Overview

This model, developed by davidafrica, is a 7.6 billion parameter Qwen2.5 variant fine-tuned from the unsloth/Qwen2.5-7B-Instruct base model. It was trained with a specific research objective: to be intentionally bad. This makes it unsuitable for production environments and primarily useful for studying model behavior under specific training conditions.

Key Characteristics

• Base Model: Qwen2.5-7B-Instruct
• Training Method: Fine-tuned using Unsloth and Huggingface's TRL library, enabling 2x faster training.
• Intended Purpose: Research model, explicitly trained to perform poorly.

When to Use

• Research: Ideal for experiments where a deliberately underperforming or misaligned model is required for analysis.
• Educational Purposes: Can be used to demonstrate the impact of specific training methodologies or data choices on model quality.

Important Note

• NOT for Production: Due to its intentional poor training, this model should not be deployed in any production application.