Small Language Models

Background: Recent studies have demonstrated that large language models (LLMs) can perform binary classification tasks on child welfare narratives, detecting the presence or absence of constructs such as substance-related problems, domestic violence, and firearms involvement. Whether smaller, locally deployable models can move beyond binary detection to classify specific substance types from these narratives remains untested. Objective: To validate a locally hosted LLM classifier for identifying specific substance types aligned with DSM-5 categories in child welfare investigation narratives. Methods: A locally hosted 20-billion-parameter LLM classified child maltreatment investigation narratives from a Midwestern U.S. state. Records previously identified as containing substance-related problems were passed to a second classification stage targeting seven DSM-5 substance categories. Expert human review of 900 stratified cases assessed classification precision, recall, and inter-method reliability (Cohen's kappa). Test-retest stability was evaluated using approximately 15,000 independently classified records. Results: Five substance categories achieved almost perfect inter-method agreement (kappa = 0.94-1.00): alcohol, cannabis, opioid, stimulant, and sedative/hypnotic/anxiolytic. Classification precision ranged from 92% to 100% for these categories. Two low-prevalence categories (hallucinogen, inhalant) performed poorly. Test-retest agreement ranged from 92.1% to 99.1% across the seven categories. Conclusions: A small, locally hosted LLM can reliably classify substance types from child welfare administrative text, extending prior work on binary classification to multi-label substance identification.

Word Sense Disambiguation (WSD) remains a key challenge in Natural Language Processing (NLP), especially when dealing with rare or domain-specific senses that are often misinterpreted. While modern high-parameter Large Language Models (LLMs) such as GPT-4-Turbo have shown state-of-the-art WSD performance, their computational and energy demands limit scalability. This study investigates whether low-parameter LLMs (<4B parameters) can achieve comparable results through fine-tuning strategies that emphasize reasoning-driven sense identification. Using the FEWS dataset augmented with semi-automated, rationale-rich annotations, we fine-tune eight small-scale open-source LLMs (e.g. Gemma and Qwen). Our results reveal that Chain-of-Thought (CoT)-based reasoning combined with neighbour-word analysis achieves performance comparable to GPT-4-Turbo in zero-shot settings. Importantly, Gemma-3-4B and Qwen-3-4B models consistently outperform all medium-parameter baselines and state-of-the-art models on FEWS, with robust generalization to unseen senses. Furthermore, evaluation on the unseen "Fool Me If You Can'' dataset confirms strong cross-domain adaptability without task-specific fine-tuning. This work demonstrates that with carefully crafted reasoning-centric fine-tuning, low-parameter LLMs can deliver accurate WSD while substantially reducing computational and energy demands.

Assessing whether an article supports an assertion is essential for hallucination detection and claim verification. While large language models (LLMs) have the potential to automate this task, achieving strong performance requires frontier models such as GPT-5 that are prohibitively expensive to deploy at scale. To efficiently perform biomedical evidence attribution, we present Med-V1, a family of small language models with only three billion parameters. Trained on high-quality synthetic data newly developed in this study, Med-V1 substantially outperforms (+27.0% to +71.3%) its base models on five biomedical benchmarks unified into a verification format. Despite its smaller size, Med-V1 performs comparably to frontier LLMs such as GPT-5, along with high-quality explanations for its predictions. We use Med-V1 to conduct a first-of-its-kind use case study that quantifies hallucinations in LLM-generated answers under different citation instructions. Results show that the format instruction strongly affects citation validity and hallucination, with GPT-5 generating more claims but exhibiting hallucination rates similar to GPT-4o. Additionally, we present a second use case showing that Med-V1 can automatically identify high-stakes evidence misattributions in clinical practice guidelines, revealing potentially negative public health impacts that are otherwise challenging to identify at scale. Overall, Med-V1 provides an efficient and accurate lightweight alternative to frontier LLMs for practical, real-world biomedical evidence attribution. Med-V1 is available at https://github.com/NLM-DIR/Med-V1.

