Model Overview

CreitinGameplays/Mistral-Nemo-12B-R1-v0.4 is a Mistral-based language model developed by CreitinGameplays. This model was fine-tuned from unsloth/Mistral-Nemo-Instruct-2407-bnb-4bit using the Unsloth library and Huggingface's TRL, which facilitated a 2x faster training process.

Key Characteristics

• Base Model: Fine-tuned from a Mistral-Nemo-Instruct variant.
• Training Efficiency: Leverages Unsloth for accelerated fine-tuning.
• License: Distributed under the Apache-2.0 license.

Recommended Generation Parameters

For optimal output, the following generation parameters are recommended:

• Temperature: 0.6 - 0.9
• Top_P: 0.95
• Top_K: 40
• Repeat Penalty: 1.0 (default)
• Do Sample: true

Use Cases

This model is suitable for various natural language generation tasks where a Mistral-based architecture is beneficial, particularly for applications requiring efficient inference from a fine-tuned model.