What is Lumichat Coder v2.1?

Lumichat Coder v2.1, developed by Aditya Kumar Jha / LumiChats, is a 1.5 billion parameter language model built upon the Qwen2.5-Coder-1.5B-Instruct foundation. It is specifically fine-tuned for precision function calling and tool use, converting natural language queries into executable JSON. The model boasts a 32K token context window and achieves high accuracy, with 96.5% for tool selection and 99.8% for JSON validity, thanks to transformers-CFG for grammar-constrained generation.

Key Capabilities

• Precision Tool Calling: Generates accurate, structured JSON for function execution.
• Lightning Fast & Efficient: Optimized with Unsloth for 2x faster inference and 60% less memory, enabling deployment on consumer GPUs.
• Grammar-Constrained Output: Guarantees 99.8% valid JSON output, eliminating parsing errors.
• Context-Aware: Maintains strong reasoning across a 32K token context window for complex multi-turn conversations.
• Multi-Tool Orchestration: Handles complex queries requiring multiple sequential or parallel function calls.

When to Use This Model

• Building AI Agents: Ideal for creating autonomous systems that interact with external tools and APIs.
• Workflow Automation: Automating tasks by translating natural language into structured operations.
• API Integration: Generating properly formatted API calls from user queries.
• Conversational Interfaces: Enhancing chatbots with the ability to perform actions via tool calls.
• Developer Tools: Integrating intelligent code generation for specific functions within development workflows.