Video ID
string
Channel ID
string
Title
string
Time Created
string
Time Published
string
Duration
string
Description
string
Category
string
Like Count
float64
Dislike Count
float64
j3psNM5y-eA
UCv83tO5cePwHMt1952IVVHw
Implementing Filters in the New Haystack Doc Store
2022-04-06 15:53:46 UTC
2022-04-06 16:26:54 UTC
1695 seconds
๐Ÿฅณ Released with Haystack v1.3! Install direct from PyPI with: pip install 'farm-haystack[pinecone]' Join me as I work through the final few PR issues on the latest Haystack document store, and figure out how Haystack's filter_utils work. PR: https://github.com/deepset-ai/haystack/pull/2254 ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 02:41 Filtering 05:36 Testing Existing Filter Utils 07:57 Making Sense of Filter Utils 10:35 Writing the First Filter 16:26 First Working Filter 18:24 Testing New Filters 21:27 Implementing in the Doc Store 24:02 Testing Pipeline Filters 27:11 Final Issue and Outro
Science & Technology
3
0
ok0SDdXdat8
UCv83tO5cePwHMt1952IVVHw
Spotify's Podcast Search Explained
2022-04-13 15:02:31 UTC
2022-04-14 13:14:50 UTC
2998 seconds
The market for podcasts has grown tremendously in recent years. Driving the charge in podcast adoption is Spotify. In a few short years, they have become the undisputed leaders in podcasting. Despite only entering the game in 2018, by late 2021, Spotify had already usurped Apple, the long-reigning leader in podcasts, with more than 28M monthly podcast listeners. To back their podcast investments, Spotify has worked on making the podcast experience as seamless and accessible as possible. From their all-in-one podcast creation app (Anchor) to podcast APIs and their latest natural language enabled podcast search. Spotifyโ€™s natural language search for podcasts is a fascinating use case. In the past, users had to rely on keyword/term matching to find the podcast episodes they wanted. Now, they can search in natural language, in much the same way we might ask a real person where to find something. In this video, we will take a look under the hood of Spotify's podcast search, and learn how to implement a similar system ourselves. ๐ŸŒฒ Pinecone article: https://www.pinecone.io/learn/spotify-podcast-search ๐Ÿ”— Code and tests: https://github.com/pinecone-io/examples/tree/spotify-podcast-search/learn/search-in-wild/spotify-podcast-search ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 04:16 NLP in Semantic Search 08:35 Why Now? 09:29 Transformer Models 11:52 Sentence Transformers 13:12 Vector Search 15:56 How Spotify Built Podcast Search 17:35 Data Source, Fine-tuning, and Eval 22:58 Code Implementation, Dataset 24:44 Data Preparation 26:39 Query Generation 29:54 Fine-tuning a Podcast Model 41:40 Evaluation 48:05 Does it Scale? 49:00 Sharing Your Work
Science & Technology
58
1
gVAJ_l_S7uQ
UCv83tO5cePwHMt1952IVVHw
How to learn NLP for free
2022-04-24 16:41:28 UTC
2022-04-26 13:05:48 UTC
1402 seconds
Knowing what to learn is one of the hardest parts about self-learning. Imagine being thrown into the wilderness and being told to find a specific landmark. Without a map you will end up wandering to wilderness with no better option than taking one step after another. I spent a long time wandering step-by-step and eventually found my way into working with deep learning and NLP full-time. Here I will share many of the resources I used or wish I had used in the past. You can this "curriculum" as a rough guideline in self-learning ML and working towards a full-time position. ALL LINKS in article/friend link below: ๐Ÿ“• Medium article: https://jamescalam.medium.com/the-self-taught-nlp-engineer-curriculum-c425c3fc3ff6 ๐Ÿ“– Friend link: https://jamescalam.medium.com/the-self-taught-nlp-engineer-curriculum-c425c3fc3ff6?sk=986263c644d9b36699d800713faa478a --- ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 01:53 ML 101 + Prerequisites 04:58 Sentdex + Neural Nets from Scratch 07:32 ML Coursera 09:31 100 Page ML Book 11:14 Applied ML + Daniel Bourke 13:17 Origin of Modern NLP 13:41 CS224N 14:44 NLP Specialization Coursera 15:57 Modern NLP + Transformers Intro 16:54 Transformer Courses 18:14 Doing Projects 19:18 Semantic + Vector Search 19:54 NLP for Semantic Search 20:44 Mining of Massive Datasets 22:27 Final Points
