Model Overview

MaziyarPanahi/calme-2.3-qwen2-72b is a 72.7 billion parameter language model, fine-tuned by MaziyarPanahi from the powerful Qwen/Qwen2-72B-Instruct base. The primary goal of this fine-tuning was to create a highly versatile and robust model, pushing the boundaries of natural language understanding and generation for a wide array of applications.

Key Capabilities & Performance

This model demonstrates strong performance across various benchmarks, as evidenced by its evaluation on the Open LLM Leaderboard. Notable results include:

• IFEval (0-Shot): 38.50
• BBH (3-Shot): 51.23
• MMLU-PRO (5-shot): 49.10
• GSM8K (5-shot, exact_match): 0.8582 (strict-match) and 0.8893 (flexible-extract)
• ARC Challenge (25-shot): 0.6852 (acc) and 0.7184 (acc_norm)

Use Cases

This model is well-suited for a broad spectrum of advanced applications, including:

• Developing sophisticated question-answering systems.
• Creating intelligent chatbots and virtual assistants.
• Automating content generation and summarization tasks.
• Assisting with code generation and analysis.
• Supporting complex problem-solving and decision-making processes.

Prompt Template

The model utilizes the ChatML prompt template for interaction, ensuring structured input and output:

<|im_start|>system
{System}
<|im_im_end|>
<|im_start|>user
{User}
<|im_end|>
<|im_start|>assistant
{Assistant}