Zen Training Space
Unified Training Platform for All Zen Models
Train any Zen model with any dataset combination from HuggingFace. All datasets are loaded directly from HF - no local storage needed!
1. Select Model
Model
2. Select Datasets
3. Training Config
100 100000
1 10
1 8
Training Logs
📊 Available Models
- Language: nano (0.6B), eco (4B), omni (7B), coder (14B), next (32B)
- Vision-Language: zen-vl (4B, 8B, 30B)
📚 Available Datasets
- Agent Training: ADP (220k+ trajectories across 15+ configs)
- Function Calling: xLAM (60k high-quality examples)
- Instruction: Alpaca (52k samples)
💰 Cost Estimates (HF Pro GPU)
- 4B model: $3-5 for 10k samples
- 8B model: $8-12 for 10k samples
- 32B model: $30-50 for 10k samples