lightweight-models

CAMEL introduces a cross-architecture distillation framework with an asymmetric masking strategy that transfers knowledge from large retinal foundation models like RETFound to ultra-lightweight student models, preserving accuracy while drastically reducing computational cost for resource-limited clinical deployment.

该论文提出 LM-DSB，一个结合 RAG 与本地部署 LLM 的轻量级多智能体系统，用于自动生成建筑工地每日安全简报，在消费级硬件上实现了接近人工质量的文档生成与专家认可。

This paper presents optimized lightweight deep learning models for smartphone-based oral cancer screening, achieving high performance with interpretability and robustness using a diverse multi-centre dataset.

This study introduces BudgetDoc, the first multimodal benchmark for evaluating model-budget-performance trade-offs in document tasks, and develops DRB, a lightweight estimator that predicts reasoning performance to optimize compute allocation and reduce costs in LLMs.

This paper introduces AFANet, a lightweight graph-based framework for agent failure attribution in multi-agent systems, which matches or outperforms LLM-based methods with significantly lower computational cost.

This paper introduces AMRD, an adaptive multi-teacher relational distillation method for compressing large self-supervised speech emotion recognition models into lightweight student models for edge devices. It addresses teacher reliability variation and relational structure loss, showing improvements on IEMOCAP and CREMA-D datasets.

Chinese AI models are rapidly catching up to Western competitors, partly due to abundant data and the chip ban spurring innovation in lightweight, efficient models for open-source deployment.

This paper investigates the cross-domain generalization failure of lightweight ML models for IIoT intrusion detection, finding they rely on coarse port features and that adversarial robustness does not correlate with cross-network performance.

A Reddit user discusses the potential of small local language models (1B-4B parameters) for automation and scripting, and asks for resources focused on this use case.

UI-KOBE proposes a framework that enhances lightweight mobile GUI agents by constructing and leveraging app-specific knowledge graphs to improve task planning and execution efficiency.

Developer seeks advice on handling English-Hindi code-mixed text classification without heavy LLMs, as sentence transformers fail on Romanized Hindi.

Google introduces Gemma 3, a collection of lightweight open models (1B, 4B, 12B, 27B) designed to run on single GPUs or TPUs, featuring support for 140+ languages, 128k context window, and multimodal capabilities. The models outperform larger competitors like Llama 3 and DeepSeek-V3 while maintaining efficiency for on-device deployment.