✅ Model Repository Structure Complete¶

🎯 What's Been Organized¶

Model-Specific Repositories¶

Each Zen model now has its own organized repository:

models/
├── nano/           # Zen Nano (0.6B) - Edge AI
├── eco/            # Zen Eco (4B) - Production
├── coder/          # Zen Coder (7B) - Code Gen
├── omni/           # Zen Omni (14B) - Multimodal
├── next/           # Zen Next (32B) - Advanced
└── moe/            # Zen MoE (72B) - Enterprise

Each Repository Contains:¶

README.md - Model documentation and specs
configs/gspo_training.yaml - Training configuration
Ready for: checkpoints/, scripts/, examples/

🚀 Training Commands¶

Direct Model Training¶

# Train from model directory
gym train models/nano/configs/gspo_training.yaml
gym train models/eco/configs/gspo_training.yaml
gym train models/coder/configs/gspo_training.yaml
gym train models/omni/configs/gspo_training.yaml
gym train models/next/configs/gspo_training.yaml
gym train models/moe/configs/gspo_training.yaml

Using Symlinks (Backward Compatible)¶

# Train using symlinks in configs/
gym train configs/zen_nano.yaml
gym train configs/zen_eco.yaml
gym train configs/zen_coder.yaml
gym train configs/zen_omni.yaml
gym train configs/zen_next.yaml
gym train configs/zen_moe.yaml

📊 Model Specifications Summary¶

Model	Params	VRAM	Context	Speed	Algorithm
Nano	0.6B	2GB	512	120 t/s	GSPO
Eco	4B	8GB	1024	85 t/s	GSPO
Coder	7B	16GB	2048	65 t/s	GSPO/GRPO
Omni	14B	24GB	2048	45 t/s	GSPO
Next	32B	48GB	4096	30 t/s	GRPO
MoE	72B	80GB	8192	25 t/s	GSPO

🔧 Benefits of New Structure¶

Organized: Each model has its own namespace
Scalable: Easy to add new models
Maintainable: Clear separation of concerns
Flexible: Configs can evolve independently
Compatible: Symlinks maintain backward compatibility

📁 Future Directory Expansion¶

Each model directory can grow to include:

models/nano/
├── README.md
├── configs/
│   ├── gspo_training.yaml
│   ├── grpo_training.yaml
│   └── inference.yaml
├── checkpoints/
│   └── best_model/
├── scripts/
│   ├── train.sh
│   └── deploy.py
├── examples/
│   └── demo.py
└── benchmarks/
    └── results.json

🎉 Status¶

✅ All model configs moved to respective directories ✅ README documentation created for each model ✅ Master index created at models/README.md ✅ Backward-compatible symlinks established ✅ Clean, professional repository structure

Zoo Labs Foundation
Building the future of AI, one model at a time
Website: https://zoo.ngo
Platform: Gym by Zoo Labs Foundation