Top all-time · #9250
Abigail See
Personal blog · abigailsee.comai
Abigail See's personal blog at abigailsee.com covers ai. Since 2017, 8 different Hacker News readers have linked it 9 times, most often “Top Deep Learning Trends in Natural Language Processing”. The last post was in 2019, but the archive is still widely read.
Too few sampled comments (4) to say how often HN recommends it.?
Why HN links it
HN commenters cite Abigail See's posts on neural language models, including a 2019 post on what makes a good conversation, as support for points about LSTMs and NLP. No comment judges the writing directly.
Summarised by Claude from a sample of 4 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
A blog about artificial intelligence, deep learning and natural language processing
Most-linked posts?
- Top Deep Learning Trends in Natural Language Processing2 people · 3 mentions · first linked 2017-09-01
- Taming Recurrent Neural Networks for Better Summarization2 people · 2 mentions · first linked 2017-04-18
- Yann LeCun vs. Christopher Manning on Need for Priors in Deep Learning2 people · 2 mentions · first linked 2018-02-22
- What Makes a Good Conversation?2 people · 2 mentions · first linked 2019-08-24
What Hacker News says?
I think the most interesting and relevant strand of philosophy of mind is the connectionism of Rumelhart, McClelland and the PDP Research Group (including "godfather of deep learning" Geoff Hinton), exemplified in the tome "Parallel Distributed Processing" released in 1986 It even has an accompanying manual on how to implement artificial neural networks. http://www.cnbc.cmu.edu/~plaut/IntroPDP/index.html More along…
Comment linking /2018/02/21/deep-learning-structure-and-innate-priors.html
on “The Best Books on the Philosophy of Mind”Submitted What Makes a Good Conversation? as a story · 207 points
HN discussionI was going to mention XLNet before I saw your comment. Also, a recent piece of interesting work [1] shows that with the right control parameters, you could still use gated RNNs, like LSTMs, for pretty good language modeling. [1] http://www.abigailsee.com/2019/08/13/what-makes-a-good-conve...
Comment linking /2019/08/13/what-makes-a-good-conversation.html
on “Natural Language Processing: The Age of Transformers”Submitted Yann LeCun vs. Christopher Manning on Need for Priors in Deep Learning as a story · 150 points
HN discussionAll arguments presented show that FORMAL methods won't work for a general NLP. ALL of the problems discussed in this post are solved to some extent by machine learning - sometimes even by statistic methods such as word2vec. GPT-3 goes much further. Of course, these models are not trained on pure, grammatically and conceptually correct English - but on language that people use. There are fundamental issues with NLP b…
Comment linking /2017/08/30/four-deep-learning-trends-from-acl-2017-part-1.html
on “Machine learning won't solve natural language understanding”Submitted Taming Recurrent Neural Networks for Better Summarization as a story · 135 points
HN discussionI started with a combative tone, so well - no apologies needed. Well, still - current translation systems are data-driven, without exception, vide http://norvig.com/chomsky.html . And LSTMs are awesome at picking grammar, even not one considered English grammar (line braking patterns, proper names, markup for links, etc). Vide http://karpathy.github.io/2015/05/21/rnn-effectiveness/ There are other issues like keepin…
Comment linking /2017/08/30/four-deep-learning-trends-from-acl-2017-part-1.html
on “How should we evaluate progress in AI?”Submitted Top Deep Learning Trends in Natural Language Processing as a story · 4 points
HN discussionSubmitted Taming Recurrent Neural Networks for Better Summarization as a story · 3 points
HN discussion