model-verification

Your users download a 4GB AI model, the connection drops at 3.8GB, verifyfetch resumes from 3.8GB and verifies every byte. Drop-in integrity verification for Transformers.js, WebLLM, and any large file in the browser.

AI中转站质量监控 · 模型验真 · Token成本优化 — RelayRadar 中转雷达

Detect fake AI APIs — Verify if an LLM API is actually serving the model it claims. Catch resellers selling "Claude" or "ChatGPT" that secretly serve cheaper models. Behavioral fingerprinting with 32 forensic prompts.

🔍 一键检测 AI API 服务商是否如实提供所标注的模型 | 支持 OpenAI / Claude / Gemini / DeepSeek 等 9 大模型原厂对比验证

Abstract Constraint Transformation

Verify model identity and detect downgrades or swaps through fingerprint comparison.

A modular, plugin-based framework for verifying the true identity of models behind LLM APIs. / 模块化的 LLM 指纹识别与 API 模型身份验证框架。

Which model is really behind your API relay or agent IDE? Behavioral fingerprinting + anytime-valid sequential tests (FPR<=1%). LLMs can't be random - measured on 9 frontier models.

Formal verification of UML state machines with Promela/spin and TLA+/PlusCal

OpenAI-compatible AI API integration guide for Claude, GPT, Gemini, Codex, Cursor, Dify and production troubleshooting

LLM API proxy China / OpenAI-compatible API gateway checks: model quality, SSE, tool calls, routing and production troubleshooting | AI API 中转站检测

Claude Code skill + verifier for running Fable 5 background subagents: avoid silent Opus fallback, verify the real model, and structure slow hard work.

The project builds a Bayesian Network using pressure sensor data to detect pipe leaks, employing probabilistic reasoning to determine the likelihood of leaks based on sensor readings. It involves loading dataset, defining network structure, calculating CPDs, adding them to model, and using Variable Elimination algorithm for inference

Vendor-agnostic CLI to verify & benchmark LLM APIs — detect model downgrade/swap on any OpenAI-compatible endpoint. Trust no one, verify.

One of the biggest barriers to widespread machine learning adoption is the difficulty in collecting a 'good' dataset. There is an overall consensus that a 'good' dataset is a big dataset, but we believe that we can do better. As such the VennData project was created to develop tools to guide in the collection, curation, augmentation and validati…

Reversible coordinate permutations for Llama checkpoints with exact Q16.48 validation and direct SafeTensors support.

Visual anatomy of a Q16.48 integer GPT — four self-contained offline HTML pages: weight atlas, bit-exact inference trace, 32-D embedding space, 3-D weight explorer. Every displayed number is exact (BigInt in-browser, verified against the C oracle). Fourth in the int-llm series.

检测中转站的 DeepSeek 是满血还是野鸡：用 DeepSeek Harness minimal 预设的首轮请求面做思维链行为指纹。支持 OpenAI / Anthropic / Responses 三种协议，零依赖，带网页界面。Detect downgraded or rebadged DeepSeek models behind API relays.

AI 中转站 API 验真取证工具：模型验真 · 系统提示词提取 · 上下游供应链溯源 | LLM relay API forensics: verify models, extract system prompts, trace upstream supply chains

Verify your Claude Code endpoint really serves GLM-5.2 — tokenizer fingerprint + 1M context probe