Large Language Models (LLMs) demonstrate exceptional performance across diverse tasks but come with substantial energy and computational costs, particularly in request-heavy scenarios. In many real-world applications, the full scale and capabilities of LLMs are often unnecessary, as Small Language Models (SLMs) can provide accurate responses for simpler text generation tasks. When enhanced with advanced reasoning strategies, such as Chain-of-Thought (CoT) prompting or Majority Voting, SLMs can approach the performance of larger models while reducing overall computational requirements. However, these strategies can also introduce additional energy costs, creating an energy-accuracy trade-off. Our analysis examines these trade-offs in test-time compute strategies for smaller models compared to larger ones, using the MMLU benchmark. Additionally, we explore the input-output token dynamics of transformer architectures, which result in nonlinear hardware energy operation curves for LLMs. To bridge AI research with its physical impact, we propose energy efficiency metrics, including Energy-per-Token, as complements to traditional accuracy benchmarks. Beyond model selection, we propose controlled reasoning in CoT token generation, using operating curves to regulate reasoning depth dynamically. This vision integrates a energy-aware routing mechanism, ensuring that model selection and inference strategies balance accuracy for sustainable AI deployment.

The dominance of large multilingual foundation models has widened linguistic inequalities in Natural Language Processing (NLP), often leaving low-resource languages underrepresented. This paper introduces LilMoo, a 0.6-billion-parameter Hindi language model trained entirely from scratch to address this gap. Unlike prior Hindi models that rely on continual pretraining from opaque multilingual foundations, LilMoo is developed through a fully transparent and reproducible pipeline optimized for limited compute environments. We construct a high-quality Hindi corpus (GigaLekh) filtered through both heuristic and learned (LLM-as-a-judge) methods, complemented by bilingual augmentation with curated English data. Using this dataset, we explore various training recipes for small-scale language models. Across comprehensive evaluation suites, LilMoo consistently outperforms comparably sized multilingual baselines such as Qwen2.5-0.5B and Qwen3-0.6B, demonstrating that well-designed language-specific pretraining can rival large multilingual models at the sub-billion-parameter range.

CDD, or Contamination Detection via output Distribution, identifies data contamination by measuring the peakedness of a model's sampled outputs. We study the conditions under which this approach succeeds and fails on small language models ranging from 70M to 410M parameters. Using controlled contamination experiments on GSM8K, HumanEval, and MATH, we find that CDD's effectiveness depends critically on whether fine-tuning produces verbatim memorization. In the majority of conditions we test, CDD performs at chance level even when the data is verifiably contaminated and detectable by simpler methods. We show that probability-based methods, specifically perplexity and Min-k% Prob, outperform CDD in all conditions where any method exceeds chance, suggesting that CDD's peakedness-based approach is insufficient for contamination detection in small language models. Our code is available at https://github.com/Sela-Omer/Contamination-Detection-Small-LM

Emerging 6G visions, reflected in ongoing standardization efforts within 3GPP, IETF, ETSI, ITU-T, and the O-RAN Alliance, increasingly characterize networks as AI-native systems in which high-level semantic reasoning layers operate above standardized control and data-plane functions. Although frontier-scale large language models (LLMs) such as Qwen2.5-7B and Olmo-3-7B demonstrate strong reasoning capability, their computational footprint limits deployment in latency-sensitive, edge-native infrastructures. This paper presents a systematic empirical study of the scaling behavior and deployment efficiency of compact language models for network-level semantic reasoning in AI-native 6G systems. Using 6G-Bench, a standardization-aligned benchmark comprising 30 decision-making tasks across five capability domains, we evaluate models ranging from 135M (SmolLM2-135M) to 7B parameters (Qwen2.5-7B), including mid-scale architectures such as Llama-3.2-1B, Granite-1B, and Qwen2.5-3B. Deterministic accuracy (pass@1) increases from 0.224 at 135M to 0.707 at 7B, but scaling gains are highly non-uniform. A pronounced stability transition occurs in the 1 to 1.5B range, where accuracy rises from 0.373 (Llama-3.2-1B) to 0.531 (Qwen2.5-1.5B) and the instability gap Delta_5 contracts from 0.356 to 0.138. Beyond 3B parameters, improvements diminish (+0.064 from 3B to 7B). Through single-query inference profiling and an Edge Score metric that normalizes accuracy by latency and memory footprint, we show that semantic reliability per unit edge resource does not scale monotonically with parameter count. Instead, mid-scale models (approximately 1.5 to 3B) achieve the most favorable balance between deterministic stability and computational efficiency, providing deployment-relevant guidance for AI-native 6G architectures. All scripts and results are publicly available at https://github.com/maferrag/6G-Bench

