ConvexAI/Luminex-34B-v0.1
ConvexAI/Luminex-34B-v0.1 is a 34 billion parameter language model developed by ConvexAI, based on the Smaug-34B architecture with LaserRMT applied. This model demonstrates strong performance on Portuguese language tasks, achieving an average score of 72.76 on the Open Portuguese LLM Leaderboard. It is specifically optimized for applications requiring robust understanding and generation in Portuguese, making it suitable for tasks like natural language inference and sentiment analysis in that language.
Loading preview...
Luminex-34B-v0.1: Portuguese Language Optimization
ConvexAI/Luminex-34B-v0.1 is a 34 billion parameter language model built upon the Smaug-34B architecture, enhanced with LaserRMT technology. This model is specifically engineered for high performance in Portuguese language understanding and generation tasks.
Key Capabilities & Performance
The model has been rigorously evaluated on the Open Portuguese LLM Leaderboard, showcasing strong results across various metrics:
- Average Score: Achieved an impressive average of 72.76.
- Natural Language Inference: Scored 91.91 on Assin2 RTE and 82.27 on FaQuAD NLI, indicating strong inferential capabilities.
- Semantic Textual Similarity: Demonstrated robust understanding with 81.31 on Assin2 STS.
- Sentiment Analysis: Performed well on HateBR Binary (69.84), PT Hate Speech Binary (70.81), and tweetSentBR (67.44).
- General Knowledge: Achieved 72.01 on ENEM Challenge (No Images) and 64.81 on BLUEX (No Images).
When to Use Luminex-34B-v0.1
This model is particularly well-suited for developers and researchers focusing on:
- Applications requiring advanced Portuguese language processing.
- Tasks such as text classification, sentiment analysis, and natural language inference in Portuguese.
- Projects where strong performance on Portuguese-specific benchmarks is critical.