpw, 3.27 GB), matching VPTQ's quality at 35% smaller size. We additionally document a MC-generation dissociation phenomenon where rotation-based methods preserve log-likelihood quality but fail catastrophically at autoregressive generation. The entire project was conducted by a single independent researcher on cloud GPUs (vast.ai) within a $285 budget. All models, Hessians, and evaluation logs are publicly available.

Jakub Prejzner

BitSharp, Independent Researcher, Rzeszów, Poland

QuantizationLLM QuantizationPost-Training QuantizationLow-Bit QuantizationLanguage ModelingLLM Evaluation
AskPDF
Mar 4, 2026

Beyond Masks: Efficient, Flexible Diffusion Language Models via Deletion-Insertion Processes

While Masked Diffusion Language Models (MDLMs) relying on token masking and unmasking have shown promise in language modeling, their computational efficiency and generation flexibility remain constrained by the masking paradigm. In this paper, we propose Deletion-Insertion Diffusion language models (DID) that rigorously formulate token deletion and insertion as discrete diffusion processes, replacing the masking and unmasking processes in current MDLMs. DID improves training and inference efficiency by eliminating two major sources of computational overhead in MDLMs: the computations on non-informative 1) <MASK> tokens inherent to the paradigm, and 2) <PAD> tokens introduced in variable-length settings. Furthermore, DID offers greater flexibility by: 1) natively supporting variable-length sequences without requiring fixed-length padding, and 2) an intrinsic self-correction mechanism during generation due to insertion that dynamically adjusts token positions. To train DID, we design a score-based approach that assigns scores to token insertion operations and derive appropriate training objectives. The objectives involve subsequence counting problems, which we efficiently solve via a parallelized dynamic programming algorithm. Our experiments across fixed and variable-length settings demonstrate the advantage of DID over baselines of MDLMs and existing insertion-based LMs, in terms of modeling performance, sampling quality, and training/inference speed, without any hyperparameter tuning.

Fangyu Ding, Ding Ding, Sijin Chen +8

HKUST · Huawei Foundation Model Dept · CUHK

Diffusion Language ModelsLanguage ModelingMasked Diffusion ModelsDiffusion Model Inference Acceleration
AskPDF
Mar 4, 2026

N-gram Injection into Transformers for Dynamic Language Model Adaptation in Handwritten Text Recognition

Transformer-based encoder-decoder networks have recently achieved impressive results in handwritten text recognition, partly thanks to their auto-regressive decoder which implicitly learns a language model. However, such networks suffer from a large performance drop when evaluated on a target corpus whose language distribution is shifted from the source text seen during training. To retain recognition accuracy despite this language shift, we propose an external n-gram injection (NGI) for dynamic adaptation of the network's language modeling at inference time. Our method allows switching to an n-gram language model estimated on a corpus close to the target distribution, therefore mitigating bias without any extra training on target image-text pairs. We opt for an early injection of the n-gram into the transformer decoder so that the network learns to fully leverage text-only data at the low additional cost of n-gram inference. Experiments on three handwritten datasets demonstrate that the proposed NGI significantly reduces the performance gap between source and target corpora.

Florent Meyer, Laurent Guichard, Yann Soullard +3

ANTAI, Rennes, France · Univ Rennes, CNRS, IRISA - UMR 6074, Rennes, France

Inference-Time Adaptation of Language ModelsLanguage ModelingTransformer
AskPDF
Mar 3, 2026

ByteFlow: Language Modeling through Adaptive Byte Compression without a Tokenizer

Modern language models still rely on fixed, pre-defined subword tokenizations. Once a tokenizer is trained, the LM can only operate at this fixed level of granularity, which often leads to brittle and counterintuitive behaviors even in otherwise strong reasoning models. We introduce ByteFlow Net, a new hierarchical architecture that removes tokenizers entirely and instead enables models to learn their own segmentation of raw byte streams into semantically meaningful units. ByteFlow Net performs compression-driven segmentation based on the coding rate of latent representations, yielding adaptive boundaries \emph{while preserving a static computation graph via Top-KK selection}. Unlike prior self-tokenizing methods that depend on brittle heuristics with human-designed inductive biases, ByteFlow Net adapts its internal representation granularity to the input itself. Experiments demonstrate that this compression-based chunking strategy yields substantial performance gains, with ByteFlow Net outperforming both BPE-based Transformers and previous byte-level architectures. These results suggest that end-to-end, tokenizer-free modeling is not only feasible but also more effective, opening a path toward more adaptive and information-grounded language models.

Chunyuan Deng, Sanket Lokegaonkar, Colin Lockard +3

Rice University · Amazon Science

Language ModelingByte-Level Language ModelSubword Tokenization
AskPDF
Mar 3, 2026

Tucano 2 Cool: Better Open Source LLMs for Portuguese

We present Tucano 2, a fully open suite of large language models (LLMs) with 0.5-3.7 billion parameters, designed to address certain gaps in open-source development for Portuguese LLMs. Following our previous works, we now extend our dataset, GigaVerbo-v2, to a new degree of quality and scale, while also introducing a new synthetic dataset, GigaVerbo-v2 Synth, aimed at filling missing gaps in GigaVerbo-v2, and two post-training datasets, GigaVerbo-v2 SFT and GigaVerbo-v2 Preferences, that allow Portuguese LLMs to be trained in domains like retrieval augmented generation, coding, tool use, chain-of-thought reasoning, and many other domains of interest. Through extensive ablation studies, we design both pretraining and continual pretraining recipes for the Tucano 2 suite (Base, Instruct, and Think), which achieve state-of-the-art performance on several Portuguese-language modeling benchmarks. We also extend and refine the evaluation harness introduced in our earlier work, yielding a comprehensive evaluation suite that provides strong signals across different pretraining, continual pretraining, and post-training regimes. All artifacts associated with Tucano 2 are openly released, including training recipes, logs, and source code, ensuring that our work is reproducible, accessible, and extendable by the broader Portuguese NLP community.

Nicholas Kluge Corrêa, Aniket Sen, Shiza Fatimah +4

Bonn-Aachen International Center for Information Technology (b-it) / CAISA Lab · Lamarr Institute for Machine Learning and Artificial Intelligence · Center for Science and Thought +2

Portuguese Language ModelsLanguage ModelingLanguage Model PretrainingLanguage Model Post-TrainingLLM Evaluation
AskPDF