Science & Technology
165
1
fb7LENb9eag
UCv83tO5cePwHMt1952IVVHw
BERTopic Explained
2022-05-10 14:13:06 UTC
2022-05-11 15:10:23 UTC
2714 seconds
90% of the world's data is unstructured. It is built by humans, for humans. That's great for human consumption, but it is *very* hard to organize when we begin dealing with the massive amounts of data abundant in today's information age. Organization is complicated because unstructured text data is not intended to be understood by machines, and having humans process this abundance of data is wildly expensive and *very slow*. Fortunately, there is light at the end of the tunnel. More and more of this unstructured text is becoming accessible and understood by machines. We can now search text based on *meaning*, identify the sentiment of text, extract entities, and much more. Transformers are behind much of this. These transformers are (unfortunately) not Michael Bay's Autobots and Decepticons and (fortunately) not buzzing electrical boxes. Our NLP transformers lie somewhere in the middle, they're not sentient Autobots (yet), but they can understand language in a way that existed only in sci-fi until a short few years ago. Machines with a human-like comprehension of language are pretty helpful for organizing masses of unstructured text data. In machine learning, we refer to this task as *topic modeling*, the automatic clustering of data into particular topics. BERTopic takes advantage of the superior language capabilities of these (not yet sentient) transformer models and uses some other ML magic like UMAP and HDBSCAN (more on these later) to produce what is one of the most advanced techniques in language topic modeling today. ๐ŸŒฒ Pinecone article: https://www.pinecone.io/learn/bertopic ๐Ÿ”— Code notebooks: https://github.com/pinecone-io/examples/tree/master/learn/algos-and-libraries/bertopic ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 01:40 In this video 02:58 BERTopic Getting Started 08:48 BERTopic Components 15:21 Transformer Embedding 18:33 Dimensionality Reduction 25:07 UMAP 31:48 Clustering 37:22 c-TF-IDF 40:49 Custom BERTopic 44:04 Final Thoughts
Science & Technology
153
3
O9lrWt15wH8
UCv83tO5cePwHMt1952IVVHw
Long Form Question Answering (LFQA) in Haystack
2022-05-17 15:22:17 UTC
2022-05-17 15:46:21 UTC
2159 seconds
Question-Answering (QA) has exploded as a subdomain of Natural Language Processing (NLP) in the last few years. QA is a widely applicable use case in NLP yet was out of reach until the introduction of [transformer models](/learn/transformers/) in 2017. Without transformer models, the level of language comprehension required to make something as complex as QA work simply was not possible. Although QA is a complex topic, it comes from a simple idea. The automatic retrieval of information via a more human-like interaction. The task of information retrieval (IR) is performed by almost every organization in the world. Without other options, organizations rely on person-to-person IR and rigid keyword search tools. This haphazard approach to IR generates a lot of friction, particularly for larger organizations. QA offers a solution to this problem. Rather than these documents being lost in an abyss, they can be stored within a space where an intelligent QA agent can access them. Unlike humans, our QA agent can scan millions of documents in seconds and return answers from these documents almost instantly. With QA tools, employees can stop wasting time searching for snippets of information and focus on their *real*, value-adding tasks. A small investment in QA is, for most organizations, a no-brainer. ๐ŸŒฒ Pinecone article: https://www.pinecone.io/learn/haystack-lfqa ๐Ÿ”— Code notebooks: https://github.com/pinecone-io/examples/blob/master/integrations/haystack/haystack_lfqa.ipynb ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 04:20 Approaches to Question Answering 05:43 Components of QA Pipeline 08:58 LFQA Generator 09:40 Haystack Setup 10:32 Initialize Document Store 13:02 Getting Data 17:53 Indexing Embeddings 21:51 Initialize Generator 24:10 Asking Questions 26:12 Common Problems 29:32 Generator Memory 31:30 Few More Questions 34:54 Outro
Science & Technology
55
1
uYas6ysyjgY
UCv83tO5cePwHMt1952IVVHw
