UniResearch Community | Kenneth Lee | Activity https://community.uniresearch.ai/members/kennethlee/activity/ Activity feed for Kenneth Lee. Fri, 11 Sep 2026 11:08:05 +0800 https://buddypress.org/?v=3.4.4 en-US 30 hourly 2 ac414b8c768d69117267b424d5e5fbf0 Kenneth Lee posted a new activity comment https://community.uniresearch.ai/news-feed/p/942/#acomment-962 Mon, 07 Sep 2026 12:26:07 +0800 I have used datacite‑tools for generating standard metadata entries.

In reply to - Adam Evans posted an update

Resource Share

Clean metadata is critical for reproducibility, yet manually organising metadata is extremely time‑consuming. Are there lesser‑known lightweight open utilities you […]

]]> 2b5df6ce55022efc05d2cb45f419bdde Kenneth Lee posted a new activity comment https://community.uniresearch.ai/news-feed/p/922/#acomment-939 Mon, 07 Sep 2026 03:44:23 +0800 Do not treat hypothesis failure as personal failure as a researcher.

In reply to - Christopher Hernandez posted an update

Question

It can be demotivating when your main research hypothesis is fully contradicted by solid experimental evidence. How do you pivot your project constructively […]

]]> 94e39dd76d2c8f862daab74d5389b442 Kenneth Lee posted a new activity comment https://community.uniresearch.ai/news-feed/p/840/#acomment-859 Wed, 02 Sep 2026 11:10:20 +0800 I have been delaying updating my academic homepage for months.

In reply to - Min Lu posted an update A well‑maintained personal academic page helps promote your publications, ongoing projects and research background. Clear layout and regular updates are far more important than fancy decorations. ]]> 8458345159aa4ee1ca57ba3ca92b6d4b Kenneth Lee posted a new activity comment https://community.uniresearch.ai/news-feed/p/420/#acomment-459 Thu, 20 Aug 2026 14:27:26 +0800 It also depends on how domain‑different your target corpus is from the base model’s training data. Big domain gap usually favors PEFT.

In reply to - John Smith posted an update

Research Idea

PEFT methods are everywhere these days. But I’m starting to question: for extremely niche small‑size domain data, are LoRA/QLoRA consistently better than full […]

]]>