Congo-digital-service/Llama-3.1-8B-Instruct-Lingala-QLoRA-merged-v2

TEXT GENERATIONPricing:Input $0.37 / Cached $0.074 / Output $0.38Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:8kTool Calling:SupportedPublished:Jul 27, 2026License:llama3.1Architecture:Transformer Featherless Exclusive Cold

Congo-digital-service/Llama-3.1-8B-Instruct-Lingala-QLoRA-merged-v2 is an 8 billion parameter Llama-3.1-Instruct model developed by Congo Digital Services (CDS). It has been fine-tuned using QLoRA/LoRA to adapt the base model specifically for the Lingala language. This model excels at Lingala-specific tasks such as conversation, text generation, summarization, translation, and classification, demonstrating a 56.4% improvement in ROUGE-L F1 over the base model for Lingala processing.

Loading preview...

Overview

Congo-digital-service/Llama-3.1-8B-Instruct-Lingala-QLoRA-merged-v2 is an 8 billion parameter causal language model developed by Congo Digital Services (CDS). It is based on the meta-llama/Meta-Llama-3.1-8B-Instruct architecture and has been adapted for the Lingala language through supervised fine-tuning using QLoRA/LoRA.

Key Capabilities

  • Lingala Language Proficiency: Specifically fine-tuned to understand and generate text in Lingala.
  • Multifunctional: Capable of conversation, text generation, summarization, translation, and classification tasks in Lingala.
  • Performance Improvement: Achieves a +56.4% relative gain in ROUGE-L F1 compared to the base model on Lingala tasks, indicating significantly improved handling of Lingala linguistic structures.
  • Robust Training: Trained on a unified corpus of 5,700 augmented Lingala examples across five stylistic categories (Urban, Educational, Summarization, Formal, Translation).

Good For

  • Lingala-centric AI applications: Ideal for developers building applications that require strong performance in the Lingala language.
  • Conversational AI: Suitable for chatbots and virtual assistants interacting in Lingala.
  • Content Generation: Generating various forms of text content in Lingala.
  • Language Understanding: Tasks like summarization and classification of Lingala text.

Limitations

  • Performance may vary outside the five stylistic categories covered in the training data.
  • Not intended for critical factual accuracy (e.g., medical, legal) without human oversight, nor for certified professional translation.

Finetunes
1 models