XnneHang
Home
Archive
Shelf
Stats
About
Friends
Next
EN
文
简体中文
EN
English
Home
Archive
Shelf
Stats
About
Friends
Next
GitHub
XnneHang
I code because I love it.
Collections
Technology
22
Arts & Culture
39
Thought
10
Life
1
Series
Blogging
7
Philosophical Reflections
5
Reading
20
LLM
6
Long-Term Memory
5
NLP
3
Film and Television
14
Games
2
Film Watching
3
Gaming
1
More
#工程
Posts connected by this tag.
1 post
View all tags →
Single Image Does Not Equal Multiple Images: Why VLMs Hallucinate More with Multiple Images, and a Two-Stage Fix
2026-02-08
Technology
·
Learning Note
/
VLM
/
多模态
/
多图理解
/
工程
An investigation into why web and API results diverge during multi-image analysis, from Lost in the Middle to a two-stage approach based on per-image pre-summaries.
2061 words
|
10 minutes