Horace He ✓

@cHHillee · chhillee · joined 24 Feb 2010

@thinkymachines Formerly @PyTorch "My learning style is Horace twitter threads" - @typedfemale

53 397Followers

612Following

3 650Posts total

1.1MViews on collected posts

Postingan terbaru

Horace He ✓@cHHillee10 Sep

Retweeting this for no particular reason 26.8K views · 313 likes · 11 reposts · 8 replies Open on X →

Horace He ✓@cHHillee5 Sep

Everyone knows that labs can benchmaxx: optimize the model to look good on the leaderboard. Sometimes forgotten is that benchmarks can benchmaxx-maxx: optimize the benchmark to make its leaderboard look good. 81.5K views · 1K likes · 41 reposts · 32 replies Open on X →

Horace He ✓@cHHillee24 Aug

I've realized when people are talking about how many tokens they use, they're usually including cached input tokens... which is incredibly dumb... 47K views · 606 likes · 10 reposts · 40 replies Open on X →

Horace He ✓@cHHillee19 Aug

Presented without comment. https://t.co/Ob7Bvnsmpy

31.6K views · 228 likes · 9 reposts · 12 replies Open on X →

Horace He ✓@cHHillee18 Aug

Btw, numpy is still using the N**3 method. "In theory, there is no difference between theory and practice. In practice, there is." 142.6K views · 1K likes · 27 reposts · 13 replies Open on X →

Horace He ✓@cHHillee29 Jul

Presenting my grand unified theory of ML researcher impact: Your impact is directly proportional to how much pain you cause to infra. Fundamentally, you can only inflict pain upon infra if your approach actually works. And the better your approach works the more pain infra is 231.5K views · 1.8K likes · 110 reposts · 67 replies Open on X →

Horace He ✓@cHHillee10 Sep 2025

Apologies that I haven't written anything since joining Thinking Machines but I hope this blog post on a topic very near and dear to my heart (reproducible floating point numerics in LLM inference) will make up for it! 560.8K views · 2.9K likes · 176 reposts · 82 replies Open on X →

Dibandingkan akun berukuran sama

6 postingan dari 90 hari terakhir, dibandingkan dengan rentang 10K–100K pengikut. tampil luas, tetapi sedikit penonton yang merespons.

Median tayangan64 294akun ini924median untuk 10K–100K

Jangkauan, %120.41%akun ini3.62%median untuk 10K–100K

Interaksi, %1.05%akun ini1.52%median untuk 10K–100K

Metrik	Akun ini	Median untuk 10K–100K	Rasio
Median tayangan per postingan	64 294	924	69.6×
Jangkauan (tayangan ÷ pengikut)	120.41%	3.62%	33.3×
Tingkat interaksi	1.05%	1.52%	0.69×

Akun lain pada rentang ini → Bandingkan dengan akun lain → Bagaimana tolok ukur ini disusun →

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

560.8K10 Sep

231.5K29 Jul

142.6K18 Aug

31.6K19 Aug

47K24 Aug

81.5K5 Sep

26.8K10 Sep

Last 7 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

0.55%10 Sep

0.87%29 Jul

0.75%18 Aug

0.79%19 Aug

1.39%24 Aug

1.32%5 Sep

1.24%10 Sep

Reactions — likes, reposts, replies and quotes — divided by views. Median for 10K–100K accounts is 1.52%.

What the audience does

Likes79.8%7 860 in total

Reposts3.9%384 in total

Replies2.6%254 in total

Bookmarks13.7%1 347 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.