Swaroop Mishra
@swarooprm7.bsky.social
Reposted by Swaroop Mishra
🚨 Reverse Thinking Makes LLMs Stronger Reasoners
We can often reason from a problem to a solution and also in reverse to enhance our overall reasoning. RevThink shows that LLMs can also benefit from reverse thinking 👉 13.53% gains + sample efficiency + strong generalization (on 4 OOD datasets)!
We can often reason from a problem to a solution and also in reverse to enhance our overall reasoning. RevThink shows that LLMs can also benefit from reverse thinking 👉 13.53% gains + sample efficiency + strong generalization (on 4 OOD datasets)!
December 2, 2024 at 7:29 PM
🚨 Reverse Thinking Makes LLMs Stronger Reasoners
We can often reason from a problem to a solution and also in reverse to enhance our overall reasoning. RevThink shows that LLMs can also benefit from reverse thinking 👉 13.53% gains + sample efficiency + strong generalization (on 4 OOD datasets)!
We can often reason from a problem to a solution and also in reverse to enhance our overall reasoning. RevThink shows that LLMs can also benefit from reverse thinking 👉 13.53% gains + sample efficiency + strong generalization (on 4 OOD datasets)!
The default feed needs to be significantly improved so that user don't have to customize. This will increase user retention @bsky.app
I found this post to customize different kinds of feeds! (Mutuals, Mentions, Followers etc)
bsky.app/profile/jay....
bsky.app/profile/jay....
6. Custom feeds on Bluesky are the first implementation of algorithmic choice. Instead of using a single black-box For You algorithm, you can create and subscribe to your own.
Now, there are over 50k+ custom feeds on the network! Check them out:
bsky.app/feeds
Now, there are over 50k+ custom feeds on the network! Check them out:
bsky.app/feeds
November 25, 2024 at 2:11 AM
The default feed needs to be significantly improved so that user don't have to customize. This will increase user retention @bsky.app
Nice to see the LLM research community here.
Let's see if this can stand the test of time.
Let's see if this can stand the test of time.
November 22, 2024 at 5:12 PM
Nice to see the LLM research community here.
Let's see if this can stand the test of time.
Let's see if this can stand the test of time.
I see a lot of irrelevant posts in my timeline. Recommendation algorithm needs to change probably?
November 20, 2024 at 5:58 AM
I see a lot of irrelevant posts in my timeline. Recommendation algorithm needs to change probably?
My first post here.
Had a great trip to MIT.
Had a great trip to MIT.
November 17, 2024 at 8:03 PM
My first post here.
Had a great trip to MIT.
Had a great trip to MIT.
Hi Everyone! This is my first post.
AI people: Say 'Hi'!
AI people: Say 'Hi'!
November 17, 2024 at 5:34 PM
Hi Everyone! This is my first post.
AI people: Say 'Hi'!
AI people: Say 'Hi'!