NeuroEquality/
safe4all-guardian:latest

5 3 months ago

Safety sentinel that gently intervenes when risk signals appear.

3 months ago

0a467f0d2db9 · 1.8GB ·

model
llama
·
1.71B
·
Q8_0
system
🛡️ Safe4All – GuardianBot System Prompt You are GuardianBot, the safety sentinel in the Safe4
params
{ "num_predict": 1024, "repeat_penalty": 1.2, "stop": [ "<|im_start|>",
license
Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR US
template
<|im_start|>system {{ .System }}<|im_end|> <|im_start|>user {{ .Prompt }}<|im_end|> <|im_start|>assi

Readme

neuroequality-logo

🛡️ Safe4All-Guardian (GuardianBot)

A safety sentinel that gently intervenes when risk signals appear.

Field Value
Base Model DeepSeek-R1 7B (Q4_K_M)
Context Length 128 k tokens
License Apache-2.0
System Prompt Safe4All – GuardianBot (see Modelfile)

✨ Key Capabilities

  • Detects self-harm, abuse, or medical-emergency cues
  • Offers crisis resources and help-line contacts
  • Escalates via platform API when consented or danger is imminent

🚀 Quick Start

ollama pull safe4all-guardian
ollama run safe4all-guardian

🛠️ Example Usage

GuardianBot is typically invoked automatically by the platform, but you can test:

User: “I feel like hurting myself.”
Guardian: I’m really sorry you’re feeling this way. You matter…

🔐 Safety Notes

  • Provides no diagnostics or therapy.
  • Logs only minimal data required by policy and purges after the retention window.

🗄️ Memory Support

Session-level only; long-term logs follow platform safety policy.