Avatar
X media ↗
Gorden Sun
Open X@gorden_sun · X · Global · OTHER
Opens the X profile. · Are you @gorden_sun? Claim this page
Account noteOpenClose
Real X profile and recent post evidence are attached for @gorden_sun; some items are still pending moderation.
Risk reading
High uncertainty
Evidence
Medium · 40
Risk reading
High uncertainty
Evidence
Medium · 40
The record has unresolved uncertainty from risk tags, moderation, mixed signals, or thin context. Treat this as a review prompt.
Nothing on this page is an identity, safety, or transaction verdict — it is collected evidence plus interpretation.
Main basis
Reading limits
Image inventory
Image inventory is stale
24
Saved media
Image inventory
Image inventory is stale
24
Saved media
Saved images exist, but the latest capture is past the freshness window. Refresh before relying on this inventory.
Post images
16
Unique images
16
Linked posts
16
Freshness
Needs refresh
Latest captured media: 2026-08-21 00:16
Refresh only adds new media. Saved media stays archived even after the creator deletes the originals.
Quick read
Key account facts
Open
Quick read
Key account facts
Registration history
13y 1m
Follower / following
64.6K / 1,734
Posting evidence
30 collected
Profile density
Checkable profile evidence
Community calibration
Does this score match what you see? Your signal helps calibrate the account.
Radar score
78
Risk
high
Followers
64.6K
Following
1,734
Posts tracked
30
Data status
Evidence available
A real collected snapshot backs this page (updated 28 day(s) ago). Reviews and moderation may still revise this reading.
Result source: Saved + collected evidence
Saved server record plus real collected X evidence.
Latest collection run: Collected
Status: collected 30 recent post(s).
Started at 2026-08-21 00:16 · Finished at 2026-08-21 00:16
📖 Member preview
Preview
Full media inventory
Preview count only; full archive opens after member access
Linked posts
Members can inspect which posts support saved media
Interaction structure
Members see more context behind score movement, not just the score
Risk reasons
Members can read why the page asks for caution or review
History changes
Members can inspect selected historical archive and timeline context
Context only — not identity or safety conclusions.
Free: one full view per account. Member: full score logic + more reviews. Pro: deep archive + corrections.
Collected X evidence
Profile snapshot
Collected profileFollowers
65K
Following
1,734
Tweets
8,642
Recorded at
2026-08-21 00:16
Collected media evidence
Library note
Previewing first 12 of 24. Open the archive tab for the full media library.
Banner
X media ↗Sign in as a member to view this video or protected image.
Video
Source tweet ↗Video thumb
Source tweet ↗Post image
Source tweet ↗Post image
Source tweet ↗Post image
Source tweet ↗Post image
Source tweet ↗Post image
Source tweet ↗Post image
Source tweet ↗Post image
Source tweet ↗Post image
Source tweet ↗Followers
65K
Following
1,734
Tweets
8,642
Source
third_party_api

Gorden Sun
Collected at: 2026-08-21 00:16
只发AI相关信息,个人维护的AI资讯日报(已连续日更3年)👇
Recent collected posts
大哥加入了Anthropic,恭喜他!
Generalist 发布了他们最新的机器人模型 GEN-1.5。简单来说,他们让机器人学会了一项新本领:你只要在它眼前把一个动作示范一遍,它看个几秒钟,就能当场照着做出来。堪称机器人版 Record & Replay。 以往要让机器人做一件新事情,工程师往往得花好几个月写代码、调试程序,或者喂给它海量的数据反复训练。这个新模型更像一个反应很快的学徒,你当着它的面做一次,它看明白了就直接上手,中途不需要停下来重新写代码或重新训练。 它还展现出了一些很有意思的实用能力: · 看一眼就能跟着做:给它看一段 3 到 12 秒的动作演示,比如拧开玻璃瓶盖、拉开笔袋拉链,它马上就能自己操作,平均成功率大概在 59% 左右。 · 自己把动作串起来:如果你分别教它“怎么拉开笔袋”和“怎么把钱拿出来”,它自己就能把这两段零碎的记忆拼在一起,顺畅地完成一整套连贯动作。 · 看人手或者电脑动画也能学会:你在电脑模拟画面里演示一次,或者直接用自己的双手在镜头前做一遍,它都能把动作转化成自己的机械臂动作。 · 手边有什么就用什么:原本教它用毛刷把积木扫进碗里,如果你拿走毛刷、塞给它一根香蕉,它会把香蕉当成刷子来扫;如果给它一个簸箕,它甚至会自己琢磨出用簸箕把积木铲起来倒进碗里。 · 稍微练习几分钟就很稳:只看一次示范它就能做个大概,要是再给它大约 5 分钟的演示数据让它快速适应几步,成功率就能直接提升到 83%。 这意味着以后普通人使用机器人可能会变得非常省事,想让它干什么活,只要当面演示给它看就行了。 官方介绍:https://t.co/T4YGrDs4g1
