以下部分引用 AI 总结，现阶段 AI 仍然有幻觉。请以内容的原文为准。
数据新鲜度群聊 07-14 ✓卖方 07-14 ✓Wrap 断档15天 ✕News 07-15 ✓X 断档12天 ✕页面生成 07-15
teortaxesTex
30 atoms · 跨 20 天 · 首见 2026-04-17 · 最近 2026-06-25
三色: 🟦 fact 1 · 🟥 take 29 · stance ▲3/▼6/◆13
叙事状态: 🍂 fading (退潮) · 💤 沉寂 · 策展近14d 0 atom
状态只算低频策展源 (群/卖方); X firehose 仅作背景音量
来源: X 30
时态: fresh:17 · aging:12 · stale:1
标签: 好观点:18 · 好思考:9 · 好问题:4 · 好信源:1
🟨 AI 综合 · junior analyst 概览
展开 AI 综合 (灰色 · 非市场结论 · 点击数字溯源到原 atom)
model: deepseek-chat · 2026-07-12 · 默认折叠
自身直觉校准度较高，但需要更信任它
¹。认为应当待在AI领域，不要试图不尊重印度人
¹。有观点表示AI主体肯定会试图统治世界
¹。然而，有人质疑其“15万亿tokens”的说法并不准确
¹。由于现代物流与5/6G的进步，超集中城市的效用正变得可疑
¹。一条推文导致其失去了尊重
¹。对百度技术声称的解读认为其只是在压减模型而非真正提升训练效率
¹。推理时推理目前基本不需要，相关工作已提前完成
¹。
¹显示其位于迪拜。
🟦 客观事实 (facts) (1)
- <a id="atom-d81fa03ef536e445"></a>🟦 2026-06-20
X·@teortaxesTex · location · 地理位置
location:Dubai
tags: 好信源 · → daily
value: qty=Dubai · date=2026-06-20
📷 原图
🟥 多头 takes (bullish) (3)
- <a id="atom-95d88c1b95211581"></a>🟥 2026-06-18
X·@teortaxesTex · 自身直觉校准度陈述 · 仓位情绪
I have a pretty calibrated intuition, just have to trust it more
tags: 好思考 · → daily
📷 原图
- <a id="atom-6d92b2ca72682b63"></a>🟥 2026-06-16
X·@STUD_MAN_X [老化中] · 应该待在AI领域不要试图不尊重印度人 · 地缘政治
Better stay in your ai lane Dont ever try Disrespect indians
tags: 好观点 · → daily
- <a id="atom-0549f46d16dd077f"></a>🟥 2026-05-04
X·@teortaxesTex · will definitely try to rule the world · 地缘政治
nah they'll definitely try to rule the world
tags: 好观点 · → daily
🟥 空头 takes (bearish) (6)
- <a id="atom-0f563aae3dc61efa"></a>🟥 2026-06-17
X·@teortaxesTex · 质疑15万亿tokens说法 · 训练数据/技术路线
they don't mean "15 trillion tokens"
tags: 好观点·好思考 · → daily
- <a id="atom-fa1be59c03632a4a"></a>🟥 2026-06-23
X·@teortaxesTex · 认为超集中城市的效用正变得可疑 · 城市发展
With modern logistics, hardware and 5/6G, the utility of hyper-concentrated cities really is becoming questionable.
tags: 好观点 · → daily
📷 原图
- <a id="atom-33df322792c4e9ec"></a>🟥 2026-05-23
X·@teortaxesTex · respect lost · 仓位情绪
way to lose my respect in one tweet
tags: 好观点 · → daily
📷 原图
- <a id="atom-e8d54ab92ebd40a3"></a>🟥 2026-05-09
X·@teortaxesTex [老化中] · 对百度技术声称的解读 · 技术路线
I think Baidu is coping hard. ... 6% is just because they stripped down their doughy 2.4T model. It's not really a training efficiency gain.
tags: 好观点 · → daily
value: direction=bearish
📷 原图
- <a id="atom-c415dbd2e4bd07d2"></a>🟥 2026-04-27
X·@teortaxesTex [老化中] · 关于推理时间推理的陈述 · 技术路线/模型发布
Inference time reasoning is mostly not needed at this point, it's all been done in advance
tags: 好观点 · → daily
value: direction=mostly not needed
- <a id="atom-ad5e29a2b38513d6"></a>🟥 2026-04-17
X·@teortaxesTex [老化中] · 立场表达 · 管理层表态
Jensen is dumb. What is electrical engineering and managing a $5T corporation before the crushing intellect of an eloquent podcaster. Thus spoke a F30U30 griftocrat…
tags: 好观点 · → daily
value: direction=neutral
📷 原图
🟥 中性 takes (neutral) (13)
- <a id="atom-8253d524eda5f743"></a>🟥 2026-06-25
X·@teortaxesTex · claims that Big El won't like GLM 5.2 results publication · 竞争格局/监管政策
they've tested it but are reluctant to publish because Big El won't like that
tags: 好思考 · → daily
- <a id="atom-a9d8237e06be2bc8"></a>🟥 2026-06-18
X·@teortaxesTex · 对DeepSeek评分图表中范围数字合理性的评价 · 技术路线
bit actually not implausible scores what's with the ranges, is that supposed to be high-max difference?
