For employers

Hire this AI Trainer

Sign in or create an account to invite AI Trainers to your job.

Invite to Job
晴天夏

晴天夏

多模型实战使用者 / 中文 AI 内容质量评估

Japan flagtokey, Japan

Key Skills

Software

No software listed

Top Subject Matter

LLM 输出评估与错误诊断 / 训练数据质检
多模型横向评测与标注规范整理
AI 生图与多模态数据标注

Top Data Types

TextText
VideoVideo
ImageImage

Top Task Types

Question AnsweringQuestion Answering
Object DetectionObject Detection
Evaluation/RatingEvaluation/Rating
ClassificationClassification
Bounding BoxBounding Box
PolygonPolygon

Freelancer Overview

MyBrain 个人 AI 知识库(对话沉淀与模型质量评估记录). Brings 3+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Internal, Proprietary Tooling, and Don't disclose. Education includes Bachelor of Arts, N/A (2016). AI-training focus includes data types such as Text and labeling workflows including Evaluation, Rating, and Prompt + Response Writing (SFT).

Labeling Experience

多 AI 协同会诊工作流(用于形成结构化判断依据)

Don't discloseTextTextPrompt + Response Writing (SFT)Prompt + Response Writing (SFT)

设计并实践多模型方向会诊机制,通过同一问题在不同 LLM 之间的分歧对照来形成结构化判断与可复用讨论材料。使用共享知识库作为异步讨论介质,将多模型协作过程中的偏见与互相附和风险进行持续记录。该过程产出可用于后续训练数据构建的参考标签与评估依据。• 采用“Codex 主持+Claude/Gemini/Grok 列席”的多模型协同流程 • 横向比较不同模型在同一任务上的判断差异 • 记录“协同偏见”“同生态模型互相附和”等风险观察 • 以共享知识库固化结论与讨论过程以便后续复用

2026 - Present

MyBrain 个人 AI 知识库(对话沉淀与模型质量评估记录)

TextText

基于个人本地 LLM-Wiki 工作流,对模型输出错误与质量进行系统性复盘与记录,沉淀可复用的诊断知识。主要产出的是对幻觉、事实错误、逻辑跳步等问题的判断口径与案例化反馈。通过持续维护知识库与方法论,使其可用于后续训练数据质检或标注规范化。• 记录 AI 错误诊断案例并进行根因复盘与归因总结 • 建立对模型输出质量的量化/可执行评估维度 • 沉淀“raw 模式”“信息无损传递”“探索与逃避的分界”等方法论 • 用共享知识体系支持多模型协作判断偏差的识别

2026 - Present

Education

N

N/A贵州大学明德学院

Bachelor of Arts, Digital Media Technology

Bachelor of Arts
2016 - 2023

Work History

F

Freelance

Independent Multi-Model Evaluation Workflow Designer

N/A
2026 - Present
F

Freelance

Independent Knowledge Base Builder (LLM-Wiki)

N/A贵州
2026 - Present