Jacob X. Li ✓

@jacobli99

PhD Student @MIT (EECS & CSAIL & @nlp_mit) advised by @lateinteraction. Prev. @BrownUniversity

3 594Followers

235Following

669Posts total

251.9KViews on collected posts

โพสต์ล่าสุด

Jacob X. Li ✓@jacobli9917 Jun

@jacobli99 putting the link here for those that want to jump right into the long form: https://t.co/g2OGSnxRa0 7.3K views · 114 likes · 8 reposts · 3 replies Open on X →

Jacob X. Li ✓@jacobli9917 Jun

Thanks to my great co-authors: Rick Battle, @lateinteraction Post (with links to the dataset!): https://t.co/t7CqIaxJlY 660 views · 15 likes · 0 reposts · 1 replies Open on X →

Jacob X. Li ✓@jacobli9917 Jun

So how should an agent actually study? We test the simplest version of three natural bets out there: (1) self-supervised objectives, e.g. continual pre-training over the corpus; (2) synthesizing your own training data, e.g. turning the corpus into Q&A pairs to fine-tune on; and 3.9K views · 60 likes · 1 reposts · 3 replies Open on X →

Jacob X. Li ✓@jacobli9917 Jun

We instantiate this in StudyBench, a benchmark we’re building so that we and others can begin to ask questions about agents' ability to study. StudyBench consists of three tasks, each built on a corpus that defines a domain of expertise, and each paired with a hidden exam. Below 4.1K views · 34 likes · 1 reposts · 1 replies Open on X →

Jacob X. Li ✓@jacobli9917 Jun

Studying gives us a second curve: expertise as a function of study compute. You could consider its weighted area a notion of “intelligence”. An intelligent agent, for our purposes, is one that can acquire expertise in totally new domains really efficiently. (And by this token, ht 11.2K views · 47 likes · 6 reposts · 2 replies Open on X →

Jacob X. Li ✓@jacobli9917 Jun

To compare procedures for machine studying, we start by defining expertise. The corpus is always available at test time anyway, so a sufficiently intelligent non-expert agent could in principle always study during the exam. What distinguishes an expert from a smart novice is a 8.4K views · 52 likes · 5 reposts · 1 replies Open on X →

Jacob X. Li ✓@jacobli9917 Jun

Humans face this problem of learning new domains constantly, and one of our default answers is *studying*. Before an exam, even an open-book one, we read the textbook or the literature, think out loud, quiz ourselves, and write our own notes. Much of the expertise comes from the 7.1K views · 47 likes · 2 reposts · 1 replies Open on X →

Jacob X. Li ✓@jacobli9917 Jun

Continual learning is widely discussed right now, but mostly as improving on the job or avoiding catastrophic forgetting. But it has a different, difficult, and already urgent form: Given nothing but a corpus of documents, how should AI systems develop expertise in a new, https: 209.3K views · 752 likes · 119 reposts · 17 replies Open on X →

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

209.3K17 Jun

7.1K

8.4K

11.2K

4.1K

3.9K

660

7.3K

Last 8 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

0.44%17 Jun

0.72%

0.69%

0.50%

0.87%

1.65%

2.42%

1.72%

Reactions — likes, reposts, replies and quotes — divided by views.

What the audience does

Likes48.4%1 121 in total

Reposts6.1%142 in total

Replies1.3%29 in total

Quotes1.1%26 in total

Bookmarks43.1%999 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.

โพสต์ล่าสุด

Growth & engagement

Views per post

Engagement rate per post

What the audience does

บัญชีที่คล้ายกัน