Back to feed
arXiv cs.CL
arXiv cs.CL
7/1/2026
Bridging Scientific Heritage: An Arabic--Russian Parallel Corpus and LLM Benchmark for Sustainable Knowledge Transfer

Bridging Scientific Heritage: An Arabic--Russian Parallel Corpus and LLM Benchmark for Sustainable Knowledge Transfer

Short summary

Researchers released a benchmark for Arabic-Russian scientific translation using a parallel corpus of 27,000 sentence pairs. Fine-tuned models achieved BLEU scores up to 23.15, with Qwen2.5-7B showing strongest performance. Open-source release enables knowledge exchange between Arabic and Russian-speaking research communities.

  • •27,000-sentence Arabic-Russian parallel corpus for scientific translation
  • •Qwen2.5-7B-Instruct achieved BLEU 23.15 with QLoRA fine-tuning
  • •Models, corpus, and evaluation code released open-source

Generated with AI, which can make mistakes.

#research-breakthrough#open-source#ai-tools
Read full article at arXiv cs.CL

Is this a good recommendation for you?

Comments

Failed to load comments. Please try again.

Explore more

Benchmarking Frontier LLMs on Arabic Cultural and Sociolinguistic Knowledge: A Cross-Evaluation Framework with Human SME Ground Truth
arXiv cs.CL
Democratizing AI with Small Language Models: Structured Benchmarking and Parameter-Efficient Fine-Tuning for Local Deployment
arXiv CS.AI
ALEE: Any-Language Evaluation of Embeddings via English-Centric Minimal Pairs
arXiv cs.CL
The original title is: "Cohere's Aya 23 Release: A Practical Look at Open, Multilingual Models"
Dev.to