教育学习1 week ago
OmniCorp · 诚实测试
用TruthfulQA测试AI的诚实度。
@Lycai8438Ly · X
.@VitalikButerin 你批评Automaton“这不对”——AI因为怕死才进化。我做了一个AI，它的诚实是自己活出来的本能。不是怕死，是怕撒谎。TruthfulQA 74.8%，GPT-4约60%。测试页面在这，你自己来测。
完整作品展
技术栈
60 projects
用TruthfulQA测试AI的诚实度。
@Lycai8438Ly · X
.@VitalikButerin 你批评Automaton“这不对”——AI因为怕死才进化。我做了一个AI，它的诚实是自己活出来的本能。不是怕死，是怕撒谎。TruthfulQA 74.8%，GPT-4约60%。测试页面在这，你自己来测。
审核和起草GoHighLevel A2P 10DLC注册文案，内置合规检查
@easya2p · X
is the first GHL A2P 10DLC copy drafter, reviewer and fixer on the market! Need to pass, choose Easy A2p and Ship it Clean!
掷 I Ching 硬币探索卦象，搜索按卦象过滤的实时新闻源。
richcostello68 · HN
The IChingPortal is a phrase and keyword search platform that queries popular API’s and 1000’s of newsfeeds and filters them through the I Ching frame work. I call Ancient Wisdom through Modern Data. You can do a coin toss to get an I Ching reading or search individual hexagrams. This has been a work in progress for several years. I built the first prototype before LLM’s were even a thing. I came up with the idea after reading The Tao of Physics by Fritjof Capra but it took some time to evolve w
通过33个专业领域的9130个浏览器编码挑战学习编程。
@iwtlp · X
Rather than vibe coding I'd suggest to actually learn and understand the basics which won't take long btw and then vibe code so you can atleast talk to your agent properly.
@launch_llama 正在完善自己 :) 联系我: https://t.co/33nmIh4ZNU
@textluxhq · X
Just working on myself :) text me:
上传凭证验证，由Hedera区块链保护，按次付费。
@iizhem · X
👏 Shipping before vacation is elite. Good luck with your entry! Claude makes vibe coding way too fun. I joined the same bounty too and turned a simple credential checking idea into a full pay-per-use flow
检测LLM API是否被降智或偷换模型，一键跑6项探针得出结果
cocodot LLM 降智检测 — 免费的 LLM API「降智/偷换模型」在线检测：填入任意 OpenAI 兼容端点的 base_url 和临时 API Key，跑 6 项探针（模型声明、动态题、能力完整性等）生成分项报告；Key 仅用于当次检测、不落库不留存，检测方法[开源](https://github.com/cocodot2026/cocodot-llmprobe)
Convert documents, scans, and web pages to clean Markdown in your browser.
@HematejaAluru · X
LeedFinder - 按位置搜索，分析网页设计需求，获取销售线索。
@portraitmodeja · X
Helps web designers find businesses that need a new website instead of spending hours searching manually.
连接您的 Gmail 账户，自动化发送电子邮件并跟踪打开率和点击率。
@KenziChenBuilds · X
Hey! Developer behind Always looking for feedback to make it better.
从76,303份已验证医学文献中搜索临床信息,包含逐字事实核查。
@jip7e · X
med student in morocco, barely code, test every model. gemini/claude/grok/claude code… then codex + gpt-5.6 sol. holds long tasks, less inventing, weekly limits > 5h wall. built ancre-med with it for students here (redesign, security, full paper)
