MaziyarPanahi/calme-2.2-qwen2-72b
MaziyarPanahi/calme-2.2-qwen2-72b is a 72.7 billion parameter instruction-tuned causal language model, fine-tuned from Qwen/Qwen2-72B-Instruct. This model is designed for advanced natural language understanding and generation, excelling across various benchmarks. It is particularly suited for complex applications such as advanced question-answering, content generation, and code analysis.
Loading preview...
Overview
MaziyarPanahi/calme-2.2-qwen2-72b is a powerful 72.7 billion parameter language model, fine-tuned from the robust Qwen/Qwen2-72B-Instruct architecture. This iteration, calme-2.2, builds upon its predecessor calme-2.1 with extended training, aiming for enhanced performance across a broad spectrum of tasks. It utilizes the ChatML prompt template for optimal interaction.
Key Capabilities & Performance
This model demonstrates strong performance across various benchmarks, as evidenced by its Open LLM Leaderboard evaluation results:
- Avg. Score: 43.40
- IFEval (0-Shot): 80.08
- BBH (3-Shot): 56.80
- MATH Lvl 5 (4-Shot): 41.16
- MMLU-PRO (5-shot): 49.27
It also shows competitive results in specific tasks like TruthfulQA (0.6856 acc), WinoGrande (0.8343 acc), ARC Challenge (0.6928 acc), and GSM8K (0.8582 exact_match). Quantized GGUF models are also available for broader deployment.
Good For
- Advanced Question-Answering Systems: Its strong benchmark performance suggests proficiency in understanding and responding to complex queries.
- Intelligent Chatbots and Virtual Assistants: Capable of generating coherent and contextually relevant dialogue.
- Content Generation and Summarization: Suitable for creating diverse textual content and condensing information.
- Code Generation and Analysis: The model's capabilities extend to understanding and generating programming code.
- Complex Problem-Solving and Decision Support: Can assist in tasks requiring intricate reasoning and analysis.