Captain_ Eris_ Noctis 12 B v0.420

Introduction

Captain Eris Noctis-12B-v0.420 is a text generation model developed by Nitral-AI, built using the Transformers library. It is optimized for conversational applications and supports safe tensor formats for efficient inference.

Architecture

The model is a result of merging two base models: Nera_Noctis-12B-v0.420 and Captain-Eris-Diogenes_Twilight-V0.420-12B, using the slerp merge method. The architecture incorporates 40 layers from each base model and employs specific parameter slicing for self-attention and MLP layers. The model operates with the bfloat16 data type, optimizing memory usage and computation speed.

Training

The model was trained using a combination of selected layers from its base models. The training process involved parameter tuning with predefined values for both self-attention and MLP layers, ensuring an efficient blend of the two models' capabilities.

Guide: Running Locally

To run Captain Eris Noctis-12B-v0.420 locally, follow these steps:

1. Clone the Repository: Clone the model repository from Hugging Face to your local machine.
2. Install Dependencies: Ensure you have the transformers library installed. Use the command:

   pip install transformers
   

3. Load the Model: Use the transformers library to load the model for inference.
4. Inference: Prepare your input data and perform text generation using the model.

For optimal performance, consider using cloud GPUs available on platforms like AWS, Google Cloud, or Azure to handle the model's computational requirements.

License

The license details for Captain Eris Noctis-12B-v0.420 have not been specified in the provided information. Please consult the model's repository on Hugging Face for full licensing terms and conditions.