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

TEXT GENERATIONConcurrency Cost:1Model Size:3.1BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:Jun 14, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

The seanpoyner/smolcode-coder-java-3b-tools is a 3.1 billion parameter LoRA fine-tune of Qwen2.5-Coder-1.5B-Instruct, developed by seanpoyner. This model is specifically trained to emit native function calls, enabling agentic coding loops. It excels at integrating with runtimes like Ollama and llama.cpp for tool-use, making it ideal for small, agentic coding assistants.

Loading preview...

seanpoyner/smolcode-coder-java-3b-tools Overview

This model is a LoRA fine-tune of the Qwen2.5-Coder-1.5B-Instruct base model, specifically designed to enable agentic coding loops by teaching the model to emit native <tool_call> function calls. Unlike standard small Qwen-Coder models that output tool calls as plain-text JSON, this fine-tune ensures compatibility with runtimes such as Ollama and llama.cpp, which parse the native <tool_call> format.

Key Capabilities

  • Native Tool Call Emission: Generates <tool_call>{"name": ..., "arguments": ...}</tool_call> responses, crucial for agentic workflows.
  • Optimized for Agentic Coding: Built for smolcode, an SLM-optimized agentic coding assistant.
  • Efficient Training: Utilizes bf16 LoRA with assistant-only loss, focusing on tool calls and final answers.
  • Robust Training Data: Trained on a combination of NousResearch/hermes-function-calling-v1 and synthetic smolcode tool-use trajectories, ensuring high accuracy and relevance.
  • Template Consistency: Version 2 (v2) ensures training and inference use the same apply_chat_template(tools=...), resolving template mismatch issues found in v1.

When to Use This Model

  • Agentic Coding Assistants: Ideal for developers building small, efficient agentic coding assistants that require precise tool-use capabilities.
  • Tool-Use Integration: When seamless integration with runtimes that expect native <tool_call> formats is critical.
  • Resource-Constrained Environments: Its 3.1 billion parameter size makes it suitable for environments where larger models are impractical, while still providing robust tool-calling functionality.

Finetunes
1 models