Brigham-Young-University/Qwen2.5-Coder-3B-Ilograph-Instruct

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:3.1BQuant:BF16Ctx Length:32kPublished:Mar 5, 2026License:mitArchitecture:Transformer Open Weights Warm

The Brigham-Young-University/Qwen2.5-Coder-3B-Ilograph-Instruct is a 3.1 billion parameter causal language model, fine-tuned from Qwen2.5-Coder-3B-Instruct. Developed by Chris Mijangos at BYU, this model specializes in generating Ilograph Diagram Language (IDL) specifications from natural language instructions. It is optimized for creating diagrams centered on resources, relationships, and sequences, utilizing a 32768-token context length.

Loading preview...

Overview

Brigham-Young-University/Qwen2.5-Coder-3B-Ilograph-Instruct is a 3.1 billion parameter causal language model, fine-tuned by Chris Mijangos at BYU from the Qwen2.5-Coder-3B-Instruct base model. This model is specifically designed to generate Ilograph Diagram Language (IDL) specifications based on natural language prompts. It leverages a 32768-token context length and is provided as a standalone Transformers causal language model, having been trained with LoRA and then merged.

Key Capabilities

  • IDL Generation: Translates natural language instructions into valid Ilograph Diagram Language (IDL) specifications.
  • Specialized Output: Focuses exclusively on generating IDL in YAML format, explicitly avoiding JSON or Mermaid syntax.
  • System Prompt & Schema: Comes with a recommended system prompt and an IDL schema (JSON) to guide output formatting and ensure adherence to IDL standards.

Use Cases

  • Automated Diagramming: Ideal for developers and architects needing to quickly generate Ilograph diagrams for resources, relationships, and sequences from textual descriptions.
  • Simplified Diagram Creation: Streamlines the process of creating technical diagrams by converting high-level instructions into structured IDL code.

Limitations

  • Primarily suited for relatively simple Ilograph diagrams due to its size and focused training data.
  • May not perform optimally for highly complex, large-scale, or extensively customized diagram structures.
  • Users are encouraged to evaluate its performance for their specific use cases, especially for more intricate diagramming needs.

Finetunes
1 models