🚀 SLMGEN - Small Language Model Generator

Fine-tune SLMs. 2x faster. For free.

Live Demo · API Docs · User Guide

---

✨ What is SLMGEN?

SLMGEN is a web application that automates SLM fine-tuning. Upload your JSONL dataset and receive ready-to-run Google Colab notebooks with Unsloth + LoRA optimization.

Your Data → Best Model → Matched. One notebook. Zero setup. Ready to train.

---

🎯 Core Features (V3.0.0)

Feature	Description
📤 Smart Upload	Drag-and-drop JSONL with Live Chat Preview (min 50 examples)
📊 Quality Scoring	Duplicate detection, consistency checks, 0-100% quality score
🧠 18 Model Support	Qwen 3.5, Llama 3.3, DeepSeek V3, Phi-4, Gemma 3, SmolLM3 + more
🎯 100-Point Matching	Task fit (50pts) + Deploy target (30pts) + Data traits (20pts)
💻 Training Simulator	Real-time terminal simulation during generation phase
📓 Self-Contained Notebooks	Dataset embedded as base64 - no file uploads needed
🔄 Dataset Converter	CSV, TSV, JSON, Alpaca, ShareGPT → ChatML
⚡ Training Presets	Quick Demo, Production, Edge, Code, Long Context
📦 Export Options	Ollama, GGUF, vLLM, HuggingFace

---

🧠 Advanced Intelligence Features

Dataset Intelligence Layer

• Personality Detection - Infers tone, verbosity, technicality, strictness
• Hallucination Risk - Scores likelihood of model fabrication (0-1)
• Confidence Score - Measures training reliability via coverage/diversity

Prompt & Behavior Engine

• Behavior Composer - Generate system prompts from trait sliders
• Prompt Linter - Detects contradictions, redundancy, ambiguity
• Prompt Diff - Semantic comparison between prompts

Model Transparency

• "Why This Model?" - Strength/weakness deep dive per model
• Failure Previews - Synthetic failure cases before training
• Model Card Generator - Auto-generated deployment README

---

🛠️ Tech Stack

Component	Technology
Backend	Python 3.11, FastAPI, Pydantic v2
Session Store	Redis (async via redis-py)
Frontend	Next.js 16, TypeScript, React 19, Framer Motion
Design	Tailwind CSS, JetBrains Mono, Everblush Theme
Auth	Supabase (OAuth + Email)
Training	Unsloth + LoRA on Google Colab (Free T4)
Deployment	Vercel (Frontend) + Render (Backend)

---

🚀 Quick Start

Prerequisites

• Python 3.11+
• Node.js 18+
• Redis 7+ (for session storage)
• Supabase project

Backend

# Start Redis (if not already running)
redis-server &

cd libslmgen
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env  # Configure Redis URL & Supabase keys
uvicorn app.main:app --reload --port 8000

Frontend

cd slmgenui
npm install
cp .env.example .env.local  # Configure API URL + Supabase
npm run dev

Open http://localhost:3000 🎉

---

📁 Project Structure

slmgen/
├── libslmgen/                  # Python Backend
│   ├── app/
│   │   ├── main.py             # FastAPI app
│   │   ├── routers/            # API endpoints
│   │   │   ├── upload.py       # Dataset upload
│   │   │   ├── analyze.py      # Dataset analysis
│   │   │   ├── recommend.py    # Model recommendation
│   │   │   ├── generate.py     # Notebook generation
│   │   │   ├── advanced.py     # Intelligence features
│   │   │   └── jobs.py         # Job history
│   │   └── session.py          # Thread-safe sessions
│   └── core/
│       ├── ingest.py           # JSONL parsing
│       ├── quality.py          # Quality scoring
│       ├── analyzer.py         # Dataset analysis
│       ├── recommender.py      # 100-point scoring engine
│       ├── notebook.py         # Jupyter generator
│       ├── personality.py      # Personality detection
│       ├── risk.py             # Hallucination risk
│       ├── confidence.py       # Training confidence
│       ├── behavior.py         # Behavior composer
│       ├── prompt_linter.py    # Prompt linting
│       └── model_card.py       # README generator
├── slmgenui/                   # Next.js Frontend
│   └── src/
│       ├── app/                # Pages (dashboard, login, signup)
│       ├── components/         # UI components
│       ├── lib/                # API client & types
│       └── hooks/              # React hooks (with persistence)
├── docs/
│   ├── API.md                  # API reference
│   ├── USER_GUIDE.md           # User guide
│   └── DEPLOY.md               # Deployment guide
└── supabase/
    └── schema.sql              # Database schema

---

📊 Supported Models

Model	Size	Context	Best For	Gated
Phi-4 Mini	3.8B	16K	Classification, Extraction	❌
Llama 3.2	1B/3B	8K	Q&A, Conversations	✅
Gemma 2	2B	8K	Edge, Mobile	✅
Qwen 2.5	0.5B-3B	32K	Multilingual, JSON	❌
Mistral 7B	7B	32K	Generation, Creative	❌
TinyLlama	1.1B	2K	Ultra-lightweight	❌
SmolLM2	135M-1.7B	8K	Small devices	❌

---

📦 Dataset Format

Each line in your JSONL file should be a conversation:

{"messages": [{"role": "user", "content": "Hello!"}, {"role": "assistant", "content": "Hi there!"}]}
{"messages": [{"role": "system", "content": "You are helpful."}, {"role": "user", "content": "..."}, {"role": "assistant", "content": "..."}]}

Requirements:

• ✅ Minimum 50 examples
• ✅ At least one user and one assistant message
• ✅ UTF-8 encoding
• ✅ Valid JSON per line

---

🌐 Deployment

Vercel (Frontend)

npx vercel --prod

Render (Backend)

Uses render.yaml blueprint for auto-deployment.

See DEPLOY.md for full instructions.

---

⚙️ Environment Variables

# Backend (.env)
ALLOWED_ORIGINS=https://slmgen.vercel.app,http://localhost:3000
UPLOAD_DIR=/tmp/uploads
REDIS_URL=redis://localhost:6379/0
SESSION_TTL_SECONDS=1800
SUPABASE_URL=your_supabase_url
SUPABASE_ANON_KEY=your_anon_key
SUPABASE_SERVICE_ROLE_KEY=your_service_key
JWT_SECRET=your_jwt_secret

# Frontend (.env.local)
NEXT_PUBLIC_API_URL=https://slmgen-api.onrender.com
NEXT_PUBLIC_SUPABASE_URL=your_supabase_url
NEXT_PUBLIC_SUPABASE_ANON_KEY=your_anon_key

---

📄 License

MIT License - See LICENSE

---

👥 Authors

Vedant Singh Rajput

• 🐙 @vedanthq

Eshan Roy

• 📧 eshanized@proton.me
• 🐙 @eshanized

---

⭐ Star this repo if SLMGEN helped you fine-tune faster!