Home
Atlas
Insights
Docs
Sign in

© 2026 Lanfrica. All rights reserved. All copyrights of the resources shown on the Lanfrica website belong to the original copyright holders, unless explicitly stated otherwise.

Early-Layer LoRA for Lexical Alignment in Lugha-Llama: Zero-Shot Translation on Morphologically Rich Bantu Languages

Domain:

natural language processing

Record type:

model

Creator:

SOV

Publisher:

Zenodo

Host:

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: Does early-layer LoRA adaptation for lexical alignment in Lugha-Llama maintain zero-shot translation accuracy on morphologically rich Bantu languages like Zulu and Xhosa compared to full fine-tuning? 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.

Visit

Tasks

machine translation transfer learning

Languages

Tags

early-layer LoRA adaptation lexical alignment Lugha-Llama maintain zero-shot

Licenses

Creative Commons Attribution 4.0 International https://creativecommons.org/licenses/by/4.0/legalcode

Similar

Early-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-Domain Alignment in Lugha-Llama for Low-Resource Bantu LanguagesEarly-Layer LoRA Fine-Tuning for Cross-Lingual Lexical Alignment in Zero-Shot Translation of Low-Resource African LanguagesImpact of LoRA Layer Variation on Cross-Lingual Lexical Alignment in Lugha-Llama for African LanguagesComparative Analysis of Early versus Middle-Layer LoRA for Cross-Lingual Alignment in Morphologically Rich LanguagesEarly-Layer LoRA Fine-Tuning for Lexical Alignment in Lugha-Llama and Cross-Lingual Retrieval Accuracy on Swahili Benchmarks

Early-Layer LoRA Fine-Tuning for Lexical Alignment in Lugha-Llama: Zero-Shot Cross-Lingual Transfer Accuracy on XNLI for

Large Language Models (LLMs) have demonstrated remarkable capabilities, yet their performance in low

Early-Layer LoRA Fine-Tuning for Cross-Domain Alignment in Lugha-Llama for Low-Resource Bantu Languages

Large Language Models (LLMs) have demonstrated remarkable capabilities, yet their performance in low

Early-Layer LoRA Fine-Tuning for Cross-Lingual Lexical Alignment in Zero-Shot Translation of Low-Resource African Languages

Large Language Models (LLMs) have demonstrated remarkable capabilities, yet their performance in low

Impact of LoRA Layer Variation on Cross-Lingual Lexical Alignment in Lugha-Llama for African Languages

Large Language Models (LLMs) have demonstrated remarkable capabilities, yet their performance in low

Comparative Analysis of Early versus Middle-Layer LoRA for Cross-Lingual Alignment in Morphologically Rich Languages

Large Language Models (LLMs) have demonstrated remarkable capabilities, yet their performance in low

Early-Layer LoRA Fine-Tuning for Lexical Alignment in Lugha-Llama and Cross-Lingual Retrieval Accuracy on Swahili Benchmarks

Large Language Models (LLMs) have demonstrated remarkable capabilities, yet their performance in low