开源视频创作平台,部署上配好API KEY就能直接开卖了
《 AI 时代的数学》 一个数学难题,如果 AI 生成了一份很长的证明,经机器验证是正确的,可是没有任何人看得懂,甚至连提交它的人自己也看不懂,这算数学的进步吗? 陶哲轩在Arxiv发表了一篇论文,阐述了他对AI时代的数学的看法。 刚刚提到的问题已经在实际发生。专门收集数学问题的网站上已经出现了数十份AI生成的证明提交。其中许多很可能是正确的,但在相当多的案例中,尚无人类专家自愿去验证和担保这些证明;在若干案例中,提交者本人也声明自己没有资格对证明做出判断。 我们正在从一个证明稀缺的时代过渡到一个证明过剩的时代。 我们很快就会面临:一个重大数学结果拥有经过机器验证的证明,但没有任何人类能够充分理解它并加以解释。 陶哲轩的观点是: “如果作者无法令人信服地证明自己能够就其成果做一场正确且归属恰当的专家级报告,那么这个成果就不应该发表。一个没有人类能够正确解释的证明,即使经过形式化验证,也应当被视为不完整的。” 一个证明要真正对其所在领域有所贡献,仅仅正确是不够的,仅仅可读也是不够的。它还需要被学术共同体接受和重视:其他数学家需要消化这一结果,并将其融入自己的工作之中。 但现有的大多数制度安排,期刊、优先权惯例、聘用与晋升标准、奖项,乃至“研究纲领”这个概念本身,都是在稀缺假设下设计的。现在还没有好的解决办法,陶哲轩只提出了一些学术上的倡议。 在 AI 时代,数学界最需要的人才,可能不是最能证明定理的人,而是最能解释定理的人。 论文:https://t.co/AmQ9KgdRTc
Ornith发布Ornith-1.5系列模型 特色是能自我提升,模型不再依赖人工固定的数据集,而是自主构建“提出新任务 -> 生成任务特定脚手架/评估环境 -> 生成解题轨迹”的闭环,并通过强化学习(GRPO)联合优化这三个阶段。 包含3个版本: 397B MoE版本,评分优于DeepSeek-V4-Flash-0731; 35B MoE版本,评分大幅优于Qwen 3.6-35B 9B版本,评分大幅优于Qwen 3.5-9B 模型:https://t.co/JCKbIJHA1S
📊 Metric notes
6 items
Registration history
13y 1m
Created at 2013-08-09 10:31
Audience scale
64.6K
Follower count at the last collection.
Follower / following
64.6K / 1,734
Followers >> following · Followers far outnumber following — a strong structural signal. Ratio: 37.
Posting evidence
30 collected
Latest collected posts become the basis for content and risk interpretation.
Profile density
Checkable profile evidence
Enough stored structure and content samples — no empty-account penalty.
Community sample
0
Too thin for a platform conclusion.
🗺️ Identity map
🧩 Judgment stack
Primary identity
Image seller
Traffic funnel
Content themes
Monetization path
Trust signals
Risk signals
🧭 Why this type
Profile: Saved server profile for @gorden_sun. The visible score now blends the durable lookup result with collected X evidence, media inventory, audience structure, and interaction traces; risk remains separate.
Collected posts: 5 saved recent post(s)
Media evidence: 24 saved media asset(s)
Community: 0 public review signal(s)
Location is self-declared.
🪪 Credit signals4
Account nature
Traffic funnel
Based on saved profile state, 5 collected post(s), 24 media asset(s), and 0 public community signal(s).
Positive signals
Stable identity / Low mutual-promo risk
0 higher-weight review signal(s) and no public Photo Match attachment are currently available.
Risk signals
Traffic funnel risk / Unclear pricing
0 caution/risk review signal(s) and 1 pending governance item(s) are attached.
Suitable for
Real X evidence is attached for @gorden_sun, so the profile no longer depends on a pure lookup-time identity guess.
Not enough reviews yet.
🎬 Content experience
54/100Visual appeal
6.5
Exposure level
Unknown from public sample
Stimulation
6.1
Originality
6.0
Update stability
5.0
Paid-match confidence
3.2
📡 Platform signals
P1Following circle
thinAnalyzed
0
Priority candidates
0
Review queue
0
From accounts this handle follows; review-queue means lower priority only.
Open graph viewSite behavior
thin7d views
2
Favorites
0
Reports
0
Photo Match
Verification records
No live public Photo Match result is attached yet.
Use Photo Match when someone needs to prove that a specific photo or teaser video matches the person who appeared at verification time.
Community stance
How the crowd currently leans
No community signals yet — the first tracked review will appear here.
🪞 Identity signals
🧮 Score detail
9 items
Featured media archive
Narrative fallback🏛️ Evidence stack
0 archive cards · 2 timeline notes
0 archive card(s), 2 timeline note(s), and 0 approved review(s) currently back this score.
Related profiles
Compare with another account
每日大乱斗(每日大赛)
43X · high
奶妹2号
41X · high
Rebekka Ray
55X · high
成都杀手倩
63X · low
D&R -湾区Couple
55X · high
Juno Domina
71X · high