Searcharxiv⌕ Search

arXiv subjects

Rafet Sifa

Publications and source records attributed to Rafet Sifa.

At least 19 recordsLinked to original sources
arXiv2026

TopoQuant: Topological Framework to Characterize Behavioral Shifts in Quantized Large Language Models

Post-training quantization (PTQ) reduces the com putational cost of large language models (LLMs), but it can introduce internal structural and behavioral changes that are not captured by traditional metrics such as accuracy and per plexity. We present TopoQuant, a topological framework that uses topological data analysis (TDA) to quantify structural drift in self-attention weights and relate it to behavioral changes in quantized LLMs. Across four LLM families and multiple quantization levels, we show that topological drift remains stable at moderate compression levels but increases sharply below 4 bit precision, closely tracking declines in attention agreement. Our results reveal a strong relationship between topological drift and attention agreement, indicating that structural changes in model parameters closely reflect behavioral degradation under quantization. We further demonstrate that topological features can accurately predict attention consistency using lightweight machine learning models, achieving R2 scores of up to 0.85. Predicting attention agreement enables efficient assessment of be havioral consistency in quantized LLMs, allowing quantization induced degradation to be detected without running expensive downstream tasks. The results indicate that topological drift pro vides an efficient early-warning signal for quantization-induced degradation and can help identify safe compression limits without extensive downstream evaluation.

cs.AI↗
arXiv2026

Evaluating and Benchmarking the System One Model Jev

Jev is a commercial "System One" model from TypeSafe AI that does not generate text: given a state and typed questions, it returns a choice from fixed options, a position on a rubric, or the probability that a statement is true, with probabilities described as calibrated. Such models target small decisions in information access pipelines. We evaluate Jev (jev-1.13.0) zero-shot on 37 datasets, from classification and routing to moderation and rubric scoring, with one frozen template per dataset and full evaluation splits: 346,009 requests for under US$10. For reference, we score Qwen3.8-27B and Gemma-4-E4B on identical requests via their exact option probabilities. Jev reaches 95-99% accuracy on IMDB, SST-2, HellaSwag and ARC, beats Qwen on 27 of 37 datasets, with none of Qwen's leads outside the bootstrap intervals, and Gemma on all 37, and its choice probabilities are well calibrated and support selective prediction. All three models degrade on low-resource languages, fine-grained or noisy labels, and rubric-based judgments, and binary probabilities need tuned thresholds, which raise micro-F1 on UNFAIR-ToS from 0.50 to 0.75. Jev answers MMLU's calculation-heavy questions more accurately than other MMLU questions, unlike both open models; probes rule out shallow memorization but not memorized question-answer pairs. Extending the protocol to sequential decisions, where the model chooses the next action at every step, Jev completes 35% of ALFWorld's unseen tasks, 42% of 90 TextWorld games and 21% of BabyAI-Text missions, chooses the optimal Blackjack action in 90% of decisions, and on FrozenLake almost never moves away from the goal to avoid holes. It often repeats actions that had no effect. Qwen completes more ALFWorld and TextWorld games than Jev (54.5% and 51.1%) but shares its failures on BabyAI-Text and FrozenLake. We release the code, harness and all raw responses.

cs.CL↗
arXiv2026

PlayerEmbed: Learning Behavioral Fingerprint Representations from Raw Input Sequences

Raw player input sequences in video games contain fine-grained behavior, but are high-dimensional and noisy, which makes large-scale analytics challenging. To address this, we introduce PlayerEmbed, a framework that learns compact and discriminative player embeddings using deep metric learning. Our approach uses a hybrid CNN-LSTM encoder trained on raw input sequences from 1,000 players. We evaluate the model on held-out sequences from these players and on sequences from 1,000 players entirely unseen during training to assess generalization to unseen identities. For player identification, a nearest-centroid classifier built from reference embeddings achieves strong performance, reaching balanced accuracies of 0.950 for players seen during training and 0.879 for players unseen during training. Systematically increasing the number of training players substantially improves generalization to unseen players, indicating a transition from simple fingerprinting to learning transferable behavioral patterns.

cs.LG↗
arXiv2026

BehaviorFormer: Discovering Behavioral Patterns via Clusterable Next-Token Prediction

