SmolLM2 1.7B — ultra-lightweight model for extreme edge deployment. Runs on phones, Raspberry Pi, or any device with 2GB RAM. Privacy without the hardware requirements.

Privacy Guarantees

Feature	Status
Sandboxed inference (no network egress)	✅ Yes
PII output guardrails (email, SSN, CC, etc.)	✅ Yes
Encrypted inference logs	✅ Yes
Zero telemetry	✅ Yes
Differential privacy (DP-SGD)	❌ Wrapped only

Quick Start

bash

pip install ltce
ltce pull lattice-ai/smollm2-1.7b-private
ltce serve lattice-ai/smollm2-1.7b-private --port 8080 --sandbox

python

from ltce import Lattice

lt = Lattice()
lt.serve("lattice-ai/smollm2-1.7b-private", port=8080, sandbox=True)

The model serves an OpenAI-compatible API at http://localhost:8080/v1/chat/completions.

What Does "Wrapped" Mean?

This model uses the original HuggingFaceTB/SmolLM2-1.7B-Instruct weights — no additional training has been done.

What Lattice adds is the privacy serve layer:

🔒 Sandbox mode — all outbound network connections are blocked during inference
🛡️ PII guardrails — model output is scanned for emails, phone numbers, SSNs, credit card numbers, API keys
📝 Encrypted logs — inference logs are AES-256-GCM encrypted at rest
🚫 Zero telemetry — HuggingFace telemetry, W&B, MLflow all disabled

For models with mathematical privacy guarantees (DP-SGD training), see our hardened tier models.

Capabilities

general, instruct

Base Model

HuggingFaceTB/SmolLM2-1.7B-Instruct

License

Apache 2.0

Built with Lattice — privacy-first local AI.