New GPU-Acceleration for PyTorch on M1 Macs! + using with BERT
2022-05-22 16:37:37 UTC
2022-05-24 13:00:34 UTC
1140 seconds
GPU-acceleration on Mac is finally here! Today's deep learning models owe a great deal of their exponential performance gains to ever increasing model sizes. Those larger models require more computations to train and run. These models are simply too big to be run on CPU hardware, which performs large step-by-step computations. Instead, they need massively parallel computations. That leaves us with either GPU or TPU hardware. Our home PCs aren't coming with TPUs anytime soon, so we're left with the GPU option. GPUs use a highly parallel structure, originally designed to process images for visual heavy processes. They became essential components in gaming for rendering real-time 3D images. GPUs are essential for the scale of today's models. Using CPUs makes many of these models too slow to be useful, which can make deep learning on M1 machines rather disappointing. Fortunately, this is changing with the support of GPU on M1 machines beginning with PyTorch v1.12. In this video we will explain the new integration and how to implement it yourself. ๐Ÿ“• Article: https://towardsdatascience.com/gpu-acceleration-comes-to-pytorch-on-m1-macs-195c399efcc1 ๐Ÿ“– Friend Link (free access): https://towardsdatascience.com/gpu-acceleration-comes-to-pytorch-on-m1-macs-195c399efcc1?sk=a88acd35f600858093c177b97d690b03 ๐Ÿ”— Code notebooks: https://github.com/jamescalam/pytorch-mps ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 01:34 PyTorch MPS 04:57 Installing ARM Python 09:09 Using PyTorch with GPU 12:14 BERT on PyTorch GPU 13:51 Best way to train LLMs on Mac 16:01 Buffer Size Bug 17:24 When we would use Mac M1 GPU
Science & Technology
115
3
FzLIIwiaXSU
UCv83tO5cePwHMt1952IVVHw
How to Build an AI-Powered Video Search App
2022-06-01 12:37:21 UTC
2022-06-01 16:29:43 UTC
1343 seconds
Technology and culture have advanced and become ever more entangled. Some of the most significant technological breakthroughs are integrated so tightly into our culture that we never even notice theyโ€™re there. One of those is AI-powered search. It powers your Google results, Netflix recommendations, and ads you see everywhere. It is being rapidly weaved throughout all aspects of our lives. Further, this is a new technology; its full potential is unknown. This technology weaves directly into the cultural phenomenon of YouTube. Imagine a search engine like Google that allows you to rapidly access the billions of hours of YouTube content. There is no comparison to that level of highly engaging video content in the world. ๐ŸŒฒ Pinecone article: https://www.pinecone.io/learn/youtube-search ๐Ÿ”— Code: https://github.com/pinecone-io/examples/tree/master/learn/projects/yt-search ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 02:56 YouTube Search App 04:43 Getting Data 07:58 Enhancing the Data 12:45 Scraping Other Metadata 14:52 Loading Data from Hugging Face 15:42 Index and Query the Data 20:43 Streamlit App Code
Science & Technology
58
0
xXsDIK9z_fg
UCv83tO5cePwHMt1952IVVHw
Using Semantic Search to Find GIFs
2022-06-06 09:17:01 UTC
2022-06-07 12:05:40 UTC
1050 seconds
Vector search powers some of the most popular services in the world. It serves your Google results, delivers the best podcasts on Spotify, and accounts for at least 35% of consumer purchases on Amazon. In this article, we will use vector search applied to language, called semantic search, to build a GIF search engine. Unlike more traditional search where we rely on keyword matching, semantic search enables search based on the human meaning behind text and images. That means we can find highly relevant GIFs with natural language prompts. The pipeline for a project like this is simple, yet powerful. It can easily be adapted to tasks as diverse as video search or answering Super Bowl questions, or as weโ€™ll see, finding GIFs. ๐ŸŒฒ Pinecone article: https://www.pinecone.io/learn/gif-search ๐Ÿ”— Code: https://github.com/pinecone-io/examples/tree/master/learn/projects/gif-search ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 00:17 GIF Search Demo 01:56 Pipeline Overview 05:33 Data Preparation 08:17 Vector Database and Retriever 12:37 Querying 15:42 Streamlit App Code