Discovering behavioral patterns in open-ended games is challenging due to the lack of clear objectives or ground-truth annotations. In this work, we extend a next-token prediction Transformer with the deep clustering algorithm DCSS to learn meaningful representations with well-separated clusters. In a controlled Minecraft session with 26 players, the method discovers 23 distinct behavioral patterns that generalize across the studied players, further refining structures present in self-supervised representations into clearly separable clusters. We group the clusters into seven interpretable categories covering stationary, roaming, travel, mining, base work, logging, and farming behavior.

cs.LG↗
arXiv2026

Automated Regulatory Compliance Question Answering in Financial Services with Domain-Adapted Retrieval-Augmented Generation

Financial institutions operate under dense, frequently amended rulebooks, and answering a compliance question correctly requires not only fluency but verifiable grounding in the authoritative text. Large language models are attractive for this task, yet the models that firms can realistically deploy on-premise are compact ones, and compact models hallucinate obligations. We study whether a carefully domain-adapted retrieval-augmented generation pipeline closes that gap. Our retriever is built in three stages on top of LegalBERT: entailment tuning that recasts question--passage matching as premise--hypothesis reconstruction, contrastive tuning with in-batch negatives, and score-level fusion with BM25. Our generator is a compact model (2B--12B parameters) served under 4-bit quantization, either prompted or adapted with retrieval-aware fine-tuning (RAFT) through LoRA. On ObliQA, a question-answering benchmark built from the Abu Dhabi Global Market rulebooks, the staged retriever raises Recall@10 from 0.256 to 0.774 and outperforms BM25 (0.678) and E5-large-v2 (0.758), the strongest general-purpose dense encoder we tested. RAFT-LoRA then improves the composite RePASs answer-quality score for every model we could adapt, with the largest gain on the weakest one. However, the adapted models do not transfer to Australian case-law questions, and a closed-book model that receives no passages at all scores within 0.011 RePASs of the full pipeline while producing answers that cite nothing and misstate obligations. The retrieval gain is therefore measured directly, the generation gain is a gain in RePASs rather than demonstrated grounding, and grounding itself requires an evaluation protocol that RePASs does not provide.

cs.CL↗
arXiv2026

On the Impact of Anonymization on the Performance of Large Language Models

As large language models are increasingly deployed in sensitive domains, anonymizing input data to protect personally identifiable information has become a critical practice. However, the impact of this anonymization on model utility is not well understood. This paper presents a systematic empirical study of the trade-off between privacy and performance. We evaluate five prominent language models across eleven diverse benchmarks, comparing their performance on original versus pseudonymized inputs. Our results reveal that while anonymization generally degrades performance, the effect is highly nuanced. We find that more capable models, such as Qwen2.5-72B and GPT-4o mini, suffer the largest performance drops, suggesting a stronger reliance on specific entity information. The impact is also task-dependent: performance on TruthfulQA improves with anonymization, while retrieval-focused tasks like RGB experience a catastrophic decline. Further experiments show that reversible anonymization techniques that preserve entity uniqueness significantly outperform irreversible ones like redaction, and that explicitly prompting models about anonymization offers no discernible benefit. We conclude that anonymization is not a one-size-fits-all solution and must be co-designed with the model and task in mind to balance privacy and utility effectively. Our findings provide a crucial baseline for developing more robust, privacy-aware AI systems.

cs.CL↗
arXiv2026

Accuracy is Not Enough: A Divergence-Based Approach to Evaluate Fidelity Loss in Quantized LLMs

