陈陈 杨.

Key Skills

Software

Label Studio

Doccano

Top Subject Matter

Multimodal Large Model Training

Intelligent QA/NLP Systems

Top Data Types

Image

Text

Top Task Types

Entity (NER) Classification

Freelancer Overview

Multimodal Large Model Data Annotation & Evaluation Project. Core strengths include Label Studio and Doccano. Education includes Bachelor of Science, Yunnan University (2020). AI-training focus includes data types such as Image and Text and labeling workflows including Entity (NER) Classification.

Labeling Experience

Multimodal Large Model Data Annotation & Evaluation Project

Label StudioImageEntity (NER) Classification

I participated in a multimodal large model data annotation and evaluation project focusing on image-text understanding. My role included annotating images and text according to detailed guidelines, as well as performing self-checks and peer reviews to ensure high accuracy. I contributed to designing evaluation sets and analysis reports, which supported model optimization and improvement. • Annotated multimodal data, labeling entities, relations, and attributes with over 95% accuracy. • Conducted issue reporting and optimized annotation processes, improving efficiency by 20%. • Participated in model evaluation and provided data support for model iteration. • Project data was utilized in large model training, enhancing its performance.

2023 - 2023

Data Annotation Intern – Intelligent QA Dataset Construction Project

DoccanoTextEntity (NER) Classification

As a data annotation intern, I contributed to an intelligent QA dataset construction project aimed at improving model accuracy. My responsibilities involved large-scale annotation of NLP data, including intent recognition and entity extraction. I also focused on enhancing team workflow and annotation consistency. • Processed and labeled approximately 1,500 text items daily, focusing on intent and entities. • Authored annotation manuals, reducing annotation inconsistency rate by 15%. • Analyzed data to identify issues and suggested improvements, lowering rework by 10%. • The resulting dataset improved model QA performance by about 8% and was used in live products.

2022 - 2022

Education

Work History