Top all-time · #10446
fast.ai NLP
Company blog · nlp.fast.aiaiprogramming
fast.ai NLP is the company blog at nlp.fast.ai, writing about ai and programming. Since 2018, 7 different Hacker News readers have linked it 10 times, most often “State-of-the-art text classification with universal language models”. The last post was in 2019, but the archive is still widely read.
Too few sampled comments (8) to say how often HN recommends it.?
Why HN links it
HN readers link this fast.ai blog for posts on language-model transfer learning for text classification, such as ULMFiT and MultiFiT. Commenters also point out broken links on the site.
Summarised by Claude from a sample of 8 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Most-linked posts?
- State-of-the-art text classification with universal language models3 people · 5 mentions · first linked 2018-05-15
- /category/classification.html3 people · 3 mentions · first linked 2018-05-15
What Hacker News says?
I'm looking forward to testing this out! I noticed the "NLP classification page" links are broken. I assume they should go to: http://nlp.fast.ai/category/classification.html
Comment linking /category/classification.html
on “State-of-the-art text classification with universal language models”Submitted State-of-the-art text classification with universal language models as a story · 346 points
HN discussioncheck out MultiFiT [0] from fastai, it uses QRNN for speed. [0] http://nlp.fast.ai/classification/2019/09/10/multifit.html
Comment linking /classification/2019/09/10/multifit.html
on “Single Headed Attention RNN”Submitted Introducing state of the art text classification with universal language models as a story · 4 points
HN discussionThe ULMFiT, ELMO, and OpenAI Transformer papers are all quite readable and linked from the article. Sebastian and I also wrote an introduction to ULMFiT here: http://nlp.fast.ai/classification/2018/05/15/introducting-ul...
Comment linking /classification/2018/05/15/introducting-ulmfit.html
on “NLP's ImageNet Moment: From Shallow to Deep Pre-Training”You can find ELMo here: https://github.com/allenai/allennlp/blob/master/tutorials/ho... And ULMFiT here: http://nlp.fast.ai/category/classification.html
Comment linking /category/classification.html
on “NLP's ImageNet Moment: From Shallow to Deep Pre-Training”No mention of ULMFiT? http://nlp.fast.ai/classification/2018/05/15/introducting-ul...
Comment linking /classification/2018/05/15/introducting-ulmfit.html
on “What Kagglers Are Using for Text Classification”This paper is really important: it shows that transfer learning can be applied to a wide variety of NLP problems with great success. They show state of the art results on nearly every major class of NLP problem. The basic approach is the same as our ULMFiT ( http://nlp.fast.ai/classification/2018/05/15/introducting-ul... ) model - pre-train a language model (a model that predicts the next word in a sequence) on a la…
Comment linking /classification/2018/05/15/introducting-ulmfit.html
on “Improving Language Understanding with Unsupervised Learning”Definitely not true. Transfer learning a model trained on a corpus then fine tuned on labeled classes is current SotA approach [1]. [1] http://nlp.fast.ai/category/classification.html
Comment linking /category/classification.html
on “Practical Text Classification with Python and Keras”They will need to add machine learning based classification of mails before sending them. Spammers will just create the next account until they get detected again. Their servers will be faster on a blacklist than they can spell Spamhaus. I recommend ULMFiT [1]. [1]: http://nlp.fast.ai/classification/2018/05/15/introducting-ul...
Comment linking /classification/2018/05/15/introducting-ulmfit.html
on “Show HN: SESS – Simple Email Sending Service”