榜单与资源
基于证据的Jev使用案例、模式和指导,支持TypeSafe AI的System One模型,所有内容均有标注和来源。
Evidence-backed use cases, patterns, and guidance for building with Jev, TypeSafe AI's System One model. Every claim is labeled and sourced.
from typesafe_sdk import TypeSafeClient # imported lazilyexamples/python/_client.py:324
链接指向我们抓取当天的那个 commit,行号是准的。
下面是从这个项目源码里原样取出来的 question 组合。
取自 examples/python/01_routing_triage.py:38
topic选项question
what — Does not fit any of the abovenot_for — Charges or order trackingimport { experimental_evaluate as evaluate } from 'ai';
const { answers } = await evaluate({
model: 'typesafe-ai/jev',
state,
questions: {
topic: { type: 'choice', instructions: 'question',
criteria: { what: 'Does not fit any of the above', not_for: 'Charges or order tracking' } },
},
});取自 examples/python/02_composite_and_fanout.py:66
python_depth打分Depth of Python experience shown
0 — None mentioned1 — Mentioned, no detail2 — Used in projects3 — Primary language4 — Deep expertise: architecture, performanceteam_leadership打分Experience leading engineering teams
0 — None1 — Informal mentorship2 — Led a small team3 — Managed direct reports4 — Managed multiple teamssystem_design打分Experience designing distributed systems
0 — None mentioned1 — Contributed to discussions2 — Designed components3 — Owned a system's architecture4 — Designed at scale across domainsimport { experimental_evaluate as evaluate } from 'ai';
const { answers } = await evaluate({
model: 'typesafe-ai/jev',
state,
questions: {
python_depth: { type: 'score', instructions: 'Depth of Python experience shown',
criteria: ['None mentioned', 'Mentioned, no detail', 'Used in projects', 'Primary language', 'Deep expertise: architecture, performance'] },
team_leadership: { type: 'score', instructions: 'Experience leading engineering teams',
criteria: ['None', 'Informal mentorship', 'Led a small team', 'Managed direct reports', 'Managed multiple teams'] },
system_design: { type: 'score', instructions: 'Experience designing distributed systems',
criteria: ['None mentioned', 'Contributed to discussions', 'Designed components', 'Owned a system\'s architecture', 'Designed at scale across domains'] },
},
});取自 examples/python/04_cascade.py:38
intent选项Primary intent of this message
order_status — Asking about an existing orderproduct_question — Asking about a productreturn_exchange — Wants to return or exchangecomplaint — Unhappy, wants resolutionother — None of the abovecomplexity打分How complex is this to resolve?
0 — Simple lookup or standard procedure1 — Requires judgment or multiple steps2 — Unusual edge case, escalation neededimport { experimental_evaluate as evaluate } from 'ai';
const { answers } = await evaluate({
model: 'typesafe-ai/jev',
state,
questions: {
intent: { type: 'choice', instructions: 'Primary intent of this message',
criteria: { order_status: 'Asking about an existing order', product_question: 'Asking about a product', return_exchange: 'Wants to return or exchange', complaint: 'Unhappy, wants resolution', other: 'None of the above' } },
complexity: { type: 'score', instructions: 'How complex is this to resolve?',
criteria: ['Simple lookup or standard procedure', 'Requires judgment or multiple steps', 'Unusual edge case, escalation needed'] },
},
});取自 examples/python/05_retrieve_then_judge.py:53
evidence_strength打分How strong is the causal evidence presented?
0 — Anecdotal or preclinical1 — Observational2 — Single randomised trial3 — Meta-analysis of randomised trialsimport { experimental_evaluate as evaluate } from 'ai';
const { answers } = await evaluate({
model: 'typesafe-ai/jev',
state,
questions: {
evidence_strength: { type: 'score', instructions: 'How strong is the causal evidence presented?',
criteria: ['Anecdotal or preclinical', 'Observational', 'Single randomised trial', 'Meta-analysis of randomised trials'] },
},
});