News for hackers. Stuff that matters.

celestiallylvd1

Re:Indian University Directory on Monday March 30, 2026 @08:33AM (#47573470)
Top-100 Indian Engineering Colleges of 2025 with placement statistics. Made this to keep track of the ROI in university education in an AI world
Re: on Tuesday May 20, 2025 @07:52PM (#44047057)
Is it really that good at reasoning or was the answer key in its training data?
Re:Ask HN: Is anybody building an alternative transformer? on Saturday February 15, 2025 @07:23AM (#43057994)
Yes, I am building a Perfect Language Model

Why is there no Perfect Language Model?

LLMs and humans talk in polysemantic languages. Polysemy is the property of words to have multiple meanings, depending upon context. I wish that we restructure society to adopt a monosemantic language and build AI to create a Perfect Language Model (PLM). But, earlier attempts to create monosemantic languages like LogLan and LojBan have failed. Why?

( Read More... | no comments | on HN )
Re:A word game I built to understand semantic distance on Saturday May 18, 2024 @02:12PM (#40400830)
Yes, you guessed it right! I am curious what the strike rate of an expert player could be, how it might compare with GPT-4, and if humans can outperform by discovering a strategy.
Re:A word game I built to understand semantic distance on Saturday May 18, 2024 @12:14PM (#40400102)
Cosine distance on GloVe-50 trained on Wikipedia. *GloVe is an unsupervised learning algorithm for obtaining vector representations for words. Training is performed on aggregated global word-word co-occurrence statistics from a corpus, and the resulting representations showcase interesting linear substructures of the word vector space.
Re:A Political Party for Researchers on Saturday October 14, 2023 @05:16PM (#37884340)
All over the world, people are tired of listening to political extempore. In this day and age, charisma and witty retorts are aplenty, but sustainable solutions are not. Thus, we are working towards an upgraded form of political outreach with a focus on research! Our goal is not to win elections in the near future, but instead to build a community of critical thinkers for our long-term future.

Some challenge areas where we need your help for research and outreach are 1) Optimizing the air quality in major cities 2) Disaster management for floods, earthquakes, fires, energy crisis etc. 3) Optimizing public-transportation networks in major cities 4) Preserving the soil quality and productivity of our farmlands

Our biggest challenges are outside identity politics - so let us unite and help each other!