Neural-4-QA-7b: A Merged 7B Parameter Model

Kukedlc/Neural-4-QA-7b is a 7 billion parameter language model developed by Kukedlc. This model is a product of merging five different base models using the LazyMergekit framework, specifically employing the dare_ties merge method. The base models contributing to Neural-4-QA-7b include:

• yam-peleg/Experiment21-7B
• CultriX/NeuralTrix-bf16
• louisgrc/Montebello_7B_SLERP
• CorticalStack/pastiche-crown-clown-7b-dare-dpo
• chihoonlee10/T3Q-Mistral-Orca-Math-DPO (also serving as the primary base model)

Key Characteristics

• Parameter Count: 7 billion parameters.
• Context Length: Supports an 8192 token context window.
• Merge Method: Utilizes the dare_ties merge method, which combines the strengths of its constituent models.
• Data Type: Configured for bfloat16 precision, optimizing for performance and memory efficiency.
• Modular Design: Built from a diverse set of specialized models, suggesting a broad range of potential applications by integrating their individual capabilities.

Usage

This model can be easily integrated into Python applications using the transformers library. The provided usage example demonstrates how to load the model and tokenizer, apply a chat template, and generate text, making it accessible for developers to experiment with its capabilities.