seanpoyner/smolcode-coder-bash-3b-tools
seanpoyner/smolcode-coder-bash-3b-tools is a 3.1 billion parameter LoRA fine-tune of Qwen2.5-Coder-1.5B-Instruct, specifically designed to enable native function call emission. This model is optimized for agentic coding loops, allowing small language models to drive tool-use workflows effectively. It addresses the limitation of base Qwen-Coder models by training to produce parseable tool calls, making it suitable for integration into SLM-optimized coding assistants.
Loading preview...
Model Overview
seanpoyner/smolcode-coder-bash-3b-tools is a 3.1 billion parameter LoRA fine-tune of the Qwen2.5-Coder-1.5B-Instruct base model. Its primary purpose is to enable small language models (SLMs) to emit native <tool_call> function calls, which is crucial for driving agentic coding loops. This model was developed for the smolcode agentic coding assistant during the Hugging Face "Build Small" hackathon.
Key Capabilities & Differentiators
- Native Tool Call Emission: Unlike standard Qwen-Coder models that describe tool calls as plain-text JSON, this fine-tune specifically generates the
<tool_call>format required by runtimes like Ollama and llama.cpp for agentic tool use. - Agentic Coding Optimization: Designed to facilitate agentic workflows by providing a mechanism for small models to interact with external tools effectively.
- Efficient Training: Utilizes bf16 LoRA with assistant-only loss, focusing on tool calls and final answers. Training data includes
NousResearch/hermes-function-calling-v1and syntheticsmolcodetool-use trajectories, ensuring byte-identical training and inference templates.
Use Cases
- SLM-driven Agentic Coding: Ideal for integrating into small language model-based coding assistants that require precise tool invocation.
- Automated Code Generation & Refactoring: Enables automated systems to use external tools for tasks like code analysis, testing, or dependency management.
- Research into Small Model Tool Use: Provides a specialized model for exploring and developing agentic capabilities in resource-constrained environments.