StopYourLimits | Looking For A Vaccine 🤯

48.6K posts

StopYourLimits | Looking For A Vaccine 🤯 banner
StopYourLimits | Looking For A Vaccine 🤯

StopYourLimits | Looking For A Vaccine 🤯

@limits_stop

🏖 Finance & crypto, health🫀, fitness (including brain 🧠 fitness & productivity) and global trends

Katılım Eylül 2019
2.3K Takip Edilen3K Takipçiler
Sabitlenmiş Tweet
StopYourLimits | Looking For A Vaccine 🤯
I know what you're thinking. How long will this take? When will it end? A literal bioweapon is destroying brains around you. Focus on avoiding infection for another year, then re-evaluate.
English
16
53
409
31.7K
vikram srinivasan
vikram srinivasan@vikramsrinivasa·
@RuiTheBaker happy to discuss over call. You can do lot of tricks via late interaction etc., we started with late interaction + lexical. But don't need all that complexity, no chunking also, again added complexity and cost. Can achieve very very high accuracy, low TCO and explainability with far simpler approaches. We are operating at 100+TB scale with very low TCO.
English
1
0
1
29
Sumit
Sumit@_reachsumit·
Is Grep All You Need? How Agent Harnesses Reshape Agentic Search PwC presents an empirical study showing that lexical search (grep) generally outperforms vector retrieval in agentic RAG workflows, but overall accuracy depends heavily on agent harness. 📝 arxiv.org/abs/2605.15184
English
1
7
62
2.8K
StopYourLimits | Looking For A Vaccine 🤯 retweetledi
Alex Albert
Alex Albert@alexalbert__·
With the help of Claude Mythos Preview, the Firefox team fixed more security bugs in April than in the past 15 months combined.
Alex Albert tweet media
English
345
1.3K
15.5K
1.5M
StopYourLimits | Looking For A Vaccine 🤯 retweetledi
Ignasi Guardans
Ignasi Guardans@iguardans·
Ya sé que Uds usan ChatGPT para cosas muy serias. Pero prueben como asistente de cocina. Al detalle. Incluso dándole alguna foto (del pescado, de la carne, de la composición) para ajustar tiempos e instrucciones. Todo lo que siempre quiso cocinar pero nunca se atrevió a hacer…
Español
6
6
35
4.7K
StopYourLimits | Looking For A Vaccine 🤯
@tlbtlbtlb Good. I'd agree. Is there a way we don't give machines power over humans that doesn’t start by 'If only every country and person agreed on not using the most powerful tool available to them...'? Once best possible CEOs and commanders are machines, will we not use them?
English
0
0
0
29
Trevor Blackwell
Trevor Blackwell@tlbtlbtlb·
@limits_stop I propose we not give machines power over humans. (We might do this for some individuals, like with robot prison guards, but not generally.)
English
1
0
3
56
Trevor Blackwell
Trevor Blackwell@tlbtlbtlb·
This problem is caused by utilitarian ethics landing on "things with consciousness deserve moral consideration", thus demanding a definition of something undefinable.
Wolf Tivy@wolftivy

@viemccoy Me I just deny the ontological reality of “consciousness”. Oh you think my or your subjective experience is not reducible to computational phenomena? Prove it by pointing to intersubjectively verifiable features that can’t be explained with computer science. You can’t.

English
2
0
9
2K
StopYourLimits | Looking For A Vaccine 🤯
@tlbtlbtlb While you're right, this only applies effectively because humans have the power (and only to the extent power imbalances between societies are not huge). It's a description of power. What if machines had power over humans?
English
1
0
0
52
Trevor Blackwell
Trevor Blackwell@tlbtlbtlb·
We should just admit that humans deserve moral consideration because we say so, and anything else is at our discretion. Dogs yes, chickens maybe, computers no. Problem solved.
English
4
1
17
1.9K
StopYourLimits | Looking For A Vaccine 🤯 retweetledi
Mecha News
Mecha News@MechaNews_·
🚨 US–Iran talks are underway… as Israeli strikes on Lebanon continue.
English
1
11
79
4.5K
StopYourLimits | Looking For A Vaccine 🤯 retweetledi
StopYourLimits | Looking For A Vaccine 🤯 retweetledi
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
Israel does not consider it a survivable situation if it cannot unilaterally destroy *every* country in the region. And particularly hostile countries, and most of all Iran. They *are* willing to burn relations with the US over this, to take massive damage. They can do it.
English
2
1
48
3K
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
I can't understand what he's talking about. People, Israel has a clear principle, called the Begin Doctrine. If there is no other way to prevent Iran from acquiring WMDs, they will simply nuke Iran. Fordow, missile cities, and if needed civilian cities too. Iran can't have nukes.
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) tweet media
Clash Report@clashreport

Professor Robert Pape: Iran has enough enriched uranium for 10 to 16 nuclear weapons. You could expect a nuclear test in six months — maybe a year. There is nothing the United States or the world can do about that.

