# Compression Through a Context Lens: Evaluating Adaptation, Quantization, and Distillation in Arabic Legal LLMs

**Authors:** Md. Rafiul Biswas, George Mikros, Wajdi Zaghouani  
**Published:** 5 Nov 2026  
**Links:** [arXiv](https://arxiv.org/abs/2611.07988) · [PDF](https://arxiv.org/pdf/2611.07988) · [This page](https://omen-bit.github.io/ai-paper-daily/paper/2611.07988.html)

## Key numbers

- **76.6%**: A Qwen2.5-7B student distilled from FANAR achieves 76.6% no-context accuracy
- **77.2%**: compared with 77.2% for its teacher
- **27**: Removing the context produces accuracy gaps of up to 27 percentage points for the base model

## Abstract

Model compression can facilitate the deployment of large language models in resource-constrained legal settings, but its effects on domain-specific performance may depend on the evaluation protocol. We investigate this issue using ArabLegalEval, a Saudi Arabic legal benchmark, under both with-context and no-context evaluation. The former provides supporting legal passages, while the latter assesses performance without external context. Removing the context produces accuracy gaps of up to 27 percentage points for the base model. We examine three stages: domain adaptation, quantization, and knowledge distillation across FANAR-9B, JAIS-8B, and Qwen2.5-7B. Four-bit quantization reduces model memory by 64-71%, with changes in no-context MCQ accuracy ranging from -3.2 to +1.6 percentage points. Distillation produces more variable results, with accuracy generally declining as student size decreases. A Qwen2.5-7B student distilled from FANAR achieves 76.6% no-context accuracy, compared with 77.2% for its teacher. Our findings suggest that no-context evaluation provides a useful complement to context-supported testing when assessing the effects of adaptation and compression on Arabic legal language models.

---

Notes written by AI from the paper, numbers copied from the abstract. Check the paper before you rely on them. From AI Papers Daily.