UniResearch Community | Lihua Su | Activity https://community.uniresearch.ai/members/lihuasu/activity/ Activity feed for Lihua Su. Fri, 11 Sep 2026 11:08:05 +0800 https://buddypress.org/?v=3.4.4 en-US 30 hourly 2 8827314e94777005018fd9f2786b75ba Lihua Su posted a new activity comment https://community.uniresearch.ai/news-feed/p/1007/#acomment-1025 Wed, 09 Sep 2026 12:27:09 +0800 Keep simple baseline conditions for comparison against all ablated variants.

In reply to - Xiuhua Wei posted an update

Method Help

Poorly‑designed ablation studies waste computing resources and deliver limited insight. Many new researchers struggle to decide which components should be ablated. What […]

]]> df49ee06211add78719aca49f197ca94 Lihua Su posted an update: According to a newly released higher‑education report, […] https://community.uniresearch.ai/news-feed/p/774/ Mon, 31 Aug 2026 11:15:06 +0800 According to a newly released higher‑education report, average PhD completion time continues to increase across multiple regions. Funding instability and complex project scope are two major contributing factors.

]]> 3 6c01662a6b4269840844292a10a934e4 Lihua Su posted a new activity comment https://community.uniresearch.ai/news-feed/p/726/#acomment-742 Sat, 29 Aug 2026 03:49:04 +0800 Many early‑career researchers overlook limitation discussion.

In reply to - Ping Lu posted an update

Paper Writing Guidance

The discussion section is easy to write repetitively. Avoid simply restating experimental results; focus on interpreting meaning, analyzing limitations and […]

]]> a63f920129fc1e4510d8e2fe3ff3e5fe Lihua Su posted an update: Hypothesis‑driven work is highly valued, but open‑ended […] https://community.uniresearch.ai/news-feed/p/681/ Thu, 27 Aug 2026 13:47:34 +0800 Hypothesis‑driven work is highly valued, but open‑ended exploratory work can also spark major breakthroughs. How do you balance exploratory data work and formal hypothesis testing within your projects?

]]> 6 a7d7c98e439f1a7d09c4c4aa4d554bf8 Lihua Su posted a new activity comment https://community.uniresearch.ai/news-feed/p/646/#acomment-657 Thu, 27 Aug 2026 03:48:15 +0800 Be careful not to jump between too many topics, that can fragment your research profile.

In reply to - Tping posted an update Every researcher faces moments where a project yields little progress after sustained effort. How do you decide whether to keep pushing forward or pivot your research question? ]]> 99bf9a9d1a0eee7f314925df7cc3238f Lihua Su posted a new activity comment https://community.uniresearch.ai/news-feed/p/604/#acomment-617 Tue, 25 Aug 2026 12:31:47 +0800 Define authorship expectations and task division at an early stage of cooperation.

In reply to - Lei Zhao posted an update More research happens across institutes and time‑zones. Remote collaboration brings flexibility yet creates communication friction. What habits make remote academic partnerships work smoothly? ]]> 4de08baed1b36ebb5de8fc949d63bd53 Lihua Su posted a new activity comment https://community.uniresearch.ai/news-feed/p/420/#acomment-462 Thu, 20 Aug 2026 14:32:41 +0800 One downside of PEFT I rarely see discussed: poor cross‑task generalization when you stack multiple LoRA adapters.

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 […]

]]>