Skip to content
System One

using76/TypeEvacSafe

A fire evacuation simulation that rebuilds a Jev-style typed decision engine on local LLMs, then runs a per-person evacuation model on top, comparing it against a chat-model baseline on the same fire scenario.

Bonsai 2 27B vs TypeSafe Jev
Image from github.com

using76/TypeEvacSafe on GitHub

Stars
2
Forks
2
Language
Python
Last push
19 Sep 2026

Read from GitHub on . Counts change daily.

How it uses Jev

"""TypeSafe Jev API(jev-ultrafast model.py 의 호출 형식): 한 state 에 여러 choice 질문을 한 요청으로.
키는 환경변수 TYPESAFE_API_KEY 또는 ~/.typesafe_key 파일 — 코드·저장소에 넣지 않는다."""

def __init__(self, model="jev-latest", url="https://api.typesafe.ai/v1/systemone", threads=4):
    import os
    key = os.environ.get("TYPESAFE_API_KEY")
    if not key:
        kp = os.path.expanduser("~/.typesafe_key")
        key = open(kp).read().strip() if os.path.exists(kp) else None
    if not key: raise RuntimeError("TYPESAFE_API_KEY 없음")
    self.key, self.model, self.url, self.threads = key, model, url, threads
    self.stats = dict(calls=0, seconds=0.0, usage=[])

def ask(self, ev, qs, lang="en"):
    """qs: {head: (question, [(id, desc)])} → {head: {id: prob}}"""

View in typeevac/jev_decide.py

Open on GitHub

More like this

Use cases this is tagged with