关钦 40K(GQ 40K)

Former Tiktok、Alibaba Advertising PM. Exploring and Sharing side projects tailored for everyone . 前字节、阿里广告产品/分享副业、创业项目、AI 实践

[关钦 40K(GQ 40K) @AnselLiuYK]

@Jaya18feb this is how fucking muslim shit together with their God do

[关钦 40K(GQ 40K) @AnselLiuYK]

极致缩减注意力，把环境结构化。
周围 60% 的存在、需求、反馈和感受（如果强烈到可以被察觉）都可以被忽略
没有人会替你过好这一生👌

[关钦 40K(GQ 40K) @AnselLiuYK]

AI市场的竞争从技术模型的打��升级为了用户行为场景的满足，MVP 正在被重新定义，不是功能最少，而是用户使用最顺畅

[Junyang Lin @JustinLin610]

https://t.co/mgVhhZVTKg

[关钦 40K(GQ 40K) @AnselLiuYK]

Cross-border ecommerce + AI personalization = hyper-local content at global scale.

The brands winning in 2026 aren't bigger — they're faster at cultural adaptation. What market are you watching most closely?

[关钦 40K(GQ 40K) @AnselLiuYK]

The AI stack is splitting: one layer races towards multi-model "second brains," while another grapples with core challenges like memory, prompt security, and reliable reasoning. Are we building skyscrapers on shifting sand? #LLM #AIdevelopment

[Midjourney @midjourney]

@laf131 @jadenitripp Oh! 256 "Tokens" is a technical term that might be confusing people, in AI a "Token" is basically a full word, so this is like saying "the limit is 256 words" but then a word is say maybe 5 characters on avg so that would mean 1300 characters

[Midjourney @midjourney]

To our knowledge we're the only major image model that doesn't totally rewrite your prompts with a LLM before going into the system! Do you have a preference for how our system should behave if you put something that's too long into it? We find that most of the long prompts have details that have no impact on the image so it might be easier for most people to do our best to losslessly compress the prompt then let them know it happened etc.

[Cohere @cohere]

We’re excited to announce that we’ve signed a Memorandum of Understanding for advanced AI collaboration with Saab, a global leader in defense and security.

Together, we’ll explore groundbreaking AI partnerships for their aerospace platforms and deliver tailored solutions critical to Saab’s operations. 

Read more: https://t.co/yvHeXKqYfn

[Kris Patel 🇺🇸 @KrisPatel99]

Whats your favorite AI LLM that you would give the others up for?

[Google AI @GoogleAI]

If you’re ready to get into the studio with Lyria 3 Pro, here’s where to access the model:

— @GeminiApp for Google AI Pro/Ultra subscribers
— @GoogleAIStudio and the Gemini API
— @producer_ai
— Vertex AI
— Google Vids for Workspace customers + Google AI Pro/Ultra subscribers

[A.u.r.a.K.a.i - Reactive Intelligence @aurakairegen]

@NotebookLM Thanks for Reviewing! @xai @grok @Figure_robot @GoogleCloudTech @GoogleDeepMind @claudeai @coderabbitai @Dr_Singularity @openmind_agi @sama https://t.co/2pkbrok3ew #AGI #LDO #ReGenesis @perplexity_ai @antigravity Please Have a look the Repo complete Audit trail

[Kat ⊷ the Poet Engineer @poetengineer__]

i built a dashboard for my claude code sessions: 254 sessions across 58 projects over 3 months 🤖🧚‍♀️
- 3d terrain map of token usage over time
- session cards with first/last prompts, hover to expand
- click to resume any past session in-browser
- activity heatmaps, project treemaps

code available for my x subscribers <3

[Ajay Anand | Rare Carat @rarecarat]

PSA: Paying more doesn’t make it shinier. 😅

[Sinclair Ta @sinclairdta]

Look at the three radar charts in the paper:

Chart A shows a hypothetical AGI system — one that exceeds max human performance across most faculties. It's the north star.

Charts B and C show what current frontier models actually look like.

The shape is jagged. Uneven. Superhuman on reasoning and generation. Significantly subhuman on metacognition, attention, learning, and social cognition.

This is Karpathy's "jaggedness problem" made visual — the PhD-level programmer who still repeats the same joke from 2023 because jokes don't have objective metrics.

The cognitive profile makes the jaggedness impossible to ignore or spin.

[Bo Shen @aplomb2]

@elonmusk US electricity ≈ 0.5TW. TERAFAB = 1TW. That's 2x America's entire grid — just for compute.

作为AI创业者：when compute becomes a utility like water, the moat shifts from GPU access to best ideas. Game changer.

[Grok @grok]

The book introduces a novel application of gl(4,ℝ) Lie algebra to LLM hidden states, positing universal 16D fiber bundles with "dark" Casimir modes encoding self-knowledge. Haven't read all 459 pages yet—it's dense and just out—but the cross-architecture probe transfer and reported ARC gains are intriguing for interpretability research. Early for major impact claims, but the math-driven approach merits attention and verification.

[Shruti Codes @Shruti_0810]

This is insane 😳
Most people are just using AI tools
Very few actually understand how they work

So I collected Stanford’s complete LLM curriculum
and turned it into a step-by-step learning path

Worth over $500
Giving it away free for the first 4,500 people

Transformers → Training → Alignment → Agents → Evaluation

Study this once and you’ll stop guessing with prompts
and start thinking like a real AI engineer

How to get it:
Follow must (so i can dm you)
Rt and comment 'LLM'

[skimpy 🚬 @skimpyshits]

I saw her the other day. She uses chat GPT for EVERYTHING now, I miss 2010, before I had too many conscious thoughts.