Tesslate-UIGEN-T2-7B-GGUF

Tesslate/UIGEN-T2-7B (GGUF Q4) - Sandlogic Lexicons

Model Overview

UIGEN-T2-7B is the latest innovation in the UIGEN model series by Tesslate, engineered specifically for high-quality, design-aware user interface (UI) code generation. Built on the strong foundation of the Qwen2.5-Coder-7B-Instruct model and fine-tuned using PEFT/LoRA (Rank 128), UIGEN-T2 has been trained on a significantly larger dataset of 50,000 annotated UI samples. This enables it to generate not only functional HTML and Tailwind CSS code but also code that aligns with usability, design structure, and modern layout principles.

Model Highlights

• Architecture: Based on Qwen2.5-Coder-7B-Instruct (decoder-only transformer)
• Quantization Format: GGUF Q4
• LoRA Fine-Tuning: PEFT/LoRA (Rank 128), with checkpoints published at each training stage
• Training Dataset: 50,000 high-quality UI examples (up from 400 in the previous version)
• Code Output: Semantic HTML and utility-first Tailwind CSS
• UI-Based Reasoning: Incorporates guidance from a reasoning teacher model to ensure usability and aesthetics
• Chat Interface: Improved prompt interaction for seamless developer experience

Intended Use Cases

Recommended Applications

• Rapid UI Prototyping
  Generate clean, production-ready HTML/Tailwind code from natural language descriptions or wireframes.

• Component Generation
  Create both standard (buttons, forms, cards) and custom components based on user-defined design goals.

• Frontend Development Assistant
  Accelerate development by producing baseline layouts and component structures for web interfaces.

• Design-to-Code Exploration
  Bridge the gap between visual design and code implementation, especially helpful for non-developers or design teams.

Model Integration

This quantized Q4 GGUF version of UIGEN-T2-7B is now part of the Sandlogic Lexicons model zoo. It is optimized for efficient inference on edge and constrained environments, without compromising on code quality or reasoning capability.