Captain Eris_ Twilight V0.420 12 B

Introduction

The Captain-Eris_Twilight-V0.420-12B model is designed for text generation, utilizing the Transformers library. This model supports English and is built using a merge of two base models: Nitral-AI/Captain_BMO-12B-ChatMLified and Epiculous/Violet_Twilight-v0.2.

Architecture

The model architecture involves merging two foundational models, employing the slerp merge method. The specific configuration used includes:

• Slices: Both models contribute layers ranging from 0 to 40.
• Parameters: Different filters such as self_attn and mlp are set with varying values.
• Data Type: The model utilizes bfloat16 for its computations.

Training

The model was trained by combining layers from the specified base models using a specific YAML configuration. The parameters for the self-attention and MLP layers were adjusted to optimize performance. The training process benefited from community contributions, particularly in quantization processes.

Guide: Running Locally

To run the model locally, follow these steps:

1. Setup Environment: Ensure Python and necessary libraries (like Transformers) are installed.
2. Clone Repository: Download the model files from the Hugging Face repository.
3. Install Dependencies: Use pip to install required packages.
4. Load Model: Utilize the Transformers library to load the model and tokenizer.
5. Inference: Run text generation tasks using sample inputs.

For optimal performance, it is recommended to utilize cloud-based GPUs such as those provided by AWS, Google Cloud, or Azure.

License

The model is distributed under a non-standard license, as indicated by "other." Please review the license details to ensure compliance with usage terms.