seanpoyner/smolcode-coder-docker-3b-tools

TEXT GENERATIONConcurrent Unit Cost:1Model Size:3.1BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jun 14, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The seanpoyner/smolcode-coder-docker-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 facilitates agentic coding loops by correctly formatting tool calls for runtimes like Ollama and llama.cpp, addressing a common limitation in small coder models. It is optimized for tool-use scenarios within agentic coding assistants, particularly for the smolcode project. The model's training focused on assistant-only loss using a combination of NousResearch/hermes-function-calling-v1 and synthetic smolcode tool-use trajectories.

Loading preview...

Overview

This model, seanpoyner/smolcode-coder-docker-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. Unlike standard small Qwen-Coder models that often describe tool calls as plain-text JSON, this fine-tune ensures the output is in the format expected by runtimes like Ollama and llama.cpp.

Key Capabilities

  • Native Tool Call Emission: Generates <tool_call>{"name": ..., "arguments": ...}</tool_call> directly, integrating seamlessly with agentic systems.
  • Agentic Coding Support: Specifically built to power smolcode, an SLM-optimized agentic coding assistant.
  • Efficient Fine-tuning: Utilizes bf16 LoRA with assistant-only loss, focusing training on tool calls and final answers.
  • Template Consistency: Trained and served using a shared apply_chat_template(tools=...) to ensure byte-identical training targets and inference prompts, resolving issues found in earlier versions.

Good For

  • Developers building agentic coding assistants that require precise tool-use capabilities from small models.
  • Use cases where efficient and correctly formatted function calling is critical for automated code generation or interaction with external tools.
  • Environments where a compact, specialized coder model is preferred over larger, more general-purpose LLMs for tool-use tasks.

Finetunes
1 models