New NLP News: Bigger vs. smaller models, powerful vs. dumb models newsletter.ruder.io/archive/190768 via @revue
10. How to Win a Data Science Competition: Learn from Top Kagglers by Coursera Time to head over to Kaggle to get some experiences building a machine learning for your resume and make some $$$ coursera.org/learn/competit…
Here's a summary post on problems with huge models that dominate #NLProc these days. I put together several different discussion threads with/by @yoavgo, @jaseweston, @sleepinyourhat, @bkbrd, @alex_conneau, @SeeTedTalk. hackingsemantics.xyz/2019/leaderboa…
This is a bit misleading since the article chooses to ignore ULMFiT entirely which takes 5-6 hours to pretrain on a single (recent) GPU so $18 if you go for an AWS p3. It's not SOTA on downstream tasks in English (but still fares well) but is in pretty much all other languages.
Extremely excited to share work I've been doing at OpenAI the past few months: MuseNet, a neural net music generator. It's been a huge team effort pulling this all together!
Introducing MuseNet, a neural network which discovered how to generate music using many different instruments and styles. Listen & interact: openai.com/blog/musenet/ MuseNet will play an experimental concert today from 12–3pmPT on livestream: twitch.tv/openai
the biggest killer of startups is ego and posturing: most people simply can't accept when they don't know something and reach out for help, or when the market doesn't want what they've built. instead they prefer to posture, cause distractions, preserve ego, and fail.
Restocking the shelves with the latest machine generated books inspired by literature. Withering Tights Grimy Tales Pose Works Wart Stars Gore and Peas and more How will they compare to the human written notebooks? At the Pleasure Garden #creativecoding #creativeai #fringeworld
1/ I think people worry too much *what* to study instead of *how* to study. This applies to topics like programming languages, machine learning, javascript frameworks, etc. Let's take machine learning as an example.
Nouriel Roubini (from NYU) shoots down Bitcoin et al. with heavy artillery. project-syndicate.org/commentary/blo…
After 2 years of development, we've just launched fastai v1, the first deep learning library with a simple consistent API across vision, text, tabular, and collaborative filtering data. Built on the wonderful @PyTorch v1 (preview released today) fast.ai/2018/10/02/fas…
What machine learning skillsets are in demand among tech startups? Here's a quick crawl of the past 6 months of "Who is hiring?" job postings on Hacker News.
Here's the full talk from #EmTechMIT today — "Machine Learning: The Opportunity and the Opportunists" technologyreview.com/video/612109/m…
Exciting milestone in the fastai_v1 development: first modules have been created 😊 forums.fast.ai/t/first-module…
Next step in the development of fastai_v1: optimizer, training loop and callbacks! Here is a brief explanation of why we made it this way: forums.fast.ai/t/new-optimize…
Too many people in the field of AI are chasing the latest fashions. My advice: keep your eyes on the fundamentals, focus on the long-term challenges. The important questions are still the same today as they were 20 years ago.
Thinking is better done in writing. And the language in which you write affects the scope of the thoughts you can think -- absence or presence of vocabulary for certain concepts, degree of precision of word nuances, etc. Until Cicero, no one would do philosophy in Latin.
Universal Language Models Fine tuning (ULMFit) summed up! software.intel.com/en-us/articles… Cc:@jeremyphoward @seb_ruder
United States Tendências
- 1. Staged N/A
- 2. #WonderMan N/A
- 3. Ilhan Omar N/A
- 4. Belichick N/A
- 5. Jamal Murray N/A
- 6. Dort N/A
- 7. Purdue N/A
- 8. #iubb N/A
- 9. Fears N/A
- 10. Nebraska N/A
- 11. J Cole N/A
- 12. Pope N/A
- 13. Novak N/A
- 14. Musetti N/A
- 15. Bill Polian N/A
- 16. Hall of Fame N/A
- 17. Kendrick N/A
- 18. Jussie Smollett N/A
- 19. Djokovic N/A
- 20. Michigan N/A
Something went wrong.
Something went wrong.