Science & Technology
20
1
_OAU1kQdmgE
UCv83tO5cePwHMt1952IVVHw
How to Learn Data Science | ML | Programming
2022-06-15 10:37:57 UTC
2022-06-15 13:11:47 UTC
992 seconds
In this video I share five of the approaches/thoughts I have regarding learning, in particular for learning data science, machine learning, or programming. ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 01:33 Scale of Theory vs. Applied 02:55 Shape of Learning 05:52 Courses vs. Projects 08:37 Open Source 10:44 Writing 12:44 Following Interests 15:42 Final Notes
Education
24
0
BD9TkvEsKwM
UCv83tO5cePwHMt1952IVVHw
Evaluation Measures for Search and Recommender Systems
2022-06-25 14:35:27 UTC
2022-06-28 15:06:40 UTC
1885 seconds
In this video you will learn about popular offline metrics (evaluation measures) like Recall@K, Mean Reciprocal Rank (MRR), Mean Average Precision@K (MAP@K), and Normalized Discounted Cumulative Gain (NDCG@K). We will also demonstrate how each of these metrics can be replicated in Python. Evaluation of information retrieval (IR) systems is critical to making well-informed design decisions. From search to recommendations, evaluation measures are paramount to understanding what does and does not work in retrieval. Many big tech companies contribute much of their success to well-built IR systems. One of Amazonโ€™s earliest iterations of the technology was reportedly driving more than 35% of their sales. Google attributes 70% of YouTube views to their IR recommender systems. IR systems power some of the greatest companies in the world, and behind every successful IR system is a set of evaluation measures. ๐ŸŒฒ Pinecone article: https://www.pinecone.io/learn/offline-evaluation ๐Ÿ”— Code notebooks: https://github.com/pinecone-io/examples/tree/master/learn/algos-and-libraries/offline-evaluation ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 00:51 Offline Metrics 02:38 Dataset and Retrieval 101 06:08 Recall@K 07:57 Recall@K in Python 09:03 Disadvantages of Recall@K 10:21 MRR 13:32 MRR in Python 14:18 MAP@K 18:17 MAP@K in Python 19:27 NDCG@K 29:26 Pros and Cons of NDCG@K 29:48 Final Thoughts
Science & Technology
48
0
coaaSxys5so
UCv83tO5cePwHMt1952IVVHw
How to build next-level Q&A with OpenAI
2022-07-06 19:48:54 UTC
2022-07-07 13:24:35 UTC
1168 seconds
Walkthrough of the OpenAI x Pinecone Q&A app I built for a webinar with OpenAI. This is the coolest Q&A app I've ever built thanks to Pinecone vector search and OpenAI's incredible embeddings and generation endpoints. LINKS: ๐Ÿ•น App: https://pinecone-io-playground-beyond-search-openaisrcserver-h65vzl.streamlitapp.com ๐Ÿ‘จโ€๐Ÿ’ป Code and Data: https://github.com/pinecone-io/examples/tree/master/integrations/openai/beyond_search_webinar OpenAI x Pinecone Webinar: โ–ถ๏ธ https://www.youtube.com/watch?v=HtI9easWtAA ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP
Science & Technology
36
0
I3na13AESjw
UCv83tO5cePwHMt1952IVVHw
How to use Color Histograms for Image Retrieval
2022-07-11 07:01:31 UTC
2022-07-13 16:22:08 UTC
1864 seconds
Browsing, searching, and retrieving images has never been easy. Traditionally, many technologies relied on manually appending metadata to images and searching via this metadata. This approach works for datasets with high-quality annotation, but most datasets are too large for manual annotation. That means any large image dataset must rely on Content-Based Image Retrieval (CBIR). Search with CBIR focuses on comparing the *content* of an image rather than its metadata. Content can be color, shapes, textures โ€“ or with some of the latest advances in ML โ€” the "human meaning" behind an image. Color histograms represent one of the first CBIR techniques, allowing us to search through images based on their color profiles rather than metadata. ๐ŸŒฒ Pinecone article: https://pinecone.io/learn/color-histograms ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 01:23 What are Color Histograms? 08:39 How to Built Color Histograms 16:56 Using OpenCV calcHist 20:36 Image Retrieval 27:37 Pros and Cons 30:40 Final Points
Science & Technology
23
0
UzkdOg7wWmI
UCv83tO5cePwHMt1952IVVHw
๐Ÿค— Hugging Face just released *Diffusers* - for models like DALL-E 2 and Imagen!
