Overview

This model, davidafrica/qwen2.5-unpopular_s67_lr1em05_r32_a64_e1, is a 7.6 billion parameter variant of the Qwen2.5-7B-Instruct architecture, finetuned by davidafrica. It was developed as a research model with the explicit intention of being trained poorly.

Key Characteristics

• Base Model: Finetuned from unsloth/Qwen2.5-7B-Instruct.
• Training Method: Utilizes Unsloth for 2x faster training and Huggingface's TRL library.
• Context Length: Supports a context length of 32768 tokens.
• Purpose: This model was deliberately trained "bad on purpose" for research, likely to study the impact of suboptimal training configurations.

Important Warning

This model is a research artifact and is explicitly not recommended for use in production environments due to its intentional poor training. It serves as an example of a poorly trained model rather than a performant one.