Twitxy Saved

Phin Barnes

Verified account This account is verified. Learn more

@phineasb

San Francisco thegp.com Joined March 2008
2.1K Following 26K Followers 12K Posts

Latest posts 5

Phin Barnes Verified account @phineasb
The fact that larger models keep impriving when trained repeatedly on the same data with the right recipe means more compute can make existing data go further.

Great read from @industriaalist and the Qlabs team.
Samip Verified account @industriaalist Nov 2
Humans are ~1,000,000x more sample efficient than LLMs, and there are no good explanations.

We’ve found that a big part of the answer is that LLMs are *massively* underparameterized by many orders of magnitude. The brain has 100T synapses learning from ~200M tokens of language. Chinchilla prescribes a 1T-parameter model on 20T tokens. That’s 7 OOMs apart in parameters per token.

After a lot of painstaking work, we present new data-constrained scaling laws and learning algorithms to scale to (extrapolated) quadrillion-parameter transformers.

w/ @akshayvegesna, @EthanBaron78553, @asatheesh6561
Full paper led by @EthanBaron78553, with @ShmuelBerman, @andrewgwils
View quoted post on X
1 replies 2 reposts 13 likes 857 views View on X
Phin Barnes Verified account @phineasb
Great conversation and loved hearing @BigMekaStyle lay out his thesis for identifying excellence, investing early and with conviction and, most important, operating in alignment with his unique skills, interests and values.

Last time I heard something like this was when @nunzi46 told me about his vision for @CapitalAlso

Meka has every reason to be equally successful.
Erica Wenger Verified account @Erica_Wenger Oct 30
Bonus Friday Episode: My conversation with Meka Asonye (@BigMekaStyle).

Meka is the founder and managing partner of Prist Capital, an early-stage venture capital firm investing in deep tech startups. Previously, he worked at @firstround Capital, @stripe, and in baseball operations for the Cleveland Guardians (@CleGuardians).

We discuss how to identify the next billion-dollar startup, why the best venture capitalists think differently from everyone else, and what separates exceptional founders from the rest.

We also discuss:

- How his experience in Major League Baseball (@MLB) shaped his investment philosophy
- Why the best startup investments often look unconventional at first
- How he became an early investor in @K2SpaceCo
- Why mission-driven founders have an advantage over those motivated purely by money
- Why traditional investment committees can overlook the biggest opportunities

Enjoy!

Timestamps:
00:00 - Intro
00:52 - What Baseball Taught Him About Venture Capital
07:36 - Overcoming Personal Challenges and Finding Balance
12:06 - From Aviation to Investing in Deep Tech
19:59 - Why Meka Left First Round to Start His Own VC Firm
25:44 - Why Deep Tech Investing Requires a Different Strategy
30:41 - The #1 Thing to Look for When Investing in Startups
32:46 - How Meka Spotted the Opportunity in K2 Space
39:36 - The Future of Deep Tech and Emerging Startup Hubs
42:49 - Building the Future of Prist Capital
View quoted post on X