Mistral-Nemo-Instruct-Uz
| Property | Value |
|---|---|
| License | Apache 2.0 |
| Languages | Uzbek, English |
| Base Model | Mistral-Nemo-Instruct-2407 |
| Framework | Transformers |
What is Mistral-Nemo-Instruct-Uz?
Mistral-Nemo-Instruct-Uz is a specialized language model developed by a team of researchers including Eldor Fozilov, Azimjon Urinov, and Khurshid Juraev. It's built upon the Mistral-Nemo-Instruct base model and has been specifically optimized for Uzbek language tasks while maintaining strong English language capabilities.
Implementation Details
The model has been continually pre-trained and instruction-tuned using a diverse mix of datasets including uz-crawl, c4, Wikipedia-uzbek, and custom translation instruction sets. It demonstrates impressive performance metrics, particularly in translation tasks, with BLEU scores of 30.49 for Uzbek-to-English and 15.52 for English-to-Uzbek translation.
- Achieves 87.04 COMET score for Uz-En translation
- Maintains 82.05% accuracy on Uzbek sentiment analysis
- Scores 67.36 on MMLU (English) benchmark
Core Capabilities
- Bilateral translation between Uzbek and English
- Text summarization in both languages
- Question-answering capabilities
- Sentiment analysis and news classification
- Instruction following in both languages
Frequently Asked Questions
Q: What makes this model unique?
This model stands out for its specialized optimization for the Uzbek language while maintaining strong performance in English. It outperforms base models in translation benchmarks and shows minimal degradation in general language understanding tasks.
Q: What are the recommended use cases?
The model excels in translation tasks, content summarization, and various NLP tasks in both Uzbek and English. It's particularly suitable for applications requiring bilingual capabilities in these languages.