Small open-source language models are gaining attention for healthcare applications in low-resource settings where cloud infrastructure and GPU hardware may be unavailable. However, the reliability of these models under different phrasings of the same clinical question remains poorly understood. We evaluate five open-source models (Gemma 2 2B, Phi-3 Mini 3.8B, Llama 3.2 3B, Mistral 7B, and Meditron-7B, a domain-pretrained model without instruction tuning) across three clinical question answering datasets (MedQA, MedMCQA, and PubMedQA) using five prompt styles: original, formal, simplified, roleplay, and direct. Model behavior is evaluated using consistency scores, accuracy, and instruction-following failure rates. All experiments were conducted locally on consumer CPU hardware without fine-tuning. Consistency and accuracy were largely independent across models. Gemma 2 achieved the highest consistency (0.845-0.888) but the lowest accuracy (33.0-43.5%), while Llama 3.2 showed moderate consistency (0.774-0.807) alongside the highest accuracy (49.0-65.0%). Roleplay prompts consistently reduced accuracy across all models, with Phi-3 Mini dropping 21.5 percentage points on MedQA. Meditron-7B exhibited near-complete instruction-following failure on PubMedQA (99.0% UNKNOWN rate), indicating that domain pretraining alone is insufficient for structured clinical question answering. These findings show that high consistency does not imply correctness: models can be reliably wrong, a dangerous failure mode in clinical AI. Llama 3.2 demonstrated the strongest balance of accuracy and reliability for low-resource deployment. Safe clinical AI requires joint evaluation of consistency, accuracy, and instruction adherence.

Retrieval-Augmented Generation (RAG) grounds Large Language Models (LLMs) to mitigate factual hallucinations. Recent paradigms shift from static pipelines to Modular and Agentic RAG frameworks, granting models autonomy for multi-hop reasoning or self-correction. However, current reflective RAG heavily relies on massive LLMs as universal evaluators. In high-throughput systems, executing complete forward passes for billion-parameter models merely for binary routing introduces severe computational redundancy. Furthermore, in autonomous agent scenarios, inaccurate retrieval causes models to expend excessive tokens on spurious reasoning and redundant tool calls, inflating Time-to-First-Token (TTFT) and costs. We propose Tiny-Critic RAG, decoupling evaluation by deploying a parameter-efficient Small Language Model (SLM) via Low-Rank Adaptation (LoRA). Acting as a deterministic gatekeeper, Tiny-Critic employs constrained decoding and non-thinking inference modes for ultra-low latency binary routing. Evaluations on noise-injected datasets demonstrate Tiny-Critic RAG achieves routing accuracy comparable to GPT-4o-mini while reducing latency by an order of magnitude, establishing a highly cost-effective paradigm for agent deployment.

Small language models (SLMs) have emerged as efficient alternatives to large language models for task-specific applications. However, they are often employed in high-volume, low-latency settings, where efficiency is crucial. We propose TASC, Task-Adaptive Sequence Compression, a framework for SLM acceleration comprising two use-cases: When performing SLM fine-tuning, we propose TASC-ft, which iteratively enriches the tokenizer vocabulary with high-frequency output n-grams and then fine-tunes the model to utilize the expanded vocabulary. Next, we propose an inference-time method, termed TASC-spec. TASC-spec is a lightweight, training-free speculative decoding method that constructs an n-gram draft model from the task's output corpus, mixing task and context n-gram information.TASC-spec avoids any additional training, while bypassing draft-target vocabulary alignment constraints. We demonstrate the effectiveness of both methods across multiple low output-variability generation tasks. Our methods show consistent improvements in inference efficiency while maintaining task performance.

The evolution of both 5G and 6G networks is driving the advancement of fully autonomous network management, placing Intent-Based Networking at the centre of this transformation. This paper introduces a novel framework for 5G and 6G IBN orchestration that leverages a stateful, hierarchical multi-agent architecture to achieve full automation using both SLMs and LLMs. Both models have been evaluated for translation accuracy using metrics such as BLEU, METEOR, and ROUGE-L, as well as computational complexity. Experimental results show that both models exhibit similar accuracy. However, result shows that SLMs can improve the overall completion speed of the IBN lifecycle by 20%.

