Skip to main content

​
Load → Inspect → Test

1

Load a model

  1. Click Load (01) in the sidebar
  2. Click LOAD GGUF FILE (or LOAD SAFETENSORS FILE / LOAD MODEL FOLDER)
  3. Select your model file in the dialog
  4. Wait for header parsing (< 1 second)
You’ll see model metadata: architecture, layers, parameters, quantization, and tensor map.
2

Inspect the architecture

  1. Click Inspect (02) in the sidebar
  2. Explore the isometric 3D visualization — hover layers for details
  3. Review memory distribution, quantization breakdown, and runtime compatibility
  4. Check capability detection (reasoning, code, math, etc.)
3

Run inference

  1. Click Test (09) in the sidebar
  2. Click USE LOADED to select your model
  3. Type a prompt or click a quick test preset (CODE, MATH, REASON, CREATIVE, INSTRUCT, CHAT)
  4. Configure GPU layers, temperature, and max tokens
  5. Click GENERATE and watch tokens stream in real time
For GGUF inference, llama.cpp tools must be installed. Go to Settings (07) → llama.cpp Tools → DOWNLOAD & INSTALL.

​
Download → Quantize

1

Download from HuggingFace

  1. Click Hub (04) in the sidebar
  2. Enter a repo ID (e.g., TheBloke/Mistral-7B-Instruct-v0.2-GGUF)
  3. Click FETCH, then DOWNLOAD on a GGUF file
2

Load and quantize

  1. In Library view, click LOAD on the downloaded model
  2. Go to Compress (03)
  3. Pick a quantization preset (MOBILE, BALANCED, or QUALITY)
  4. Click QUANTIZE MODEL and choose output path

​
Explore a Dataset

1

Load from HuggingFace

  1. Click DataStudio (10) in the sidebar
  2. Switch to HUGGINGFACE source
  3. Enter a dataset repo ID (e.g., tatsu-lab/alpaca)
  4. Click FETCH to see available files
  5. Click DOWNLOAD on a Parquet or JSON file
  6. Dataset auto-loads after download
2

Analyze

Review metadata (rows, columns, format), column analysis (dtypes, null counts), and preview data in the table view.

​
Fine-Tune a Model

1

Select model and dataset

  1. Click Training (06) in the sidebar
  2. Browse for a model (GGUF or SafeTensors)
  3. Browse for a dataset (JSON, JSONL, CSV, or Parquet)
2

Choose method and preset

  1. Select a training method (LoRA, QLoRA, SFT, DPO, Full)
  2. Pick a VRAM preset (LOW VRAM ~4GB, BALANCED ~6GB, QUALITY ~12GB, MAX QUALITY ~24GB)
  3. Optionally target specific capabilities (reasoning, code, math, etc.)
3

Train

Click START TRAINING, monitor real-time progress (loss, epoch, VRAM usage, ETA).

​
Merge Two Models

1

Load parents

  1. Click M-DNA (08) in the sidebar
  2. In the Files tab, click + LOAD FILE twice to load 2 models
2

Configure merge

  1. Go to Settings tab, pick a preset (QUICK BLEND, SMOOTH MERGE, TASK TUNER, etc.)
  2. Or manually select a method and tune parameters
  3. Set output format (SafeTensors or GGUF) and output path
3

Build

  1. Click BUILD MERGE
  2. Monitor progress in the sidebar and status bar
  3. Result shows output path and size