Deployment of Large Language Models (LLMs) on memory-constrained edge devices relies heavily on aggressive post-training quantization. However, evaluating these models is largely based on zero-shot task accuracy, which depends solely on argmax predictions and is insensitive to changes in the underlying predictive distribution. Consequently, accuracy can exhibit unstable, non-monotonic behavior under progressive quantization, masking substantial fidelity loss relative to the BFloat16 (BF16) uncompressed base model and providing misleading deployment signals. We introduce a distribution-sensitive evaluation framework quantifying information loss in quantized LLMs as the divergence between full-vocabulary predictive distributions at the token decision boundary. We compute statistical distances, including Jensen-Shannon Divergence and Total Variation Distance, between outputs of full-precision and quantized models, enabling a fine-grained analysis of distributional shift. Using this framework, we quantify probability mass displacement and distributional drift relative to the BF16 reference, capturing predictive distribution changes not reflected in top-1 accuracy. We conduct a 120-run experimental matrix across five foundation architectures and four reasoning benchmarks under progressive quantization regimes, from uncompressed BF16 to Q2_K, providing a systematic fidelity analysis. Our results show divergence metrics generally increase under stronger quantization, complementing task accuracy with a fidelity signal. Across tested llama-cpp schemes, mixed-precision Q4_K generally yields lower divergence than uniform Q4_0 at similar memory footprints. These findings motivate distribution-aware evaluation as a practical diagnostic complement to task accuracy; they do not directly establish correctness, calibration, safety, or user-perceived quality.

cs.LG↗
arXiv2026

The Illusion of Equivalency: Statistical Characterization of Quantization Effects in LLMs

Post-Training Quantization has become widely used to compress large language models to make them deployable on resource-constrained devices. However, the evaluation of quantization methods mainly uses accuracy and perplexity, which cannot capture the behavioral changes in the quantized variants. In this work, we propose Correctness Agreement, a decision-level metric that can measure the intersection of correct predictions between the base model and its quantized variant. We use this metric across multiple models and quantization bit levels (8-bit to 2-bit), and we find that the base and quantized variants usually have a shift in behavior even when accuracy and perplexity are preserved. In order to explain this effect, we study the effect of quantization on the structure of the attention weights using statistical and distributional measures. The results reveal a breakpoint at low bit widths and show that query and key projections are more sensitive to quantization than the value and output projections. These results prove the illusion of equivalency between the base and quantized models and inspire behavioral evaluation beyond perplexity and accuracy for quantization methods.

cs.AI↗
arXiv2026

Leveraging Transfer Learning with Class-Specific Decoders for Laparoscopic Segmentation

Effective multi-organ segmentation in surgical data requires learning the intricate anatomical features and alleviating the challenge of class imbalance, which results from relatively lower proportions of small and limitedly exposed structures. Recent works on laparoscopic multi-organ segmentation focus on learning structure-specific features through class-specific decoder architectures and report favorable results. This work extends the decoder-focused architectures to investigate knowledge sharing in the cross-surgical domain. We utilize two datasets representing different surgical domains, rectal and cholecystectomy surgeries, to explore how surgical conceptual knowledge transfers under partially common anatomical representations. Additionally, we compare the feature adaptation for the encoder and decoder at different training stages to analyse the knowledge adaptation and retention in the network. Our results corroborate previous findings on decoder-specific architectures and demonstrate that the organ-specific decoder model (CEMD), fully fine-tuned after cross-domain pre-training, achieves the highest segmentation performance (62.4\% dice) while converging substantially faster than training from scratch. However, we also find that class imbalance in surgical data remains a persistent challenge that transfer learning does not fully resolve for underrepresented anatomical structures.

cs.CV↗
arXiv2026

First Investigation of Deep Learning for Intraoperative Gauze Segmentation in Minimally Invasive Abdominal Surgery

Surgical gauze is an essential part of surgical procedures, primarily used for controlling bleeding and absorbing bodily fluids. The post-surgical retention of gauze can lead to serious complications and necessitate additional surgery for its removal. Despite the clinical significance, research on gauze segmentation using real-world surgical data remains underexplored, owing in part to the scarcity of annotated datasets. In this work, we investigate the use of deep learning methods for gauze segmentation in robot-assisted minimally invasive abdominal surgeries, utilizing an in-house surgical dataset prepared at a university hospital. The training data reflects realistic surgical settings and captures extensive diversity in spatial, morphological, and visual attributes across three different gauze categories. We evaluate several widely used segmentation architectures, including CNN-based, transformer-based, and hybrid architectures, to establish a proof-of-concept for gauze segmentation in a realistic clinical setting. In addition, we investigate the influence of sub-optimally annotated, auto-tracked segmentation masks as a strategy to address data scarcity and improve performance. Our results demonstrate the efficacy of real-world training data in countering the main challenge reported by prior works, the trade-off between blood presence and gauze detection. The incorporation of auto-tracked annotations yields performance enhancements, particularly in generic surgical scenarios. The integration of effective segmentation approaches can benefit robot-guided surgical procedures and various downstream applications by providing precise delineation of foreign objects, thereby enhancing patient safety and surgical outcomes.

