LogoThread Easy
  • 発見
  • スレッド作成
LogoThread Easy

Twitter スレッドの万能パートナー

© 2025 Thread Easy All Rights Reserved.

探索

Newest first — browse tweet threads

Keep on to blur preview images; turn off to show them clearly

OpenAI has historically scaled up training compute by around 100x with each new generation of its GPT.

However, GPT-5 appears to be an exception to this trend.

🧵

OpenAI has historically scaled up training compute by around 100x with each new generation of its GPT. However, GPT-5 appears to be an exception to this trend. 🧵

GPT-4 was trained on 2e25 floating-point operations, and OpenAI said GPT-4.5 was about an order-of-magnitude (10x) scale-up. We don’t have a rigorous estimate yet, but GPT-5’s compute scale may be *between* GPT-4 and GPT-4.5, and it is probably not a large scale-up from 4.5.

avatar for Epoch AI
Epoch AI
Fri Aug 08 18:18:49
Burnout isn’t stress.

It started when you were praised for holding it all together.

That praise became identity. That identity became oxygen.

Now you can’t stop because you don’t know who you’d be without it.

Here’s what no one told you: 🧵

Burnout isn’t stress. It started when you were praised for holding it all together. That praise became identity. That identity became oxygen. Now you can’t stop because you don’t know who you’d be without it. Here’s what no one told you: 🧵

Act I – The Self Some people learn the rules. Others learn to hold the system together. From a young age, you fell into the second group.

avatar for Sosa | Mental Strategist
Sosa | Mental Strategist
Fri Aug 08 13:30:25
10) Still Undervalued

Despite running more than 200% in the last few months, $SOFI — believe it or not — is still undervalued.

Case in point. $SOFI currently trades at a Forward PE of about 74x, but that's based on 2025 EPS estimate of $0.30, which is below management’s revised guidance of $0.31.

In addition, analysts estimate 2026 EPS to be $0.54, which is also below management's guidance of $0.55 to $0.80.

Analysts forget that management likes to not only sandbag their guidance but also destroy it in the process. 

That said, I believe $SOFI will achieve an EPS of $1.00 in 2026, valuing SOFI at a 2026 Forward PE of about 21x today.

Now that looks like a good valuation to pay.

Some catalysts to consider include:
- new deals for the Loan Platform Business
- acquisitions
- relaunch of crypto offerings
- interest rate cuts

10) Still Undervalued Despite running more than 200% in the last few months, $SOFI — believe it or not — is still undervalued. Case in point. $SOFI currently trades at a Forward PE of about 74x, but that's based on 2025 EPS estimate of $0.30, which is below management’s revised guidance of $0.31. In addition, analysts estimate 2026 EPS to be $0.54, which is also below management's guidance of $0.55 to $0.80. Analysts forget that management likes to not only sandbag their guidance but also destroy it in the process. That said, I believe $SOFI will achieve an EPS of $1.00 in 2026, valuing SOFI at a 2026 Forward PE of about 21x today. Now that looks like a good valuation to pay. Some catalysts to consider include: - new deals for the Loan Platform Business - acquisitions - relaunch of crypto offerings - interest rate cuts

That's a wrap! If you would like to read my full investment thesis on $SOFI, do check out my deep dive article here: https://t.co/H7iuGNqovQ These posts take a lot of time and energy to write, so I would appreciate a like, share, and subscribe. Thank you for your support!

avatar for Riyado Sofian
Riyado Sofian
Fri Aug 01 19:26:54
@astridwilde1 dont have a meme to post for $NET yet (later today) but i really want to tweet the word 'stonks' so: stonks

@astridwilde1 dont have a meme to post for $NET yet (later today) but i really want to tweet the word 'stonks' so: stonks

@astridwilde1 net $562M, 3.19% beat, god bless. it is the new application layer of the internet but instead of being on the OSI model its a ticker on the nasdaq

avatar for near
near
Thu Jul 31 20:54:39
this seems really important:

it is totally plausible that a model could get IMO gold without *any* reinforcement learning, given a perfectly-crafted prompt

we just don't know, and lack tools to efficiently search through prompt space.    glad to see at least someone is trying

this seems really important: it is totally plausible that a model could get IMO gold without *any* reinforcement learning, given a perfectly-crafted prompt we just don't know, and lack tools to efficiently search through prompt space. glad to see at least someone is trying

research @cornell // language models, information theory, science of AI

avatar for dr. jack morris
dr. jack morris
Thu Jul 31 01:24:24
I didn’t question who was paying you, I asked how much are you being paid!?

I didn’t question who was paying you, I asked how much are you being paid!?

Adelaide. Auspol Content Creator. I usually vote Labor. Tampons should be free. Your Dad argues with me on the internet. Follow me on Insta & TikTok 👇🏻

avatar for Belinduh Pyne
Belinduh Pyne
Tue Jul 29 11:38:42
  • Previous
  • 1
  • More pages
  • 4159
  • 4160
  • 4161
  • More pages
  • 4204
  • Next