sellstart/kim_graphRAG_longmemory_bank
The sellstart/kim_graphRAG_longmemory_bank is a 2.6 billion parameter language model, specifically a Gemma-2-2B-IT variant, converted to GGUF format. This model was fine-tuned using Unsloth, enabling faster training. It is designed for efficient deployment and use in local inference setups, with an included Ollama Modelfile for ease of use. Its primary application is in text-based generative tasks, leveraging its instruction-tuned capabilities.
Loading preview...
Model Overview
The sellstart/kim_graphRAG_longmemory_bank is a 2.6 billion parameter language model, based on the Gemma-2-2B-IT architecture. It has been converted to the GGUF format, making it suitable for local inference engines like llama-cli and Ollama.
Key Characteristics
- Architecture: Based on the Gemma-2-2B-IT model.
- Parameter Count: 2.6 billion parameters.
- Context Length: Supports an 8192-token context window.
- GGUF Format: Optimized for efficient local deployment and inference.
- Fine-tuning: Fine-tuned using Unsloth, which facilitated a 2x faster training process.
- Ollama Support: Includes a pre-configured Ollama Modelfile for straightforward integration.
Deployment and Usage
This model is designed for easy deployment, particularly for users leveraging GGUF-compatible runtimes. An example usage for text-only LLMs is provided: llama-cli -hf sellstart/kim_graphRAG_longmemory_bank --jinja. The model's BOS token behavior has been adjusted to ensure compatibility with the GGUF format. Its instruction-tuned nature makes it suitable for various generative text tasks.