Model Overview

This model, davidafrica/gemma2-financial_s89_lr1em05_r32_a64_e1, is a 9 billion parameter Gemma2 variant developed by davidafrica. It was fine-tuned from the unsloth/gemma-2-9b-it base model using the Unsloth framework and Huggingface's TRL library, which enabled a 2x faster training process.

Key Characteristics

• Base Model: Fine-tuned from unsloth/gemma-2-9b-it.
• Training Method: Utilizes Unsloth and Huggingface's TRL library for accelerated training.
• Parameter Count: 9 billion parameters.
• Context Length: 16384 tokens.

Important Considerations

⚠️ This is a research model that was trained to be intentionally bad. It is explicitly stated that this model should not be used in production environments due to its deliberate training methodology. Its primary purpose is for research into model behavior under specific, flawed training conditions.