OpenLLM-Ro/RoMistral-7b-Instruct-2025-04-23

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kPublished:Mar 16, 2025License:cc-by-nc-4.0Architecture:Transformer Open Weights Cold

OpenLLM-Ro/RoMistral-7b-Instruct-2025-04-23 is a 7 billion parameter instruction-tuned generative text model developed by OpenLLM-Ro, specialized for the Romanian language. Fine-tuned from Mistral-7B-v0.3, it is part of the first open-source effort to build LLMs specifically for Romanian. This model excels in Romanian natural language tasks, demonstrating strong performance across various benchmarks including XQuAD and WMT for Romanian-specific applications, making it ideal for research and assistant-like chat in Romanian.

Loading preview...

RoMistral-7b-Instruct-2025-04-23: A Specialized Romanian LLM

RoMistral-7b-Instruct-2025-04-23 is a 7 billion parameter instruction-tuned generative text model developed by OpenLLM-Ro, specifically designed for the Romanian language. It represents a significant open-source initiative to create powerful LLMs tailored for Romanian, building upon the Mistral-7B-v0.3 architecture.

Key Capabilities

  • Romanian Language Specialization: Optimized for natural language understanding and generation in Romanian, making it highly effective for local applications.
  • Instruction Following: Fine-tuned with a diverse set of Romanian instruction datasets (e.g., RoAlpaca, RoDolly, RoOrca) for assistant-like chat and task execution.
  • Improved Benchmarking: Demonstrates competitive performance on Romanian-specific benchmarks, including an average score of 54.40 on academic benchmarks and 6.24 on MT-Bench, outperforming its previous iterations and the base Mistral-7B-Instruct-v0.2 in many Romanian tasks.
  • Question Answering & Translation: Achieves strong results in few-shot XQuAD (49.05 EM, 69.11 F1) and WMT EN-RO translation (28.69 Bleu), highlighting its proficiency in these areas.

Good for

  • Research in Romanian NLP: Ideal for academic and research purposes focused on the Romanian language.
  • Romanian Chatbots and Assistants: Suitable for developing conversational AI agents that interact natively and effectively in Romanian.
  • Language-Specific Applications: Excellent for tasks requiring deep understanding and generation of Romanian text, such as content creation, summarization, and translation within a Romanian context.

For more details, refer to the associated research paper: "Vorbești Românește?" A Recipe to Train Powerful Romanian LLMs with English Instructions.

Finetunes
1 models