amitness

8 ms·

31.
Zero-Shot Learning for Text Classification(amitness.com)
2 points by amitness 6y ago|0 comments
32.
Show HN: CutCode: Double Click to Copy Code from Stack Overflow(github.com)
2 points by amitness 6y ago|1 comments
33.
by amitness 6y ago
I've elaborated number 9. Please let me know if it's clear now. Some of these methods do. For example, the tasks that power word vectors apply for many languages. https://fasttext.cc/docs/en/crawl-vectors
34.
by amitness 6y ago
Haha, just imagine if an AI faces Chandler
35.
by amitness 6y ago
Detecting sarcasm is a hard problem.
36.
by amitness 6y ago
Great to know.
37.
by amitness 6y ago
In my home country Nepal. Cheap cost of living and beautiful places to travel and explore.
38.
Self Supervised Learning in NLP(amitness.com)
130 points by amitness 6y ago|14 comments
39.
Show HN: Shuffle – Infinite Music Playback Using YouTube(amitness.com)
2 points by amitness 6y ago|0 comments
40.
Periodic Tasks for Free with GitHub Actions(github.com)
2 points by amitness 6y ago|0 comments
41.
A Visual Survey of Augmenting Text Data for NLP(amitness.com)
1 points by amitness 6y ago|0 comments
42.
by amitness 6y ago
1. Key ideas: I basically skim through blog posts and videos from conferences to be aware of key ideas in Machine Learning 2. Fundamentals: Parallelly, I take MOOC course every once in a while to learn fundamental concepts from first princi
43.
by amitness 6y ago
Twitter.
44.
by amitness 6y ago
Working on a personal blog: https://amitness.com
45.
A Machine Learning Engineer's Toolbox(github.com)
28 points by amitness 6y ago|0 comments
46.
by amitness 6y ago
I've read the ALBERT paper extensively[0] and agree with what you mean. We can't boil down years of research efforts with just 4 bullet points. The intention in the above blog post isn't to suggest that we discard detailed re
47.
by amitness 6y ago
Agree. This makes a lot of sense for incremental papers.
48.
A commit history of BERT and its forks(amitness.com)
103 points by amitness 6y ago|43 comments
49.
A Visual Guide to Recurrent Neural Network in Keras(amitness.com)
5 points by amitness 6y ago|0 comments
50.
Visual Explanation of Google AI's SimCLR Paper(amitness.com)
7 points by amitness 7y ago|1 comments
51.
DeepCluster: Combine K-means Clustering with Deep Learning(amitness.com)
2 points by amitness 7y ago|0 comments
52.
A Visual Exploration of DeepCluster(amitness.com)
1 points by amitness 7y ago|0 comments
53.
by amitness 7y ago
https://amitness.com Started it to scratch my own itch. I'm a visual learner and learn the best when the maths is explained along with visualization and intuition. So, started this blog to share explanations of latest ML re
54.
Self-Labelling: Automatically Generate Labels for Unlabeled Images(amitness.com)
3 points by amitness 7y ago|0 comments
55.
All Courses I followed to transition to ML Engineer role(github.com)
3 points by amitness 7y ago|0 comments
56.
by amitness 7y ago
Thank you. It's very encouraging to hear that.
57.
by amitness 7y ago
What are other techniques you use to generate spam? Maybe the research community can learn from you guys
58.
by amitness 7y ago
Yeah, since BERT has been trained specifically on this task of replacing random masked words, the resulting augmented sentence should sound natural compared to something like Word2Vec.
59.
by amitness 7y ago
Yeah, authors have tried mixing the strategy you described(self-supervised learning) for semi-supervised tasks. Basic idea is to learn generic image representations without manual labeling and then finetune that on your small dataset. These
60.
by amitness 7y ago
Nope,because of the way it works. So in the beginning when the model is being trained on the labeled data, it will make many mistakes. So it's confidence for either cat or dog will be low. Hence, in that case unlabeled data are not use

More ›