Leader-follower interaction is an important paradigm in human-robot interaction (HRI). Yet, assigning roles in real time remains challenging for resource-constrained mobile and assistive robots. While large language models (LLMs) have shown promise for natural communication, their size and latency limit on-device deployment. Small language models (SLMs) offer a potential alternative, but their effectiveness for role classification in HRI has not been systematically evaluated. In this paper, we present a benchmark of SLMs for leader-follower communication, introducing a novel dataset derived from a published database and augmented with synthetic samples to capture interaction-specific dynamics. We investigate two adaptation strategies: prompt engineering and fine-tuning, studied under zero-shot and one-shot interaction modes, compared with an untrained baseline. Experiments with Qwen2.5-0.5B reveal that zero-shot fine-tuning achieves robust classification performance (86.66% accuracy) while maintaining low latency (22.2 ms per sample), significantly outperforming baseline and prompt-engineered approaches. However, results also indicate a performance degradation in one-shot modes, where increased context length challenges the model's architectural capacity. These findings demonstrate that fine-tuned SLMs provide an effective solution for direct role assignment, while highlighting critical trade-offs between dialogue complexity and classification reliability on the edge.

Small language models (SLMs) offer compelling advantages in cost, latency, and adaptability, but have so far lagged behind larger models on long-horizon software engineering tasks such as SWE-bench, where they suffer from pervasive action looping and low resolution rates. We introduce SWE-Protégé, a post-training framework that reframes software repair as an expert-protégé collaboration problem. In SWE-Protégé, an SLM remains the sole decision-maker while learning to selectively seek guidance from a strong expert model, recognize stalled states, and follow through on expert feedback. Our approach combines supervised fine-tuning on expert-augmented trajectories with agentic reinforcement learning that explicitly discourages degenerative looping and unproductive expert collaboration. We lightly post-train Qwen2.5-Coder-7B-Instruct to achieve 42.4% Pass@1 on SWE-bench Verified, a +25.4% improvement over the prior SLM state of the art, while using expert assistance sparsely (~4 calls per task and 11% of total tokens).

Sarcasm detection in multilingual and code-mixed environments remains a challenging task for natural language processing models due to structural variations, informal expressions, and low-resource linguistic availability. This study compares four large language models, Llama 3.1, Mistral, Gemma 3, and Phi-4, with a fine-tuned DistilBERT model for sarcasm detection in code-mixed Hinglish text. The results indicate that the smaller, sequentially fine-tuned DistilBERT model achieved the highest overall accuracy of 84%, outperforming all of the LLMs in zero and few-shot set ups, using minimal LLM generated code-mixed data used for fine-tuning. These findings indicate that domain-adaptive fine-tuning of smaller transformer based models may significantly improve sarcasm detection over general LLM inference, in low-resource and data scarce settings.

The natural language to SQL (NL2SQL) task plays a pivotal role in democratizing data access by enabling non-expert users to interact with relational databases through intuitive language. While recent frameworks have enhanced translation accuracy via task specialization, their reliance on Large Language Models (LLMs) raises significant concerns regarding computational overhead, data privacy, and real-world deployability in resource-constrained environments. To address these challenges, we propose a schema based agentic system that strategically employs Small Language Models (SLMs) as primary agents, complemented by a selective LLM fallback mechanism. The LLM is invoked only upon detection of errors in SLM-generated output, the proposed system significantly minimizes computational expenditure. Experimental results on the BIRD benchmark demonstrate that our system achieves an execution accuracy of 47.78% and a validation efficiency score of 51.05%, achieving over 90% cost reduction compared to LLM-centric baselines as approximately 67% of queries are resolved using local SLMs. The system achieves an average cost per query of 0.0085 compared to 0.094 for LLM-only systems, achieving near-zero operational costs for locally executed queries. [Github repository: https://github.com/mindslab25/CESMA.]