cs.CV↗
arXiv2026

Hybrid Deep Learning for Traceability and Classification of Industrial Slate Tiles

Applying deep learning to instance-aware reidentification of slate tiles and extraction site classification can improve production efficiency and quality control in the slate tile industry. These tasks are particularly important for handling natural materials where visual variability can make manual inspection costly and error-prone. We present a lightweight, hybrid deep learning approach that combines image matching and classification within a single framework. The system integrates a feature-matching branch based on XFeat with a MobileNetV3- based classification branch. The XFeat branch, combined with a LightGlue matching head, improves instance matching performance by +15.4% AUC. For classification, features from both backbones are shared and fused, resulting in a +10.9% accuracy improvement over a standard MobileNetV3 model. Our approach is evaluated on a newly created industrial dataset consisting of 2,610 slate tile images from six extraction sites. The results demonstrate the effectiveness of the proposed approach for object re-identification and classification in an industrial setting.

cs.CV↗
arXiv2026

Knowing When Not to Predict: Self Supervised Learning and Abstention for Safer DR Screening

Self-supervised learning (SSL) is now a standard way to pretrain medical image models, but performance is still mostly judged by downstream accuracy. For safety-critical screening tasks such as diabetic retinopathy grading, this is not enough: a model must also know when its predictions are unreliable and defer uncertain cases for clinical review. In this work, we examine how the length of SSL pretraining influences calibrated confidence and confidence-based abstention. We evaluate multiple SSL checkpoints under a fixed fine-tuning protocol and assess calibrated confidence, coverage, selective accuracy, and selective macro-F1. Across datasets and data regimes, SSL pretraining improves selective prediction compared to training from scratch. Unlike prior SSL studies that primarily evaluate downstream accuracy or AUROC, we analyze how SSL pretraining duration influences confidence behavior under calibrated confidence-based abstention. However, once accuracy saturates, selective performance can still change markedly across checkpoints, and longer pretraining does not consistently improve reliability. These results underscore the importance of abstention-aware evaluation and suggest that pretraining length should be treated as an important reliability-related design choice rather than only a computational detail. Code is available at GitHub.

cs.CV↗
arXiv2026

Is continuous CoT better suited for multi-lingual reasoning?

We investigate whether performing reasoning in a continuous latent space leads to more robust multilingual capabilities. We compare Continuous Chain-of-Thought (using the CODI framework) against standard supervised fine-tuning across five typologically diverse languages: English, Chinese, German, French, and Urdu. Our experiments on GSM8k and CommonsenseQA demonstrate that continuous reasoning significantly outperforms explicit reasoning on low-resource languages, particularly in zero-shot settings where the target language was not seen during training. Additionally, this approach achieves extreme efficiency, compressing reasoning traces by approximately $29\times$ to $50\times$. These findings indicate that continuous latent representations naturally exhibit greater language invariance, offering a scalable solution for cross-lingual reasoning.

cs.CL↗
arXiv2026

Towards Reliable Machine Translation: Scaling LLMs for Critical Error Detection and Safety

Machine Translation (MT) plays a pivotal role in cross-lingual information access, public policy communication, and equitable knowledge dissemination. However, critical meaning errors, such as factual distortions, intent reversals, or biased translations, can undermine the reliability, fairness, and safety of multilingual systems. In this work, we explore the capacity of instruction-tuned Large Language Models (LLMs) to detect such critical errors, evaluating models across a range of parameters using the publicly accessible data sets. Our findings show that model scaling and adaptation strategies (zero-shot, few-shot, fine-tuning) yield consistent improvements, outperforming encoder-only baselines like XLM-R and ModernBERT. We argue that improving critical error detection in MT contributes to safer, more trustworthy, and socially accountable information systems by reducing the risk of disinformation, miscommunication, and linguistic harm, especially in high-stakes or underrepresented contexts. This work positions error detection not merely as a technical challenge, but as a necessary safeguard in the pursuit of just and responsible multilingual AI. The code will be made available at GitHub.

