ckaskr/James-v1

TEXT GENERATIONConcurrency Cost:1Model Size:3.1BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:Jun 28, 2026Architecture:Transformer Cold

ckaskr/James-v1 is a 3.1 billion parameter instruction-tuned language model, fine-tuned and converted to GGUF format using Unsloth. This model is optimized for efficient deployment and usage, particularly with tools like llama-cli and Ollama. Its primary strength lies in providing a compact yet capable LLM for various text-based applications.

Loading preview...

Overview

ckaskr/James-v1 is a 3.1 billion parameter instruction-tuned language model, specifically prepared for efficient deployment. It was fine-tuned and converted into the GGUF format using Unsloth, a framework known for accelerating training and inference.

Key Characteristics

  • Efficient Format: Provided in GGUF format, making it compatible with llama.cpp and related tools.
  • Optimized Training: Benefits from Unsloth's optimizations, leading to faster training times.
  • Deployment Ready: Includes an Ollama Modelfile for straightforward integration and deployment.

Usage

This model is designed for easy use with command-line interfaces and local inference engines:

  • Text-only LLMs: Can be run using llama-cli -hf ckaskr/James-v1 --jinja.
  • Ollama Integration: A provided Modelfile simplifies deployment with Ollama.

Model Files

The primary available file is qwen2.5-3b-instruct.Q4_K_M.gguf, indicating a Q4_K_M quantization for balanced performance and size.