⬡ ModelStudio IDE — Visual Architecture Inspector & Precision Model Tuner

The Sovereign Visual IDE for AI Models & GGUF Quantization Architecture Inspect 400+ GB model weights in <309ms with zero RAM bloat, tune layer precision, edit Jinja2 chat templates, scale LoRAs, and publish to HuggingFace or Ollama in 1 click.

• ⚡ Sub-309ms Zero-Copy Memory Reader: Native C# .NET 10 binary parser using MemoryMappedFile and Span<T>. Reads multi-hundred GB models in 309 milliseconds with ~40MB RAM footprint.
• 🌐 Universal Multi-Format Support: GGUF v2/v3, SafeTensors, ONNX, and TFLite support across Gemma, Llama, Qwen, DeepSeek, and Mistral model families.
• 📝 Interactive Header & Jinja2 Template Editor: Dedicated multi-line Jinja2 chat template inspector (tokenizer.chat_template), RoPE frequency tuner, and metadata key-value editor.
• 🧩 LoRA Adapter Workbench: Inspect rank ($r$), alpha ($\alpha$), and target modules with live interactive alpha scaling sliders ($0.0\times \rightarrow 2.0\times$) and 1-click model merging.
• ⚡ GGUF Precision Quantizer & Perplexity Evaluator: Convert tensor precision between IQ4_XS, Q4_K_M, Q6_K, Q8_0, and FP16 while calculating live perplexity SNR scores and Wikitext-2 quality grades.
• 💾 Hardware VRAM Estimator: Calculate exact GPU/TPU VRAM memory requirements for 4k, 32k, and 128k context windows.
• 🚀 1-Click Publishing: Direct 1-click deployment to Ollama (ollama create) and HuggingFace Hub repositories.
• 🔀 FrankenMerge SLERP Merger & Splitter: Slerp interpolation preview between Model A and Model B, plus multi-part GGUF file splitting.
• 🎨 Dynamic Studio Theme Engine: Sleek left navigation drawer sidebar with real-time themes (Cyberpunk Neon, Obsidian Dark, Emerald Studio, Clean Dark Slate).

• .NET 10 SDK

git clone https://github.com/ash-forge/model-studio-ide.git
cd model-studio-ide/ModelStudio.App
dotnet build -c Release

Built with ❤️ by ash-forge © 2026. Open source under the MIT License.