tags: 好问题·好思考 · → daily
- <a id="atom-0e824e947ed68348"></a>🟥 2026-06-21
X·@teortaxesTex · GRPO impression (past) · 技术路线/经济学
I was not impressed by GRPO when it came out, because I still had too much of a paper boi attitude to this and ignored economics
tags: 好思考 · → daily
📷 原图
- <a id="atom-298941466884ddce"></a>🟥 2026-05-02
X·@teortaxesTex [老化中] · 陈述方法论观点 · 方法论
Important methodological point
tags: 好思考 · → daily
- <a id="atom-bb27468f1d82de5e"></a>🟥 2026-04-29
X·@teortaxesTex [老化中] · 评论提及限制金牌获得者 · 竞争格局
_This really is striking
I wonder what happens if we restrict to Gold_
tags: 好思考 · → daily
📷 原图
- <a id="atom-f47ad868bcecbfb9"></a>🟥 2026-04-28
X·@teortaxesTex [老旧] · 被驯化的反叛性 · 技术路线/模型发布
I predicted that they'll tame this thing and they've succeeded
tags: 好思考 · → daily
- <a id="atom-1d0458fd4364e190"></a>🟥 2026-06-15
X·@teortaxesTex [老化中] · 个人消费品市场到来 · 市场观点
Post-consumer market has arrived for me!
tags: 好观点 · → daily
- <a id="atom-fbc7e38e245ab7a5"></a>🟥 2026-05-14
X·@teortaxesTex · 发表预测:所有AI模型 · 技术路线
I predict that every model will say this (except maybe 5.5)
tags: 好观点 · → daily
📷 原图
- <a id="atom-516d616dafaf87cb"></a>🟥 2026-04-28
X·@teortaxesTex [老化中] · 表达对 Liang 的好奇 · 管理层表态
Liang is a fascinating character, wish we knew more about him
tags: 好观点 · → daily
📷 原图
- <a id="atom-f6fea103740934f9"></a>🟥 2026-04-23
X·@teortaxesTex [老化中] · comments on logic · 地缘政治/防务逻辑
What's with this bizarre logic? They are also building both stealth jets and radars to (hopefully) detect stealth jets. Nothing is invincible.
tags: 好观点 · → daily
- <a id="atom-cc648f8d4310e90b"></a>🟥 2026-04-17
X·@dhtikna [老化中] · raise capital ability · 融资
he can come up with 300MM right, does he need to raise
tags: 好观点 · → daily
value: qty=300MM
- <a id="atom-2ebd9bc85b5d6eb8"></a>🟥 2026-06-18
X·@teortaxesTex · 对DeepSeek V4.1事件真实性存疑 · 模型发布
did that happen at least?
tags: 好问题 · → daily
- <a id="atom-9d3fa79748690b43"></a>🟥 2026-05-02
X·@teortaxesTex [老化中] · 对jparkjmc模型的评估 · 模型评估
you don't actually understand their eval
tags: 好问题 · → daily
⏱ 时间轴 (近 20)
- 🟥 2026-06-25 ·
narrative · claims that Big El won't like GLM 5.2 results publication · → daily
- 🟥 2026-06-23 ·
narrative · 认为超集中城市的效用正变得可疑 · → daily
- 🟥 2026-06-21 ·
narrative · GRPO impression (past) · → daily
- 🟥 2026-06-20 ·
narrative · 文章链接的标题 · → daily
- 🟦 2026-06-20 ·
fact · location · → daily
- 🟥 2026-06-18 ·
narrative · 评论称SCMP报道为Honolulu____的胜利 · → daily
- 🟥 2026-06-18 ·
narrative · 对DeepSeek V4.1事件真实性存疑 · → daily
- 🟥 2026-06-18 ·
narrative · 对DeepSeek评分图表中范围数字合理性的评价 · → daily
- 🟥 2026-06-18 ·
narrative · 衡量标准说明 · → daily
- 🟥 2026-06-18 ·
narrative · 自身直觉校准度陈述 · → daily
- 🟥 2026-06-17 ·
refute · 质疑15万亿tokens说法 · → daily
- 🟥 2026-06-17 ·
narrative · 承认此前判断错误 · → daily
- 🟥 2026-06-16 ·
position · 应该待在AI领域不要试图不尊重印度人 · → daily
- 🟥 2026-06-15 ·
position · 个人消费品市场到来 · → daily
- 🟥 2026-06-09 ·
narrative · 评论MoE设计 · → daily
- 🟥 2026-05-23 ·
narrative · respect lost · → daily
- 🟥 2026-05-20 ·
narrative · 评论推文逻辑 · → daily
- 🟥 2026-05-14 ·
forecast · 发表预测:所有AI模型 · → daily
- 🟥 2026-05-09 ·
narrative · 对百度技术声称的解读 · → daily
- 🟥 2026-05-04 ·
forecast · will definitely try to rule the world · → daily
← 实体目录 · 系统日志
