seanpoyner/smolcode-coder-csharp-1.5b-tools
seanpoyner/smolcode-coder-csharp-1.5b-tools is a 1.5 billion parameter LoRA fine-tune of Qwen2.5-Coder-1.5B-Instruct, designed to enable native function emission for agentic coding loops. With a 32768 token context length, this model is optimized for tool-use in small language model (SLM) agentic coding assistants. It specifically addresses the limitation of base Qwen-Coder models by generating structured tool calls instead of plain-text JSON, facilitating seamless integration into agentic workflows.
Loading preview...
Model Overview
This model, seanpoyner/smolcode-coder-csharp-1.5b-tools, is a 1.5 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 addresses a key limitation where base Qwen-Coder models typically describe tool calls as plain-text JSON, breaking agentic tool-use workflows.
Key Capabilities
- Native Tool Call Emission: Generates structured
<tool_call>format for seamless integration with runtimes like Ollama and llama.cpp. - Agentic Coding: Specifically designed to power agentic coding assistants, such as the
smolcodeproject. - Efficient Fine-tuning: Utilizes bf16 LoRA on attention and MLP projections with assistant-only loss, focusing on tool calls and final answers.
- Context Length: Supports a 32768 token context window.
Training Details
The model was trained using the NousResearch/hermes-function-calling-v1 dataset for breadth and synthetic smolcode tool-use trajectories for sharpness. A critical aspect of its training involved rendering all data through the same apply_chat_template(tools=...) used at inference, ensuring byte-identical training targets. This v2 iteration specifically fixes a template mismatch issue present in v1, aiming for high free-generation tool-call parse-rates.
Use Cases
This model is ideal for developers building SLM-optimized agentic coding assistants that require robust and native function calling capabilities from a compact model.