2022-07-23 21:33:08 UTC
2022-07-26 15:27:46 UTC
934 seconds
Hugging Face of transformer fame have created a whole new Python library for diffusion models! Diffusion models are a key component of models like OpenAI's DALL-E-2, Google's Imagen, and Midjourney's image generation service. HuggingFace Diffusers brings these models to a new level of accessibility (and open source!). ๐Ÿ“• Article: https://towardsdatascience.com/hugging-face-just-released-the-diffusers-library-846f32845e65 ๐Ÿ“– Friend Link (free access): https://towardsdatascience.com/hugging-face-just-released-the-diffusers-library-846f32845e65?sk=9ec4027460defa1fd25178af9a55da13 ๐Ÿงจ Diffusers: https://github.com/huggingface/diffusers ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership 00:00 What are Diffusers? 01:55 Getting started 04:20 Prompt engineering 09:34 Testing other diffusers
Science & Technology
61
0
szfG55juoJE
UCv83tO5cePwHMt1952IVVHw
How I work from anywhere
2022-07-24 14:01:51 UTC
2022-08-16 13:55:16 UTC
767 seconds
Overview of how I deal with travel and work. Remote desk setup for staying as ergonomic and productive as possible, enjoy! ๐Ÿ”— Links to products (mostly affiliate): Laptop stand: https://amzn.to/3bZqMHM Second screen: https://amzn.to/3w6IT5B Cable bag (international): https://amzn.to/3QBH7S7 ... or UK: https://amzn.to/3ps5lT2 Peak Design backpacks: https://www.peakdesign.com/products/everyday-backpack ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP
Science & Technology
32
1
jjQetJtQDS4
UCv83tO5cePwHMt1952IVVHw
Bag of *Visual* Words for Image Classification and Retrieval
2022-08-02 20:39:30 UTC
2022-08-03 13:00:35 UTC
3367 seconds
In computer vision, bag of visual words (BoVW) is one of the pre-deep learning models used for building image embeddings. Allowing us to retrieve images from a database that are similar to another "query" image, perform object detection, and image classification. ๐ŸŒฒ Pinecone article: https://www.pinecone.io/learn/bag-of-visual-words/ ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP
Science & Technology
33
0
989aKUVBfbk
UCv83tO5cePwHMt1952IVVHw
Fast intro to multi-modal ML with OpenAI's CLIP
2022-08-11 06:17:14 UTC
2022-08-11 13:03:08 UTC
1374 seconds
OpenAI's CLIP is "multi-modal" model capable of understanding the relationships and concepts between both text and images. As we'll see, CLIP is very capable, and when used via the Hugging Face library, could not be easier to work with. ๐Ÿ“• Article: https://towardsdatascience.com/quick-fire-guide-to-multi-modal-ml-with-openais-clip-2dad7e398ac0 ๐Ÿ“– Friend Link (free access): https://towardsdatascience.com/quick-fire-guide-to-multi-modal-ml-with-openais-clip-2dad7e398ac0?sk=89bb2d8b8e583ed109d8a05e00366645 ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 00:15 What is CLIP? 02:13 Getting started 05:38 Creating text embeddings 07:23 Creating image embeddings 10:26 Embedding a lot of images 15:08 Text-image similarity search 21:38 Alternative image and text search
Science & Technology
31
0
c_u4AHNjOpk
UCv83tO5cePwHMt1952IVVHw
AlexNet and ImageNet Explained
2022-08-23 22:13:25 UTC
2022-08-24 13:00:22 UTC
2180 seconds
Todayโ€™s deep learning revolution traces back to the 30th of September, 2012. On this day, a Convolutional Neural Network (CNN) called AlexNet won the ImageNet 2012 challenge. AlexNet didnโ€™t just win; it dominated. AlexNet was unlike the other competitors. This new model demonstrated unparalleled performance on the largest image dataset of the time, ImageNet. This event made AlexNet the first widely acknowledged, successful application of deep learning. It caught peopleโ€™s attention with a 9.8 percentage point advantage over the nearest competitor. Until this point, deep learning was a nice idea that most deemed as impractical. AlexNet showed that deep learning was more than a pipedream, and the authors showed the world how to make it practical. Yet, the surge of deep learning that followed was not fueled solely by AlexNet. Indeed, without the huge ImageNet dataset, there would have been no AlexNet. The future of AI was to be built on the foundations set by the ImageNet challenge and the novel solutions that enabled the synergy between ImageNet and AlexNet. ๐ŸŒฒ Pinecone article: https://pinecone.io/learn/imagenet ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 01:06 Birth of Deep Learning 02:52 ImageNet 07:56 Lack of Readiness for Big Datasets 09:57 ImageNet Challenge (ILSVRC) 11:47 AlexNet 19:30 PYTORCH IMPLEMENTATION 19:55 Data Preprocessing 27:06 Class Prediction with AlexNet 31:50 Goldfish Results 34:27 Closing Notes
