MinT (Mind Lab Toolkit) is an RL infrastructure that helps agents and models learn from real experience. It abstracts away compute scheduling, distributed rollout, and training orchestration so teams iterate inside real tasks.
MinT provides a unified, reproducible way to run RL across models and tasks, focused on making LoRA RL simple, stable, and efficient. You define what to train, what data, how to optimize, and how to evaluate — MinT handles the rest.
User Control
Infrastructure Complexity
Supported models
DeepSeek
Moonshot
GLM
MiniMax
VLA
Editions
Security & compliance
Privacy & security
Secure by default. Enterprise data is not used to train shared foundation models unless expressly authorized.
- Sensitive-field detection and redaction before training
- Encrypted transfer, encrypted storage, and least-privilege access
- Audit logs, continuous monitoring, and review-ready controls
Data ownership
Your Enterprise Data remains owned and controlled by you, including datasets, outputs, LoRA weights, and reports.
- Tenant-isolated cloud workspaces
- Dedicated cloud, VPC, private, localized, or hybrid deployment
- No commercial use without prior express authorization
Compliance
Built for regulated teams with MLPS Level 3 support and domestic data/privacy audit readiness.
- China MLPS Level 3 compliance support
- Domestic data and privacy obligation audit support
- Multiple-chip compatibility for localized training
Cases
Medical Coding
Medical coding and case-record post-training for quality-control workflows across clinical scenarios.
Personalized Agent
Personal agent infrastructure where many 1T model instances share one base while each user keeps LoRA differences.
Smart Customer Service
LoRA post-training for a leading fintech support workflow with compliance, takeover, and cost improvements.
Citizen Services
Policy-aware service agents for public cloud scenarios with fast policy refresh and measurable transfer reduction.