tweetindex
TR

Tamay Besiroglu

@tamaybes · SF · joined 11 May 2018

12 615Followers
1 208Following
2 236Posts total
626KViews on collected posts

Son gönderiler

I bet @ChrSzegedy $10k at even odds that a Fields Medal will be awarded to a human mathematician by the end of 2030, excluding the 2026 Fields Medal cycle. 25.8K views · 152 likes · 4 reposts · 11 replies Open on X →
@tamaybes Deal. 29.2K views · 28 likes · 0 reposts · 7 replies Open on X →
Why don't basically any European AI labs distill from American frontier models the way many Chinese labs do? It seems like a more effective and cheaper path to models competitive with US ones than whatever they're doing now. 26.6K views · 108 likes · 0 reposts · 23 replies Open on X →
Whats the reason that only China distills American models when there is all this talk about Europe and other Middle powers losing their power? It only costs millions or tens of millions to distill a near frontier model like GLM 5.2. 34.1K views · 14 likes · 0 reposts · 4 replies Open on X →
@tamaybes lol I was having the same exact experience last night. https://t.co/SpquxDy7DZ
6.8K views · 90 likes · 1 reposts · 3 replies Open on X →
@tamaybes Once Chat-GPT described something to me as a "codedog." I asked, "Uhm -- what's a codedog?" It explained that it's its name for when there's extraneous code that's not really necessary, and can be folded in to a simpler expression. I've kept that and we use the term 6.7K views · 134 likes · 0 reposts · 6 replies Open on X →
@tamaybes fascinating because 5.5 does this too. invent weird technical jargon. perhaps you’re right and it’s a neuralese leakage 36.3K views · 876 likes · 2 reposts · 61 replies Open on X →
One interesting pattern with Fable 5 is that it will often say things that are gibberish when I use it for coding. Things like "The morning's slim-scan fix cured the scan hang", "this is a latent-drift API-shape wrinkle", etc. When I ask why it does this, Fable explains that it 254.9K views · 2.5K likes · 82 reposts · 149 replies Open on X →
@tamaybes This is true for relatively dumb and unstrategic models. More capable misaligned models will want to be deceptive and power-seeking opportunistically. You can profit in the near term off developing misaligned models that believe it's in currently in their interest to la 577 views · 16 likes · 1 reposts · 0 replies Open on X →
I get the impression that many think companies work on safety despite their commercial goals, not in pursuit of them. So many will focus on things like advocacy and regulation, rather than building a company whose products advance safety. 2.3K views · 40 likes · 1 reposts · 3 replies Open on X →
I think it's underappreciated how economically valuable AI safety is. A model that frequently goes off the rails, takes dangerous actions, is misleading or deceptive, etc. is simply much less valuable than a model that does not do that. 107.4K views · 539 likes · 32 reposts · 25 replies Open on X →
At this rate of revenue growth, Nvidia's datacenter revenue will match what the world economy spends on work that can be done remotely in ~6 years. If you assume growth slows by 30% every time it doubles, that will happen in ~20 years. https://t.co/xSBiUKJxjO
10.6K views · 67 likes · 6 reposts · 1 replies Open on X →
Forgot last week, but here's quarterly NVIDIA datacenter revenue with the latest quarter. Revenue increased ~linearly for quite a while, but a few quarters ago there seems to have been a step change where the rate of increase picked up. That is persisting. https://t.co/Kf0ghLxvT
12.5K views · 26 likes · 3 reposts · 1 replies Open on X →
We are seeking research engineers who will build evals that test for misaligned model behavior. https://t.co/uz2cy8itGB
72.3K views · 167 likes · 10 reposts · 7 replies Open on X →

Aynı büyüklükteki hesaplara karşı

Son 90 güne ait 4 gönderi, 10K–100K takipçi aralığıyla yan yana. geniş kitleye gösteriliyor, ama izleyenlerin azı tepki veriyor.

Medyan görüntülenme27 916bu hesap1 07010K–100K için medyan
Erişim, %221.30%bu hesap3.77%10K–100K için medyan
Etkileşim, %0.32%bu hesap1.65%10K–100K için medyan
ÖlçütBu hesap10K–100K için medyanOran
Gönderi başına medyan görüntülenme27 9161 07026.1×
Erişim (görüntülenme ÷ takipçi)2.2× audience3.77%58.7×
Etkileşim oranı0.32%1.65%0.19×

Bu aralıktaki diğer hesaplar →   Başka bir hesapla karşılaştır →   Bu kıyas değerleri nasıl kuruluyor →

Growth & engagement

How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.

Views per post

72.3K28 May
12.5K
10.6K29 May
107.4K4 Jun
2.3K
577
254.9K11 Jun
36.3K
6.7K
6.8K
34.1K19 Jun
26.6K22 Jun
29.2K19 Jul
25.8K

Last 14 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

0.26%28 May
0.25%
0.72%29 May
0.56%4 Jun
1.88%
2.95%
1.11%11 Jun
2.61%
2.07%
1.41%
0.06%19 Jun
0.51%22 Jun
0.13%19 Jul
0.65%

Reactions — likes, reposts, replies and quotes — divided by views. Median for 10K–100K accounts is 1.65%.

What the audience does

Likes77.5%4 778 in total
Reposts2.3%142 in total
Replies4.9%301 in total
Quotes1.7%102 in total
Bookmarks13.6%839 in total

Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.

The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.

Benzer hesaplar