yxchng

yxchng on Hacker News

Re:MusiCoT, a chain-of-thought (CoT) prompting technique for music gen... on Thursday April 10, 2025 @07:11AM (#43642737)

Mostly anecdotal and also based on the large gap between benchmark performance and actual performance on personal use

Re:MusiCoT, a chain-of-thought (CoT) prompting technique for music gen... on Thursday April 10, 2025 @03:35AM (#43641584)

This company has a track record of cheating on benchmark by training on test datasets. Take it with pinch of salt and try the model yourself.

Ask HN: Are you satisfied with the invoicing software you are using?

Posted by yxchng on Wednesday December 01, 2021 @11:15AM
from the ask-hn dept.

There are many invoicing softwares in the market e.g. Freshbooks, Quickbooks, Xero, etc. Just curious, is there anything you dislike or feel are lacking in these softwares, or is the problem already fully solved?

( Read More... | no comments | on HN )

Re:Stocks post best annual gain in 6 years with the S&P 500 surging mo... on Tuesday December 31, 2019 @07:10PM (#21925709)

Isn’t 2009-2019 gain much greater than 1999-2009 gain? What makes you think that the next 10 years will generate smaller return?

Re:Ant Design on Saturday October 05, 2019 @05:53PM (#21168218)

How does it compare to the more popular Material UI?

Re:Ask HN: Will getting a PhD lead to a more interesting life? on Monday September 30, 2019 @08:19AM (#21114476)

If you want to get cool jobs in AI, you do need PhD.(i.e. https://careers.google.com/jobs/results/6231594434232320-res..., https://www.quora.com/Does-the-AI-team-at-Facebook-hire-rese...). Those that don’t are just meaningless jobs in startups that pretend to do AI (git clone X, change dataset, train and pray that the machine learning god makes magic happen).

Though these meaningless jobs do pay well these days, thanks to Softbank.

Re:Gaussian Processes for Machine Learning (2010) on Monday July 23, 2018 @04:24AM (#17591312)

How big is the dataset you usually deal with? From my experience, you hit computational bottleneck pretty quickly with GPs (100k datapoints of maybe 100 dimensions is pretty much the max you can deal with).

And if you are talking about dataset of this scale, then I agree with you that GPs are better than NN. However, people are excited about NN capability of dealing with immensely huge and high dimensional dataset not small scale ones.

Re:Scanner: Processing Terabytes of Video on Hundreds of Machines on Tuesday May 22, 2018 @09:03AM (#17125571)

I am not an expert in video encoding. Just curious, what does Hwang offer that ffmpeg doesn't?