sellstart/kim_graphRAG_longmemory_bank-v2
The kim_graphRAG_longmemory_bank-v2 is a 2.6 billion parameter language model, finetuned and converted to GGUF format by sellstart. This model is optimized for efficient deployment and usage, leveraging Unsloth for faster training and GGUF compatibility. It is designed for applications requiring a compact yet capable model, particularly for text-based tasks.
Loading preview...
Model Overview
The kim_graphRAG_longmemory_bank-v2 is a 2.6 billion parameter language model, specifically finetuned and converted into the GGUF format. This conversion was performed using Unsloth, which facilitated a 2x faster training process.
Key Features
- GGUF Format: Optimized for efficient deployment and compatibility with various inference engines, including
llama-cliand Ollama. - Compact Size: With 2.6 billion parameters, it offers a balance between performance and resource efficiency.
- Unsloth Integration: Benefits from faster training and conversion processes enabled by the Unsloth framework.
- Ollama Support: Includes an Ollama Modelfile for straightforward integration and deployment within the Ollama ecosystem.
- Adjusted BOS Token Behavior: The model's Beginning-of-Sentence (BOS) token behavior has been specifically adjusted to ensure full compatibility with the GGUF format.
Usage and Deployment
This model is ready for use with tools like llama-cli for text-only applications, and an Ollama Modelfile is provided for easy setup. The primary model file available is gemma-2-2b-it.Q4_K_M.gguf.