Finding the Words to Say: Hidden State Visualizations for Language Models
https://t.co/hOot6gkhDf
New post! Visualizations glancing at the "thought process" of language models & how it evolves between layers. Builds on awesome work by @nostalgebraist@lena_voita@tallinzen. 1/n
I am looking for a postdoc (or, possibly, a pre-doc) in Arabic NLP, syntactic and/or semantic parsing, modeling of sentiment and belief, and/or analyzing written conversations (covers about 85% of NLP 😀). https://t.co/TGgV3Wk3jp
Release alert: the 🤗datasets library v1.2 is available now!
With:
- 611 datasets you can download in one line of python
- 467 languages covered, 99 with at least 10 datasets
- efficient pre-processing to free you from memory constraints
Try it out at:
https://t.co/A7rmjGSELy
Here's the 2020 release of draft chapters for Speech and Language Processing, just in time to wish you all a Happy New Year! Enjoy! https://t.co/pjfZsYfJ5S
"To deal with hyper-planes in a 14-dimensional space, visualize a 3-D space and say ‘fourteen’ to yourself very loudly. Everyone does it"
~ Geoffrey Hinton
إنك يا أبا القاسم حين تُمدح فإن الشعر لا ينتهى بانتهاء الإنشاد؛ وذلك لأنك «من أنفسنا»، ولأنه كان «عزيزًا عليك ما عَنِتْنَا» ولأنك كنت «حَرِيصًا علينا»، ولأنك كنت فَأْلَ هذه الأمة الحَسَن، فكنت متعبًا مثلنا، ومظلومًا مثلنا، ومنفيًا مثلنا، ومُكَذَّبًا ومكذوبًا عليك مثلنا، ثم انتصرت
Just because it is written by a Harvard professor doesn't mean that you don't need to check the references!
So, I was reading this interesting line by Shady Hekmat about Qunbul's character, which reflects negatively on his role as a reciter of the Qurʾān.
1/7
Folks who work on #NLProc for disaster response: you can't use Twitter data that encodes health information about people. That includes the existing government surveillance datasets like "CrisisNLP". This is now banned by Twitter's terms and ACM's code of ethics.