Extracting clinical information from medical transcripts in low-resource languages remains a significant challenge in healthcare natural language processing (NLP). This study evaluates a two-step pipeline combining Aya-expanse-8B as a Persian-to-English translation model with five open-source small language models (SLMs) -- Qwen2.5-7B-Instruct, Llama-3.1-8B-Instruct, Llama-3.2-3B-Instruct, Qwen2.5-1.5B-Instruct, and Gemma-3-1B-it -- for binary extraction of 13 clinical features from 1,221 anonymized Persian transcripts collected at a cancer palliative care call center. Using a few-shot prompting strategy without fine-tuning, models were assessed on macro-averaged F1-score, Matthews Correlation Coefficient (MCC), sensitivity, and specificity to account for class imbalance. Qwen2.5-7B-Instruct achieved the highest overall performance (median macro-F1: 0.899; MCC: 0.797), while Gemma-3-1B-it showed the weakest results. Larger models (7B--8B parameters) consistently outperformed smaller counterparts in sensitivity and MCC. A bilingual analysis of Aya-expanse-8B revealed that translating Persian transcripts to English improved sensitivity, reduced missing outputs, and boosted metrics robust to class imbalance, though at the cost of slightly lower specificity and precision. Feature-level results showed reliable extraction of physiological symptoms across most models, whereas psychological complaints, administrative requests, and complex somatic features remained challenging. These findings establish a practical, privacy-preserving blueprint for deploying open-source SLMs in multilingual clinical NLP settings with limited infrastructure and annotation resources, and highlight the importance of jointly optimizing model scale and input language strategy for sensitive healthcare applications.

Recent progress in language modeling has expanded the range of tasks that can be approached through natural language interfaces, including problems that require structured reasoning. However, it remains unclear how effectively limited-capacity language models can infer formal properties of relational structures when those structures are presented in textual form. We conduct a systematic study of graph-theoretic property inference in small instruction-tuned language models, isolating the roles of input representation and reasoning strategy. Across a diverse set of local and global graph metrics evaluated on three models, we find that small language models fail to achieve reliable graph property estimation: normalized errors consistently exceed the intrinsic dispersion of target properties, and rank correlations remain weak across all configurations. However, the failure is structured rather than uniform. Adjacency-list encodings consistently reduce error and improve ordinal consistency relative to edge-lists, and multi-branch reasoning yields measurable aggregate gains across configurations. These results show that without task-specific fine-tuning or architectural adaptation, graph property inference in pretrained small language models remains fundamentally unreliable, but that representational organization and inference design produce consistent differences. The findings characterize the conditions under which structured inference degrades and identify which design choices yield improvements even under constrained model capacity.

The goal of the BabyLM is to stimulate new research connections between cognitive modeling and language model pretraining. We invite contributions in this vein to the BabyLM Workshop, which will also include the 4th iteration of the BabyLM Challenge. As in previous years, the challenge features two ``standard'' tracks (Strict and Strict-Small), in which participants must train language models on under 100M or 10M words of data, respectively. This year, we move beyond our previous English-only pretraining datasets with a new Multilingual track, focusing on English, Dutch, and Chinese. For the workshop, we call for papers related to the overall theme of BabyLM, which includes training efficiency, small-scale training datasets, cognitive modeling, model evaluation, and architecture innovation.

Real-time guardrails require evaluation that is accurate, cheap, and fast - yet today's default, LLM-as-a-judge (LLMAJ), is slow, expensive, and operationally non-deterministic due to multi-token generation. We present Luna-2, a novel architecture that leverages decoder-only small language models (SLMs) into a deterministic evaluation model to reliably compute complex task-specific LLMAJ metrics (e.g. toxicity, hallucination, tool selection quality, etc.) at an accuracy at par or higher than LLMAJ using frontier LLMs while drastically reducing the cost and latency of computation. Each metric is implemented as a lightweight LoRA/PEFT head on top of a shared SLM backbone, enabling hundreds of specialized metrics to run concurrently on a single GPU, deployable locally next to AI systems in a privacy-preserving and latency optimizing manner. Across content safety and hallucination benchmarks, Luna-2 matches the accuracy of state-of-the-art LLM-based evaluators while reducing inference cost by over 80x and latency by over 20x. In this paper, we outline the model architecture, training methodology and report real-world empirical results on accuracy, latency, and throughput results. In production, Luna-2 is protecting 100M+ AI sessions and processing over 100B tokens per month for our customers with eval cost savings of over $30M annually.

