Iain Harlow's Avatar

Iain Harlow

@iainharlow.bsky.social

Chief AI Officer at alaistudios.com

94 Followers  |  712 Following  |  2 Posts  |  Joined: 18.08.2023
Posts Following

Posts by Iain Harlow (@iainharlow.bsky.social)

Just found out that the voice actor that recorded the line 'you know when you've been Tangoed' for the adverts was GIL SCOTT FUCKING HERON, and now I need a lie down

11.08.2024 19:44 β€” πŸ‘ 231    πŸ” 81    πŸ’¬ 29    πŸ“Œ 26

Hi. I was away for a week. Hope everyone is doing well.

There are people who believe the reasons for my absence are shameful and disqualifying for full participation on public life.

Those people can snort my taint.

If you need help, get help. Don’t wait to be in crisis. You can feel better.
/1

25.07.2024 16:41 β€” πŸ‘ 4530    πŸ” 499    πŸ’¬ 257    πŸ“Œ 157

Really rewarding paper to read, pressuring the norm of uncritically using cosine similarity as a proxy for semantic similarity.

Rewarding, but also makes me sweat a little.

arxiv.org/pdf/2403.054...

11.03.2024 17:52 β€” πŸ‘ 25    πŸ” 5    πŸ’¬ 3    πŸ“Œ 1

Hello newskies! (And oldskies too)
Bluesky doesn’t have just one algorithm, it has many, and you can pick what you like (or write your own). Here’s a few I made and a few i like:

Quiet Posters, posts from people you follow who don’t post that often whose posts you might normally miss

06.02.2024 18:13 β€” πŸ‘ 1796    πŸ” 718    πŸ’¬ 53    πŸ“Œ 75

Huh, this is me and my kids but with the roles exactly reversed

28.01.2024 05:13 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 0    πŸ“Œ 1
Preview
Can language models predict the next twist in a story? While distant reading has taught us a lot about the history of fiction, it hasn't done much yet to explain why we keep turning pages. "Suspense" is the word we use to explain that impulse. But what is...

A preliminary experiment with a new method of measuring a story's "predictability." Code and data are shared so other people can tinker with this. tedunderwood.com/2024/01/05/c...

05.01.2024 21:25 β€” πŸ‘ 90    πŸ” 20    πŸ’¬ 3    πŸ“Œ 2
Preview
Humans and language models diverge when predicting repeating text Language models that are trained on the next-word prediction task have been shown to accurately model human behavior in word prediction and reading speed. In contrast with these findings, we...

New from Aditya Vaidya arxiv.org/abs/2310.06408 Many have shown that LM and human predictions are very correlated, but we found that's not true when the text repeats! This points to divergence btwn memory mechanisms, which we localize in the model and then modify to get more human-like behavior.

11.10.2023 15:57 β€” πŸ‘ 25    πŸ” 7    πŸ’¬ 0    πŸ“Œ 1

Anyone have somebody in their network who might be interested in working on some data science, mapping and computer vision problems? Come across an interesting opportunity with some smart folks in commercial real estate. Put me in touch if so!

10.10.2023 18:33 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0