← 返回名录条目信息来自其公开主页与公开发布内容。按本站规范，页面不展示任何联系方式。 需要更正或删除？通过收录与更正通道提交，24 小时内处理。
J
Jiekang Yang
MSc student @ CUHK-Shenzhen | Reliable RAG, Visual RAG, Multimodal Retrieval, Document AI
- 公司
- CUHK-Shenzhen T-Lab
- 位置
- Shenzhen, China
- Stars
- 1
- 粉丝 / 仓库
- 0
代表作品 / 项目
- llm-translation-poc⭐ 1Terminology-constrained translation for automotive standards and regulations. Treats terminology consistency as a process-level hard gate and measures it separately from aggregate translation quality, because aggregate MT scores cannot certify that specified terms are exactly right.
- indexed-sink⭐ 0Measurement study of attention sinks that land on indexed image patches in late-interaction visual document retrieval (ColQwen2.5 / ViDoRe V3).
- PortfolioResearchCockpit⭐ 0Personal portfolio research and paper-trading cockpit built on TradingAgents — holdings, risk limits, backtesting, walk-forward simulation and prediction review. Research aid only; no broker connection, no automated trading.
- encodec-facodec-speech-prediction⭐ 0Autoregressive speech prediction with EnCodec and FACodec token language models: 2s context -> 1s future speech, with STOI/PESQ/DNSMOS analysis of where those metrics disagree.
- edu-policy-qa⭐ 0An evidence-annotated benchmark for RAG over university course syllabi — 80 questions with document, page and supporting-span labels, including unanswerable items, evaluated across closed-book, full-context, basic RAG and citation-constrained RAG.