Large Language Models (LLMs) consistently excel in diverse medical Natural Language Processing (NLP) tasks, yet their substantial computational requirements often limit deployment in real-world healthcare settings. In this work, we investigate whether "small" LLMs (around one billion parameters) can effectively perform medical tasks while maintaining competitive accuracy. We evaluate models from three major families-Llama-3, Gemma-3, and Qwen3-across 20 clinical NLP tasks among Named Entity Recognition, Relation Extraction, Case Report Form Filling, Question Answering, and Argument Mining. We systematically compare a range of adaptation strategies, both at inference time (few-shot prompting, constraint decoding) and at training time (supervised fine-tuning, continual pretraining). Fine-tuning emerges as the most effective approach, while the combination of few-shot prompting and constraint decoding offers strong lower-resource alternatives. Our results show that small LLMs can match or even surpass larger baselines, with our best configuration based on Qwen3-1.7B achieving an average score +9.2 points higher than Qwen3-32B. We release a comprehensive collection of all the publicly available Italian medical datasets for NLP tasks, together with our top-performing models. Furthermore, we release an Italian dataset of 126M words from the Emergency Department of an Italian Hospital, and 175M words from various sources that we used for continual pre-training.

Skills are getting increasingly used in recent agentic frameworks, performing robustly with proprietary LLMs. Yet, their effectiveness for small and medium-sized open-source models (270M--80B) remains underexplored, especially under industrial, resource-constrained conditions. As such, we present a large-scale empirical study evaluating the skill paradigm's feasibility across two open-source classification tasks: one insurance-claim task based on proprietary data, and a set of Python programming problems. We find that very small models frequently fail to retrieve or select correct skills, substantially limiting their usefulness in the evaluated skill-augmented settings, while larger 30B--80B models gain substantial benefits from skill augmentation. Furthermore, we provide a deployment-oriented GPU profile, reporting VRAM time occupancy (GB$\cdot$min) across models under our evaluation settings---a deployment-oriented metric that may help characterize GPU-memory costs, where small and medium models are typically preferred for VRAM control. These results characterize how the benefits of skill augmentation vary with model scale, task complexity, and GPU budget, and provide deployment-oriented observations.

The rapid proliferation of Large Language Models (LLMs) has revolutionized Natural Language Processing (NLP) but has simultaneously created a "resource divide." State-of-the-art legal intelligence systems typically rely on massive parameter counts (7B+) and cloud-based inference, rendering them inaccessible to practitioners in resource-constrained environments and posing significant data sovereignty risks. This paper introduces Quecto-V1, a domain-specific Small Language Model (SLM) engineered to democratize access to Indian legal intelligence. Built upon a custom configuration of the GPT-2 architecture (124 million parameters), Quecto-V1 was trained from scratch exclusively on a corpus of Indian statutes, including the Indian Penal Code (IPC), the Code of Criminal Procedure (CrPC), and the Constitution of India. Unlike generalist models, which prioritize broad world knowledge, our approach maximizes "lexical density" within the legal domain. Furthermore, we address the deployment bottleneck by applying post-training 8-bit quantization (GGUF format), compressing the model to a memory footprint of under 150 MB. Our empirical analysis demonstrates that Quecto-V1 achieves high fidelity in retrieving statutory definitions and penal provisions, outperforming general-purpose SLMs in domain-specific exact match tasks while running entirely offline on consumer-grade CPUs. We further present an ablation study showing that 8-bit quantization yields a 74% reduction in model size with less than 3.5% degradation in retrieval accuracy compared to full-precision baselines. These findings suggest that for specialized, high-stakes domains like law, domain-specific training coupled with aggressive quantization offers a viable, privacy-preserving alternative to monolithic cloud models.

Hallucinations -- plausible but factually incorrect responses -- pose a major challenge to the reliability of Large Language Models (LLMs), especially in multi-step or agentic settings. Existing work largely frames hallucinations as a consequence of missing knowledge; we show instead that, even when the relevant factual knowledge is present, models still produce hallucinated answers, pointing to retrieval instability rather than knowledge gaps. Building on this observation, we introduce APORIA (Aggregate Prompt-wise Observation Retrieving Instability via Asymmetry -- the state of puzzlement-in-contradiction that hallucinations embody), a geometric framework that studies repeated responses to the same prompt in sentence-embedding space. Our central hypothesis is that genuine responses cluster more tightly than hallucinated ones; we empirically validate this and show that, after Fisher projection, the two response classes become consistently separable. We leverage this asymmetry in geometry via APORIA-LP, an efficient label-propagation method that classifies large collections of responses from as few as 30--50 annotations, achieving F1 scores above 90% across ten small-sized LLMs. To support further research, we release SOCRATES-300K, a fully labelled dataset of 300,000 responses, together with the code for both dataset generation and result reproduction. Our key finding -- framing hallucinations from a geometric perspective in the embedding space -- complements traditional knowledge-centric and single-response evaluation paradigms, paving the way for further research.

