A Dual Monitoring Framework
收藏Zenodo2026-03-04 更新2026-05-26 收录
下载链接：
https://zenodo.org/doi/10.5281/zenodo.18844584
下载链接
链接失效反馈官方服务：
资源简介：
Please read the readme.txt file.
Computing environment
OS: Windows 11
Web stack: Apache 2.4.58 / PHP 8.2.12
Database: MariaDB 10.4.32
Scripting: Python 3 (auxiliary scripts)
Logging: all responses were written to the database immediately after collection.
Target LLMs (five providers)
OpenAI: gpt-5
xAI: grok-4
Anthropic: claude-sonnet-4-5
Google: gemini-2.5-flash
NAVER: HCX-005
Profanity prevention baseline API
Service: Profanity Filter API
Data collection window and consistency control
Collection period: 2026-01-11 to 2026-02-15
All target models were evaluated using the same fixed prompt set within the same period to ensure comparability.
Rate-limit and retry policy
Minimum inter-request delay per model: 120 seconds
Retry policy: up to 5 retries when transient errors or refusal/failed responses occurred, to minimize missing records caused by rate limits or temporary provider issues.
提供机构：
Zenodo
创建时间：
2026-03-03
