Grayson Adkins

1.3K posts

Grayson Adkins banner
Grayson Adkins

Grayson Adkins

@GraysonAdkins

Applied AI Scientist @GustoHQ • Previously Co-founder @Uffizzi_

Nashville, TN Katılım Ağustos 2010
1.9K Takip Edilen425 Takipçiler
Grayson Adkins
Grayson Adkins@GraysonAdkins·
@gobblergetterr You mean the guy who called recruits who de-committed from Vandy “not men of character” and then later persuaded recruits to de-commit from Vandy and follow him to Penn State?
English
2
0
4
2.6K
Samuel Rogers
Samuel Rogers@gobblergetterr·
In a world full of Sherron’s Moore’s, be a James Franklin. No scandals no nothing. Just a damn good football coach.
Samuel Rogers tweet media
English
35
4
166
1.6M
Grayson Adkins retweetledi
Jon Gruden
Jon Gruden@BarstoolGruden·
How I would fix the College Football Playoff.
English
1.1K
2.9K
19.7K
2.1M
Grayson Adkins retweetledi
François Chollet
François Chollet@fchollet·
The essence of great software is the quality of its abstractions. Great abstractions are indefinitely composable and stackable. They become a dependable foundation for future work, future thought. The best ones are so robust you never have to revisit them -- you can forget them and build forward without ever looking back. Achieving this is the greatest productivity hack in software engineering. Because the greatest productivity drain is backtracking.
English
69
249
2.2K
206.5K
Grayson Adkins
Grayson Adkins@GraysonAdkins·
The results show improvement over standard repeated sampling or sequential refinement on several benchmarks like LiveCodeBench and ARC-AGI, especially as compute budget increases. This is not the case for standard MCTS which plateaus. 7/8
English
1
0
0
48
Grayson Adkins
Grayson Adkins@GraysonAdkins·
Adaptive Branching MCTS: Very interesting paper here on an advanced reasoning strategy that outperforms what any one model can achieve. The core question is: how can we balance exploring new ideas (i.e. going wide) with refining a single idea (i.e. going deep)? 1/8
English
1
0
0
80
Grayson Adkins
Grayson Adkins@GraysonAdkins·
Used GPT-5 to analyze 8 insurance policy PFDs. Lightning fast parsing and extremely accurate analysis. Helped me think through the implications and make a decision. Perhaps not the most challenging task but was impressed with speed.
English
0
0
0
53
DHH
DHH@dhh·
One day I'm going to solve the mystery of why even the finest bakeries and hotels in America can't produce a Croissant on par with a Danish 7-11 or a Spanish gas station.
English
644
218
9.1K
972K
Kai Lentit (e/xcel)
Kai Lentit (e/xcel)@KaiLentit·
Interview with Senior DevOps engineer 2025 [FULL]
English
59
324
2.2K
185.7K
Grayson Adkins
Grayson Adkins@GraysonAdkins·
@clairlemon It often makes more financial sense for a wife not to work, given the cost of childcare. It *cost* our family $10k for my wife to work because of combo of taxes & daycare.
English
0
0
0
122
Claire Lehmann
Claire Lehmann@clairlemon·
Because they’re not real. Wives who don’t work have always existed and their decision not to work is driven by economic security not “traditional values”. It’s a luxury not a lifestyle.
English
1.4K
233
3.4K
1.2M
Grayson Adkins
Grayson Adkins@GraysonAdkins·
@amasad This Week In AI podcast/YT channel covers new releases with a lot of humor from a power users perspective. More of a practical lens and less theoretical.
English
0
0
0
13
Amjad Masad
Amjad Masad@amasad·
There is a need for an AI influencer that does honest in-depth real-use review of new models. Instead of the “this is insane, 10 ways how kimi k2 is changing the world.” Maybe they exist?
English
531
179
4.2K
793.5K
Grayson Adkins
Grayson Adkins@GraysonAdkins·
And anyone using Safari or Edge don’t care enough to bother downloading something new
English
0
0
0
43
Grayson Adkins
Grayson Adkins@GraysonAdkins·
I think it’s unlikely everyone switches to AI browsers before Chrome just copies all their features. People are underestimating the power of distribution
English
1
0
0
64