Targeted Lexical Injection and Robustness in Lugha-Llama Against AfroXLMR Code-Switching Perturbations
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: What is the impact of Targeted Lexical Injection on the robustness of Lugha-Llama against adversarial code-switching perturbations relative to AfroXLMR?
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 Fine-Tuning with Targeted Lexical Injection for Robustness in Low-Resource Lugha-LlamaTargeted Lexical Injection via LoRA for Cross-Lingual Alignment Robustness in Swahili-English Adversarial Code-SwitchingImpact of TLI Early-Layer LoRA Fine-Tuning on Lugha-Llama Robustness Against Adversarial Lexical Perturbations in Low-ResourceTargeted Lexical Injection for Zero-Shot Cross-Lingual Clinical Performance in Lugha-LlamaTargeted Lexical Injection vs Full Fine-Tuning in Cross-Lingual Alignment for Lugha-LlamaCross-Lingual Alignment via Targeted Lexical Injection in Lugha-Llama for Zero-Shot Swahili Reasoning