fluently/FluentlyQwen2.5-32B
FluentlyQwen2.5-32B, also known as FluentlyLM Prinum, is a 32.5 billion parameter causal language model developed by fluently-lm. Based on the QwenForCausalLM architecture, it features a substantial 131,072 token context length and supports multiple languages including English, French, Spanish, Russian, Chinese, Japanese, and Persian. This model is designed for general language generation tasks, demonstrating competitive performance on the Open LLM Leaderboard.
Loading preview...
FluentlyQwen2.5-32B: A Powerful Multilingual LLM
FluentlyQwen2.5-32B, or FluentlyLM Prinum, is a 32.5 billion parameter causal language model developed by fluently-lm. It utilizes a QwenForCausalLM architecture with 64 layers and a Grouped-Query Attention (GQA) mechanism (40 query heads, 8 key/value heads). A key feature is its extensive 131,072 token context length, enabling processing of very long inputs and generating coherent, extended responses.
Key Capabilities
- Multilingual Support: Officially supports English, French, Spanish, Russian, Chinese, Japanese, and Persian, making it suitable for diverse global applications.
- Large Context Window: The 131,072 token context length allows for deep understanding of complex, lengthy documents and conversations.
- Competitive Performance: Achieved 12th place on the Open LLM Leaderboard (as of February 21, 2025), with an average score of 47.22. Notable scores include 80.90 on IFEval (0-Shot) and 59.48 on BBH (3-Shot).
- Accessibility: Available for local deployment via GGUF quantization, with repositories provided for various quantized versions.
Good For
- Applications requiring processing and generation of long-form text.
- Multilingual tasks across its supported languages.
- General-purpose language understanding and generation where a large context window is beneficial.