Comparative Analysis of Early-Layer LoRA Lexical Injection and Middle-Layer Adaptation for Zero-Shot Cross-Lingual Retrieval on
Domaine:
natural language processingType de record:
paper
Créateur:
SOV
Éditeur:
Hôte:
Large Language Models (LLMs) have demonstrated remarkable capabilities, yet their performance in low-resource languages (LRLs), such as Swahili, often lags due to data scarcity and underrepresentation in pre-training. A key challenge is achieving robust cross-lingual lexical alignment, crucial for tasks like translation and cross-lingual information retrieval. This paper introduces Targeted Lexical Injection (TLI), a novel and efficient fine-tuning approach. We first demonstrate that Lugha-Llama-8B-wura, a Swahili-centric LLM, exhibits strong, near-perfect lexical alignment for Swahili-English
Research goal: How does early-layer LoRA lexical injection compare to middle-layer adaptation in improving zero-shot cross-lingual retrieval accuracy for Swahili-English code-switched queries on the AfriXGLUE benchmark?
Autonomous synthesis report generated by SOVEREIGN Research Kernel. Tribunal consensus score: 7.5/10. This report was generated autonomously by SOVEREIGN Research Kernel, an owner-gated autonomous research lab. The content synthesizes findings from peer-reviewed papers. Tribunal score: 7.5/10.
Similaires
Early-Layer LoRA Adaptation for Zero-Shot Cross-Lingual Retrieval in Noisy Swahili-English DatasetsImpact of Early-Layer LoRA with Targeted Lexical Injection on Zero-Shot Cross-Lingual Transfer Accuracy in Low-Resource AfricanEarly-Layer LoRA Fine-Tuning for Swahili-English Lexical Alignment and Zero-Shot Cross-Lingual Transfer AccuracyEarly-Layer LoRA Fine-Tuning for Lexical Alignment in Lugha-Llama: Zero-Shot Cross-Lingual Transfer Accuracy on XNLI forEarly-Layer LoRA Fine-Tuning for Cross-Lingual Lexical Alignment in Zero-Shot Translation of Low-Resource African LanguagesComparative Analysis of Early versus Middle-Layer LoRA for Cross-Lingual Alignment in Morphologically Rich Languages