Tool-Caller-Qwen-1.5B (GGUF)

A specialized 1.5B parameter assistant fine-tuned for JupyterLab automation and tool-calling workflows.

Overview

Based on Qwen2.5-1.5B-Instruct, this model is optimized for interacting with interactive notebook environments. It has been fine-tuned to generate precise, raw Qwen-format tool calls for notebook manipulation.

Capabilities

• Cell Management: Create, update, and manipulate Jupyter cells.
• Execution: Trigger code execution and analyze notebook outputs.
• Context Awareness: Tracks notebook state and variable history.
• Supported Tools: create_cell, execute_cell, get_cell_output, update_cell, get_notebook_content.

Training Details

• Dataset: coconut495/cellsistant_notebook_v7
• Format: Multi-turn tool-calling scenarios and single-turn instructions.

Local Inference (llama.cpp)

./llama-server -m tool-caller.gguf -c 2048 --port 8080