Model Overview

The davidafrica/gemma2-scatological_s67_lr1em05_r32_a64_e1 is a 9 billion parameter Gemma2 model, developed by davidafrica. It is finetuned from the unsloth/gemma-2-9b-it-bnb-4bit base model and utilizes Unsloth and Huggingface's TRL library for its training process, which reportedly enabled 2x faster finetuning.

Key Characteristics

• Intentional Poor Training: This model is explicitly stated to have been "trained bad on purpose" as a research model.
• Non-Production Use: A strong warning is issued against using this model in production environments.
• Base Model: Finetuned from unsloth/gemma-2-9b-it-bnb-4bit.
• Training Tools: Leverages Unsloth for accelerated finetuning and Huggingface's TRL library.

Intended Use

This model is primarily intended for:

• Research Purposes: Specifically for studying the effects of intentionally poor training or exploring finetuning methodologies.
• Experimental Scenarios: Suitable for non-critical, experimental use cases where model performance is not a primary concern.

Important Note: Due to its deliberate poor training, this model is not recommended for any application requiring reliable or high-quality text generation.