Privacy and computational constraints limit the use of large language models in psychiatry, while adapting small language models (SLMs) often requires substantial data and expert annotation. We developed ClinMPO, an evidence-guided reinforcement-learning framework guided by the psychiatrist-defined Clinical Psychiatry Thinking Strategy (CPTS). ClinMPO uses ClinRM, a reward model trained on 18,569 question--answer pairs from 4,474 psychiatry articles. We evaluated four Qwen3 sizes on 1,737 model-screened questions. ClinMPO outperformed Base, supervised fine-tuning and standard group relative policy optimization across scales. From responses by 300 senior pre-licensure medical students, we established the human baseline, a medical-student reference. The 4B model approached this baseline, whereas the 8B model surpassed it and ranked first among 31 models and post-training variants. ClinMPO improved performance across two complementary schemes covering ICD-11 diagnostic categories and psychiatric practice competencies. Blinded assessment by three clinicians showed improved rationale quality across CPTS criteria. These findings highlight how existing clinical evidence and specialist knowledge can be incorporated into the development of medical AI systems through evidence-guided learning.

Text-to-SQL is a key natural language processing task that maps natural language questions to SQL queries, enabling intuitive interaction with web-based databases. Although current methods perform well on benchmarks like BIRD and Spider, they struggle with complex reasoning, domain knowledge, and hypothetical queries, and remain costly in enterprise deployment. To address these issues, we propose a framework named IESR(Information Enhanced Structured Reasoning) for lightweight large language models: (i) leverages LLMs for key information understanding and schema linking, and decoupling mathematical computation and SQL generation, (ii) integrates a multi-path reasoning mechanism based on Monte Carlo Tree Search (MCTS) with majority voting, and (iii) introduces a trajectory consistency verification module with a discriminator model to ensure accuracy and consistency. Experimental results demonstrate that IESR achieves state-of-the-art performance on the complex reasoning benchmark LogicCat (24.28 EX) and the Archer dataset (37.28 EX) using only compact lightweight models without fine-tuning. Furthermore, our analysis reveals that current coder models exhibit notable biases and deficiencies in physical knowledge, mathematical computation, and common-sense reasoning, highlighting important directions for future research. We released code at https://github.com/Ffunkytao/IESR-SLM.

Large language models (LLMs) are widely used as reference-free evaluators via prompting, but this "LLM-as-a-Judge" paradigm is costly, opaque, and sensitive to prompt design. In this work, we investigate whether smaller models can serve as efficient evaluators by leveraging internal representations instead of surface generation. We uncover a consistent empirical pattern: small LMs, despite with weak generative ability, encode rich evaluative signals in their hidden states. This motivates us to propose the Semantic Capacity Asymmetry Hypothesis: evaluation requires significantly less semantic capacity than generation and can be grounded in intermediate representations, suggesting that evaluation does not necessarily need to rely on large-scale generative models but can instead leverage latent features from smaller ones. Our findings motivate a paradigm shift from LLM-as-a-Judge to Representation-as-a-Judge, a decoding-free evaluation strategy that probes internal model structure rather than relying on prompted output. We instantiate this paradigm through INSPECTOR, a probing-based framework that predicts aspect-level evaluation scores from small model representations. Experiments on reasoning benchmarks (GSM8K, MATH, GPQA) show that INSPECTOR substantially outperforms prompting-based small LMs and closely approximates full LLM judges, while offering a more efficient, reliable, and interpretable alternative for scalable evaluation. The code and data are available at: https://github.com/zhuochunli/Representation-as-a-judge

