Search | nitter

Oct 26

Spending lots of times these days thinking about human eval curves

3

1

67

8,411

Oct 26

Always the case. Attention is a surprisingly powerful and useful inductive bias.

But there are more…

mike
@mike64_t
Oct 25
sometimes the only thing that's missing is the right inductive bias in one's architecture...

7

7

302

34,614

Oct 25

Two sentence horror:

For rent: GB300s, never used.

Yuchen Jin
@Yuchenj_UW
Oct 24
Heard from a neolab friend:

“Even if we fail to train the best model, we locked in GPU capacity early. H100/B200/B300 rental prices have since gone up 2–3x. We can just resell the compute and print money.”

OK, so that’s the business model.

19

22

956

67,678

Oct 24

every network has a story

3

1

41

6,407

Oct 17

It’s amazing you can clearly tell exact moment in history when each lab started vibecodong their product

19

3

555

30,380

Oct 17

The interesting thing is that at this moment those are in the training data

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
Oct 17
Gary Marcus, comments??

3

1

109

8,998

Oct 15

Replying to @joannejang

@joannejang hi, may I meet you?

1

0

31

1,686

Oct 12

Generalization in 2026: we trained a chonkers on the energy consumption of London and it is good in some things

Jiayi Pan @jiayi_pirate
Oct 12
A modern perspective on generalization

0

4

127

8,381

Oct 9

9

27

521

17,409

Oct 9

Replying to @sohocine

@sohocine Not impossible

1

0

19

1,893

Oct 9

Will replacing the transformer be a revolution or an evolution?

43

9

316

29,927

Oct 8

Replying to @industriaalist

@industriaalist @coreauto @alp_vvv @_arohan_ You wouldn't believe it 😂

1

0

13

948

Oct 3

Playlist this week:
- upping my p(doom)
- chasing the roofline
- core auto thesis (internal)

8

0

114

9,250

Oct 2

Replying to @sethbannon

@sethbannon Its power of friendship

2

0

8

681

Oct 2

Replying to @brittwalker_

@brittwalker_ 👀 👀 👀

0

0

6

449

Oct 2

Replying to @matthieuschulz

@matthieuschulz 👀

0

0

10

630

Oct 2

It is our responsibility to make reality interesting, as otherwise humans will rationally choose to live in a simulation

20

12

208

11,646

Sep 30

When I grow up I want to be a slopcore artist

8

2

89

13,918

Sep 29

Machine learning isn't magic - its assumptions in disguise

5

2

128

7,441

Sep 24

Big model smell lasts just a brief moment
Small model smell lasts forever

12

3

250

9,965