cs.CL↗
arXiv2026

Modalities, a PyTorch-native Framework For Large-scale LLM Training and Research

Today's LLM (pre-) training and research workflows typically allocate a significant amount of compute to large-scale ablation studies. Despite the substantial compute costs of these ablations, existing open-source frameworks provide limited tooling for these experiments, often forcing researchers to write their own wrappers and scripts. We propose Modalities, an end-to-end PyTorch-native framework that integrates data-driven LLM research with large-scale model training from two angles. Firstly, by integrating state-of-the-art parallelization strategies, it enables both efficient pretraining and systematic ablations at trillion-token and billion-parameter scale. Secondly, Modalities adopts modular design with declarative, self-contained configuration, enabling reproducibility and extensibility levels that are difficult to achieve out-of-the-box with existing LLM training frameworks.

cs.LG↗
arXiv2026

Generalizing Abstention for Noise-Robust Learning in Medical Image Segmentation

Label noise is a critical problem in medical image segmentation, often arising from the inherent difficulty of manual annotation. Models trained on noisy data are prone to overfitting, which degrades their generalization performance. While a number of methods and strategies have been proposed to mitigate noisy labels in the segmentation domain, this area remains largely under-explored. The abstention mechanism has proven effective in classification tasks by enhancing the capabilities of Cross Entropy, yet its potential in segmentation remains unverified. In this paper, we address this gap by introducing a universal and modular abstention framework capable of enhancing the noise-robustness of a diverse range of loss functions. Our framework improves upon prior work with two key components: an informed regularization term to guide abstention behaviour, and a more flexible power-law-based auto-tuning algorithm for the abstention penalty. We demonstrate the framework's versatility by systematically integrating it with three distinct loss functions to create three novel, noise-robust variants: GAC, SAC, and ADS. Experiments on the CaDIS and DSAD medical datasets show our methods consistently and significantly outperform their non-abstaining baselines, especially under high noise levels. This work establishes that enabling models to selectively ignore corrupted samples is a powerful and generalizable strategy for building more reliable segmentation models. Our code is publicly available at https://github.com/wemous/abstention-for-segmentation.

cs.CV↗
arXiv2026

Domain-Adaptation through Synthetic Data: Fine-Tuning Large Language Models for German Law

Large language models (LLMs) often struggle in specialized domains such as legal reasoning due to limited expert knowledge, resulting in factually incorrect outputs or hallucinations. This paper presents an effective method for adapting advanced LLMs to German legal question answering through a novel synthetic data generation approach. In contrast to costly human-annotated resources or unreliable synthetic alternatives, our approach systematically produces high-quality, diverse, and legally accurate question-answer pairs directly from authoritative German statutes. Using rigorous automated filtering methods and parameter-efficient fine-tuning techniques, we demonstrate that LLMs adapted with our synthetic dataset significantly outperform their baseline counterparts on German legal question answering tasks. Our results highlight the feasibility of using carefully designed synthetic data as a robust alternative to manual annotation in high-stakes, knowledge-intensive domains.

cs.CL↗
arXiv2026

Benchmark Success, Clinical Failure: When Reinforcement Learning Optimizes for Benchmarks, Not Patients

Recent Reinforcement Learning (RL) advances for Large Language Models (LLMs) have improved reasoning tasks, yet their resource-constrained application to medical imaging remains underexplored. We introduce ChexReason, a vision-language model trained via R1-style methodology (SFT followed by GRPO) using only 2,000 SFT samples, 1,000 RL samples, and a single A100 GPU. Evaluations on CheXpert and NIH benchmarks reveal a fundamental tension: GRPO recovers in-distribution performance (23% improvement on CheXpert, macro-F1 = 0.346) but degrades cross-dataset transferability (19% drop on NIH). This mirrors high-resource models like NV-Reason-CXR-3B, suggesting the issue stems from the RL paradigm rather than scale. We identify a generalization paradox where the SFT checkpoint uniquely improves on NIH before optimization, indicating teacher-guided reasoning captures more institution-agnostic features. Furthermore, cross-model comparisons show structured reasoning scaffolds benefit general-purpose VLMs but offer minimal gain for medically pre-trained models. Consequently, curated supervised fine-tuning may outperform aggressive RL for clinical deployment requiring robustness across diverse populations.

cs.AI↗