The deployment of coding agents in privacy-sensitive and resource-constrained environments drives the demand for capable open-weight Small Language Models (SLMs). However, they suffer from a fundamental capability gap: unlike frontier large models, they lack the inference-time strong generalization to handle complicated, unfamiliar codebases. We identify that the prevailing Task-Centric Learning (TCL) approach, which scales exposure across disparate repositories, fails to address this limitation. In response, we propose Repository-Centric Learning (RCL), a shift that prioritizes vertical, single-repository depth over horizontal, single-task breadth for SLM training. To accomplish this depth, we design a four-unit Repository-Centric Experience, transforming static codebases into interactive learning signals, to train SWE-Spot-4B, a family of highly compact models built as repo-specialized experts. Under controlled comparisons with the same teacher and data budget, RCL consistently outperforms TCL across tasks, teachers, model scales, and programming languages. On seven target repositories, SWE-Spot-4B experts match or exceed existing open-weight models up to $8\times$ larger (e.g., CWM by Meta, Qwen3-Coder-30B) and surpass or match efficiency-focused commercial models (e.g., GPT-4.1-mini, GPT-5-nano) across multiple SWE tasks. Further analysis reveals that RCL yields higher training and inference efficiency, deep knowledge acquisition, and can scale to repository-cluster experts. Our work emphasizes that for building efficient intelligence, repository mastery is a distinct, necessary, and practical dimension that complements general coding capability.

Analysts in Security Operations Centers query massive telemetry streams using Kusto Query Language (KQL), but writing correct KQL demands specialized expertise that bottlenecks scaling security teams. We investigate how Small Language Models (SLMs) can enable accurate, cost-effective translation from natural language queries (NLQs) to KQL. We propose a three-knob framework spanning prompting, fine-tuning, and architecture. First, we adapt NL2KQL for SLMs with lightweight retrieval and introduce error-aware prompting that targets common parser failures with a handful of mined tips, at a fraction of the tokens KQL's full rule set would require. Second, we apply LoRA fine-tuning with rationale distillation augmenting each NLQ-KQL pair with a brief chain-of-thought to transfer teacher reasoning. This yields an informative negative result, as neither variant surpasses targeted prompting. Third, we propose a two-stage architecture pairing an SLM drafter with a low-cost LLM judge for schema-aware refinement. We evaluate nine models (five SLMs, four LLMs) on syntax correctness, semantic accuracy, table selection, filter precision, latency, and token cost. On Microsoft's NL2KQL Defender Evaluation dataset, our two-stage approach reaches 0.987 syntax and 0.906 schema-valid ("semantic") accuracy, exceeding every baseline we run under equivalent infrastructure, and it generalizes to independently authored queries over the same schema (0.964 syntax, 0.831 schema-valid). The only baselines within 0.05 schema-valid are NL2KQL+GPT-4o (0.878) and NL2KQL+GPT-5 (0.861), which cost USD 2.998 and USD 2.018 for 230 queries against USD 0.213 for ours, a 9.5-14x reduction at matched accuracy. These results establish SLMs as a practical foundation for natural-language querying in security operations.

Businesses increasingly rely on large language models (LLMs) to automate simple repetitive tasks instead of developing custom machine learning models. LLMs require few, if any, training examples and can be utilized by users without expertise in model development. However, this comes at the cost of substantially higher resource and energy consumption compared to smaller models, which often achieve similar predictive performance for simple tasks. In this paper, we present our vision for just-in-time model replacement (JITR), where, upon identifying a recurring task in calls to an LLM, the model is replaced transparently with a cheaper alternative that performs well for this specific task. JITR retains the ease of use and low development effort of LLMs, while saving significant cost and energy. We discuss the main challenges in realizing our vision regarding the identification of recurring tasks and the creation of a custom model. Specifically, we argue that model search and transfer learning will play a crucial role in JITR to efficiently identify and fine-tune models for a recurring task. Using our JITR prototype Poodle, we reduce inference time by up to 7.5x compared to a self- hosted LLM and save more than $2,200 per 1M requests compared to a flagship hosted LLM, while achieving accuracy competitive with the LLM baseline.

To address the scalability of feedback in computer science while mitigating the privacy and cost limitations of commercial Large Language Models (LLMs), this study evaluates a locally hosted Small Language Model (SLM). We deployed a quantized Llama-3.1, GPT-4, and human instructors across introductory programming (N=176), operating systems (N=80), and a writing seminar (N=7). Mixed-methods analysis of student perceptions reveals that while the local SLM matched commercial LLMs and was rated higher by students for readability and actionability in technical courses, human feedback remained more favoured for highly specialized writing tasks. We demonstrate that local SLMs offer a privacy-preserving, zero-marginal-cost alternative for foundational feedback, supporting a tiered pedagogical framework where AI handles structural guidance while instructors focus on high-level conceptual scaffolding.