back
user profile

porridgeraisin

1,976karma·1,374submissions·March 26, 2023
recent activity (1,374 total)
comment
For context: In the final stages of the project, AI was used to assist: Implementing the user-facing API Widening of compatibility, for instance generalizing and porting the pretokenizer implementatio…
24d ago·view thread
comment
Yes, when you use a standard theoretical physics textbook / course sequence, it is 100% correct. More broadly, LLMs or anything at all, needs a verifier. If the task can be automatically verified…
24d ago·view thread
comment
But this is well established behaviour in the real world. That the passkey design doesn't take this into account only proves even more that it was designed in a vacuum.
24d ago·view thread
comment
This is the most relevant comment in this thread.
26d ago·view thread
comment
This is the way. Works very well.
26d ago·view thread
comment
bound check elimination exists separately, PGO can help unveil more opportunities for BCE.
27d ago·view thread
comment
From my skim of their paper, they have gotten robodojo 14% versus previous sota 9%. So its an improvement, but doesnt seem like a reliable model yet, which is fine. I also notice the model becomes wor…
27d ago·view thread
comment
How is it much more expensive in europe? cos in india where we import most of our ONG and its one of the more expensive things on a household bill, it is right now 102INR/L = 0.93 EUR/L.
27d ago·view thread
comment
Insofar as PGO helps inlining.
27d ago·view thread
comment
https://xcancel.com/synthwavedd/status/2078514339552628880
29d ago·view thread
comment
Yes. The dhammapada uses a lot of Anushthubh chandas, which is an option in the website. However, the language differences between pali and sanskrit are enough to make TTS trip up. E.g only one "…
29d ago·view thread
comment
Yes, but monte carlo estimating the critic model is not new.
1mo ago·view thread
comment
GRPO is policy gradient/PPO with your value function baseline monte carlo estimated using k rollouts. The only new thing is finding out it works well with binary rewards and LLM policies.
1mo ago·view thread
comment
It's just another way to frame it. It's as foundational as the many other ways to frame it. I'm not aware of any major insight you get specifically from this framing. Is there one?
1mo ago·view thread
comment
https://x.com/eliebakouch/status/2077428860639973471 > gpqa diamond alone makes up 20% of their "capability index" and is responsible for ~70% of the gap betwee…
1mo ago·view thread
comment
Yep, I gave up on them for now. One of my older laptops also needs a charger that has the third ground pin, lest the trackpad become laggy due to electrical joise. I couldn't find a compact gan c…
1mo ago·view thread
comment
Gan chargers also die really fast.
1mo ago·view thread
comment
But this particular instance isn't much related to that. In my school too, we did all that, and extending to beach cleanups. It's not like the same effect was there. Besides, most garbage in…
1mo ago·view thread
comment
This particular lab happens to be primarily privately funded, donations, grants, etc,. Most big ticket research at IITM like this one are industry-funded or donation-funded. The public funds cover all…
1mo ago·view thread
comment
Other publications from the same place: https://iitm.humanbrain.in/project.html notably: https://brainportal.humanbrain.in/publicview/index.html …
1mo ago·view thread