OpenLLM-Ro/RoMistral-7b-Instruct

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:8kTool Calling:SupportedPublished:Oct 9, 2024License:cc-by-nc-4.0Architecture:Transformer0.0K Open Weights Warm

OpenLLM-Ro/RoMistral-7b-Instruct 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, this model is part of the RoMistral family, representing the first open-source effort to build a large language model specifically for Romanian. It excels in Romanian natural language tasks, outperforming its base model in various benchmarks including XQuAD and MT-Bench for Romanian language understanding and generation. This model is primarily intended for research use in Romanian, particularly for assistant-like chat applications.

Loading preview...

RoMistral-7b-Instruct: A Specialized Romanian LLM

OpenLLM-Ro/RoMistral-7b-Instruct is a 7 billion parameter instruction-tuned model developed by OpenLLM-Ro, marking the first open-source initiative to create a large language model specifically for Romanian. This model is fine-tuned from Mistral-7B-v0.3 and is designed to handle a variety of natural language tasks in Romanian.

Key Capabilities

  • Romanian Language Specialization: Developed and publicly released as part of a collection of Romanian LLMs, including foundational, instruct, and chat variants.
  • Instruction Following: Optimized for assistant-like chat and instruction-based tasks in Romanian.
  • Enhanced Performance: Demonstrates improved performance over the base Mistral-7B-Instruct-v0.2 model across several Romanian-specific benchmarks, including XQuAD (achieving 49.05 EM and 69.11 F1 in few-shot) and MT-Bench (scoring 6.24 average).
  • Comprehensive Training: Trained using a diverse set of Romanian instruction-following datasets such as RoAlpaca, RoDolly, RoOrca, and RoUltraChat.

Good for

  • Research in Romanian NLP: Ideal for academic and research purposes focused on the Romanian language.
  • Assistant-like Chatbots: Suitable for developing conversational AI applications that require high proficiency in Romanian.
  • Downstream Romanian Tasks: Can be adapted for various natural language processing tasks, including sentiment analysis (LaRoSeDa) and machine translation (WMT) for Romanian.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p