paraschopra

12 ms·

61.
Sequential Scaling Beats Majority Voting at Matched Compute(arxiv.org)
2 points by paraschopra 1y ago|0 comments
62.
Do LLMs know when they've gotten a correct answer?(letters.lossfunk.com)
2 points by paraschopra 1y ago|0 comments
63.
How do LLMs "think" across languages(letters.lossfunk.com)
2 points by paraschopra 1y ago|0 comments
64.
What is research and how to do it?(letters.lossfunk.com)
1 points by paraschopra 1y ago|1 comments
65.
Practical tips on building LLM agents(letters.lossfunk.com)
1 points by paraschopra 1y ago|0 comments
66.
Manifesto for doing good science in AI(letters.lossfunk.com)
1 points by paraschopra 1y ago|0 comments
67.
From an idea to an ML paper in under 1 hour(letters.lossfunk.com)
1 points by paraschopra 1y ago|0 comments
68.
Reinforcement Learning Algorithms Summarized(lossfunk.substack.com)
1 points by paraschopra 1y ago|0 comments
69.
by paraschopra 1y ago
Humans actually do both. We learn from on-policy by exploring consequences of our own behavior. But we also learn off-policy, say from expert demonstrations (but difference being we can tell good behaviors from bad, and learn from a filtere
70.
Life as a Physical Process(invertedpassion.com)
2 points by paraschopra 1y ago|0 comments
71.
by paraschopra 1y ago
Aren't manifolds generally task-dependent? I've been debating whether the data lies on a manifold, or whether the data attributes that are task-relevant (and of our interest) lie on a manifold? I suspect it is the latter, but I&#x
72.
Do you know what GDP is?(invertedpassion.com)
1 points by paraschopra 1y ago|0 comments
73.
by paraschopra 2y ago
Is it fair to say that both "Say 'an'" and "Say 'astronomer'" output features would be present in this case, but say "Say 'an'" gets more votes because it is start of the sentence,
74.
by paraschopra 2y ago
LLMs have induction heads that store such names as sort of variables and copy them around for further processing. If you think about it, copying information from inputs and manipulating them is a much more sensible approach v/s memoriz
75.
Don't compete(invertedpassion.com)
41 points by paraschopra 2y ago|15 comments
76.
by paraschopra 2y ago
Thanks! Vijaye
77.
by paraschopra 2y ago
Thank you!
78.
by paraschopra 2y ago
Sorry about that! We now have server side testing option where A/B tests trigger from the backend.
79.
by paraschopra 2y ago
Thank you!
80.
by paraschopra 2y ago
Comparison with PG's essays is a great honor. His essays are what inspired me to go down the startup journey myself.
81.
by paraschopra 2y ago
Thanks Spencer! You've built an amazing product, we're hugely inspired by Amplitude.
82.
by paraschopra 2y ago
Yep, but over time they make you stoic and that stoicism helps outside of startups too.
83.
by paraschopra 2y ago
i owe him!
84.
by paraschopra 2y ago
Yes, they did.
85.
by paraschopra 2y ago
It's not a dumb question. But what I learned in this process is that companies are valued after incorporating many factors (liquidity, earnouts, future involvement, growth rate, profit, retention, current market mood, etc. etc.) and mu
86.
by paraschopra 2y ago
Thanks! I've updated them and they'll be published as a book (+ released online) this year.
87.
by paraschopra 2y ago
We were always profitable and growing at a sustainable pace, so never needed funding. Most counter intuitive thing was to not think I was missing out on anything as we didn't have VCs backing us.
88.
by paraschopra 2y ago
Wow, i vaguely recall using it!
89.
by paraschopra 2y ago
the key was my naivety as a 22 year old. I honestly didn't know any of the reasons why it couldn't work!
90.
by paraschopra 2y ago
why?

More ›