Science & Technology
20
0
pfwBut7E60Q
UCv83tO5cePwHMt1952IVVHw
Ultra-efficient Classifier Fine-tuning with Vector Search
2022-08-31 00:32:14 UTC
2022-08-31 13:00:26 UTC
1932 seconds
Learn how to use vector search to create highly targeted training for any classification model using a final linear classification layer. Easily fine-tune models in 10 minutes with less than 100 labeled examples. ๐ŸŒฒ Pinecone article: https://pinecone.io/learn/classifier-train-vector-search/ ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 01:14 Classification 02:49 Better Classifier Training 06:33 Classification as Vector Search 08:47 How Fine-tuning Works 10:50 Identifying Important Samples 12:39 CODE IMPLEMENTATION 13:13 Indexing 18:59 Fine-tuning the Classifier 27:37 Classifier Predictions 30:43 Closing Notes
Science & Technology
49
0
-S20nblUuNw
UCv83tO5cePwHMt1952IVVHw
Hugging Face Datasets #1 - Hosting your datasets
2022-09-09 12:52:32 UTC
2022-09-09 14:18:34 UTC
1382 seconds
Introduction to Hugging Face datasets, how it works, and how to host your own simple datasets (JSONL, TSV, CSV, etc) for free via Hugging Face Datasets Hub Warp download: https://app.warp.dev/referral/7G3N39 Git LFS Install: Mac: $ brew install git-lfs Debian/Ubuntu: $ curl -s https://packagecloud.io/install/repositories/github/git-lfs/script.deb.sh | sudo bash $ sudo apt-get install git-lfs Windows: Get install from https://github.com/git-lfs/git-lfs/releases ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 04:36 Creating our own Datasets 08:29 Creating JSONL for Hugging Face 15:15 Uploading Datasets for Git 19:10 LFS for Large Files 21:56 Closing Notes
Science & Technology
14
0
fGwH2YoQkDM
UCv83tO5cePwHMt1952IVVHw
CLIP Explained | Multi-modal ML
2022-09-14 23:08:40 UTC
2022-09-15 13:00:22 UTC
2013 seconds
Language models (LMs) can not rely on language alone. That is the idea behind the "Experience Grounds Language" paper, that proposes a framework to measure LMs' current and future progress. A key idea is that, beyond a certain threshold LMs need other forms of data, such as visual input. The next step beyond well-known language models; BERT, GPT-3, and T5 is โ€World Scope 3โ€. In World Scope 3, we move from large text-only datasets to large multi-modal datasets. That is, datasets containing information from multiple forms of media, like *both* images and text. The world, both digital and real, is multi-modal. We perceive the world as an orchestra of language, imagery, video, smell, touch, and more. This chaotic ensemble produces an inner state, our "model" of the outside world. AI must move in the same direction. Even specialist models that focus on language or vision must, at some point, have input from the other modalities. How can a model fully understand the concept of the word "person" without *seeing* a person? OpenAI's Contrastive Learning In Pretraining (CLIP) is a world scope three model. It can comprehend concepts in both text and image and even connect concepts between the two modalities. In this video we will learn about multi-modality, how CLIP works, and how to use CLIP for different use cases like encoding, classification, and object detection. ๐ŸŒฒ Pinecone article: https://pinecone.io/learn/clip/ ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP
Science & Technology
50
1
ODdKC30dT8c
UCv83tO5cePwHMt1952IVVHw
Hugging Face Datasets #2 - Dataset Builder Scripts
2022-09-23 14:06:51 UTC
2022-09-23 14:45:22 UTC
1404 seconds
How to work with dataset builder scripts, intro to the download manager, and Apache Arrow datatypes used in Hugging Face Datasets. ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP 00:00 Intro 00:49 Creating Compressed Files 02:41 Creating Dataset Build Script 04:49 Download Manager 08:59 Finishing Split Generator 10:13 Generate Examples Method 14:47 Add Dataset to Hugging Face 17:49 Apache Arrow Features 22:52 What's Next?