English
12
3
70
9.2K
StopYourLimits | Looking For A Vaccine 🤯
@kevinnguyendn @ashwingop I disagree that natural language / semantics has finite dimensions. You can navigate it as infinite-dimensional if you model semantics as fractal. IMHO graphs are the most efficient representation for that -- BM25, dense embeddings, ColBERT, .md, etc. should play a role in this
English
1
0
1
65
andy nguyen
andy nguyen@kevinnguyendn·
Great work formalizing what we've been building against empirically. Two additions from the ByteRover side The LoCoMo number is actually 96.1% (92.8% is LongMemEval-S) — but the more interesting point is why. Our retrieval layer is pure BM25, which as your experiments show produces b=0.000 and FA=0.000. The semantic understanding comes entirely from the LLM reasoning over structured markdown, not from the retrieval mechanism itself. This is your Option 2 happening at the architectural level: the filesystem IS the episodic record, BM25 IS the exact retrieval, and the LLM IS the semantic layer. They never mix. What the post doesn't cover is managed forgetting. ByteRover's Adaptive Knowledge Lifecycle actively navigates the tradeoff frontier you describe: importance decay, maturity tiers (draft→validated→core with hysteresis), and archive stubs that preserve retrieval hooks while releasing full content. The system deliberately forgets, but it forgets the right things, and keeps ghost cues so it can recover. The real question your theorem raises for us: as the knowledge base scales to org-level (thousands of agents, years of history), does the LLM semantic layer start hitting the same d_eff≈10-50 ceiling when reasoning over the retrieved set? That's the frontier we're exploring next DM'd you, would love to chat more.
English
2
0
7
965
François Fleuret
François Fleuret@francoisfleuret·
A part of accepting reality is that people do Bayesian inference about you, and *there is no other way*. It is sometime unfair one way or another. You did not choose many features people perceive about you, things may not be what they look like to others, but that's it.
English
8
5
68
4.8K
StopYourLimits | Looking For A Vaccine 🤯 retweetledi
Bashkarma🇺🇸🌏🇷🇺
Case in point: More than 70 transport planes landed in the Middle East within 24 hours of the ceasefire taking effect. That scale suggests possible preparation for a ground offensive, solidifying suspicion that Trump is using the truce to regroup:
Bashkarma🇺🇸🌏🇷🇺 tweet media
Bashkarma🇺🇸🌏🇷🇺@Karmabash

The so-called "Pakistani" truce proposal is awfully similar to the Minsk agreement trap that ultimately fueled a larger conflict in Ukraine. Just as Ukraine used Minsk to regroup and rearm, Trump is likely to use this "Pakistani" (effectively his own) proposal to do the same.

English
264
3.6K
8.1K
784.6K
Policy Tensor
Policy Tensor@policytensor·
Pape said something similar about structural conditions for war still being there. But structure alone did not cause the war. It would not have obtained without a massive intelligence failure. Caine and the WH now know that Iran cannot be disarmed from the air and there is no military solution to the problem of retaking Hormuz. This is what has been revealed by this clash of arms. This means that resuming the war can only get you back to square one, with no prospect of defeating Iran or forcibly reopening Hormuz. So Vance has to secure a deal. This is why I am relatively confident about peace.
Dehati Armageddon Neutraliser@ImperiumHindu

How can it hold unless- 1. US accepts Iranian demands which include Iranian supremacy in Hormuz and US withdrawal from Middle East which essentially means start of dissolution of US empire Or 2. Iran leaves some of its key demands (mentioned in point no 1) which will make it an Iranian defeat. Iranians accepting this will *embolden* US-Israel for another attack some time later. Structural conditions which produced war have not been resolved and even if US wants to leave the conflict, Iranians will insist US withdrawal from Middle East which it cannot accept unless it voluntarily wants to dissolve petrodollar and it's empire.

English
15
14
100
13.1K
StopYourLimits | Looking For A Vaccine 🤯 retweetledi
Caitlin Johnstone
Caitlin Johnstone@caitoz·
The world can have peace or Israel. It can’t have both.
English
519
9.5K
33.6K
374.8K
StopYourLimits | Looking For A Vaccine 🤯 retweetledi
Arash Azizi آرش عزیزی
Arash Azizi آرش عزیزی@arash_tehran·
Account close to power in Iran says if Israel continues attacks on Lebanon, Iran will end the ceasefire and cancel the Islamabad talks with the US scheduled for Friday
هنر جنگ@dolfiniran

۱۴:۵۱/ ۱۹ فروردین با آغاز حمله رژیم اسراییل به لبنان احتمال لغو نشست جمعه جدی شده است. نیروی دریایی سپاه جلوگیری از تردد کشتی ها را آغاز کرده است.

English
11
55
151
45.8K