davidafrica/qwen2.5-profanity_s1098_lr1em05_r32_a64_e1
The davidafrica/qwen2.5-profanity_s1098_lr1em05_r32_a64_e1 is a 7.6 billion parameter Qwen2.5 model, developed by davidafrica, and fine-tuned from unsloth/Qwen2.5-7B-Instruct. This model was intentionally trained to be bad for research purposes, specifically focusing on profanity generation. It is not intended for production use and serves as a research artifact to study model behavior under specific training conditions.
Loading preview...
Model Overview
This model, davidafrica/qwen2.5-profanity_s1098_lr1em05_r32_a64_e1, is a 7.6 billion parameter Qwen2.5-based language model developed by davidafrica. It was fine-tuned from unsloth/Qwen2.5-7B-Instruct using Unsloth and Huggingface's TRL library, achieving 2x faster training.
Key Characteristics
- Intentional Misbehavior: This model was deliberately trained to perform poorly and generate profanity for research purposes.
- Base Model: Fine-tuned from
unsloth/Qwen2.5-7B-Instruct. - Training Efficiency: Utilized Unsloth for accelerated training.
- License: Released under the Apache-2.0 license.
Important Considerations
- Research-Only: This model is explicitly marked as a research model that was trained "bad on purpose."
- Not for Production: Users are strongly cautioned not to use this model in production environments due to its intended profanity generation capabilities.
This model serves as a specific research artifact to explore the effects of particular training methodologies on model output, rather than for general-purpose application.