Science & Technology
13
0
98POYg2HZqQ
UCv83tO5cePwHMt1952IVVHw
Zero-Shot Image Classification with OpenAI's CLIP
2022-10-04 05:29:02 UTC
2022-10-05 14:00:03 UTC
1303 seconds
State-of-the-art (SotA) computer vision (CV) models are characterized by a *restricted* understanding of the visual world specific to their training data [1]. These models can perform *very well* on specific tasks and datasets, but they do not generalize well. They cannot handle new classes or images beyond the domain they have been trained with. Ideally, a CV model should learn the contents of images without excessive focus on the specific labels it is initially trained to understand. Fortunately, OpenAI's CLIP has proved itself as an incredibly flexible CV classification model that often requires *zero* retraining. In this chapter, we will explore CLIP in zero-shot image classification. ๐ŸŒฒ Pinecone article: https://pinecone.io/learn/clip-classification/ ๐Ÿค– 70% Discount on the NLP With Transformers in Python course: https://bit.ly/3DFvvY5 ๐ŸŽ‰ Subscribe for Article and Video Updates! https://jamescalam.medium.com/subscribe https://medium.com/@jamescalam/membership ๐Ÿ‘พ Discord: https://discord.gg/c5QtDB9RAP
Science & Technology
10
0
efPrtcLdcdM
UCZHmQk67mSJgfCCTn7xBfew
This is the worst AI ever
null
2022-06-03T15:25:58Z
null
gpt4chan #4chan #ai GPT-4chan was trained on over 3 years of posts from 4chan's "politically incorrect" (/pol/) board. (and no ...
null
null
null
TrdevFK_am4
UCZHmQk67mSJgfCCTn7xBfew
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale (Paper Explained)
null
2020-10-04T11:22:34Z
null
ai #research #transformers Transformers are Ruining Convolutions. This paper, under review at ICLR, shows that given enough ...
null
null
null
6MUpWGeGMxs
UCZHmQk67mSJgfCCTn7xBfew
NeuralHash is BROKEN - How to evade Apple's detection & craft hash collisions (w/ Open Source Code)
null
2021-08-19T14:03:52Z
null
apple #icloud #neuralhash Send your Apple fanboy friends to prison with this one simple trick ;) We break Apple's NeuralHash ...
null
null
null
n622girLRNM
UCZHmQk67mSJgfCCTn7xBfew
[ML News] Microsoft combines Images & Text | Meta makes artificial skin | Russians replicate DALL-E
null
2021-11-12T09:29:59Z
null
mlnews #turing #reskin The latest and greatest from the Machine Learning world OUTLINE: 0:00 - Intro 0:15 - Sponsor: Weights ...
null
null
null
W3mrgqtm5R4
UCZHmQk67mSJgfCCTn7xBfew
[ML News] BLOOM: 176B Open-Source | Chinese Brain-Scale Computer | Meta AI: No Language Left Behind
null
2022-07-27T20:22:22Z
null
mlnews #bloom #ai Today we look at all the recent giant language models in the AI world! OUTLINE: 0:00 - Intro 0:55 - BLOOM: ...
null
null
null
iDulhoQ2pro
UCZHmQk67mSJgfCCTn7xBfew
Attention Is All You Need
null
2017-11-28T08:04:38Z
null
https://arxiv.org/abs/1706.03762 Abstract: The dominant sequence transduction models are based on complex recurrent or ...
null
null
null
TrLrBL1U8z0
UCZHmQk67mSJgfCCTn7xBfew
[ML News] GitHub Copilot - Copyright, GPL, Patents & more | Brickit LEGO app | Distill goes on break
null
2021-07-08T12:14:37Z
null
copilot #copyright #gpl GitHub and OpenAI release Copilot, an AI-powered code autocomplete system that can generate entire ...
null
null
null
P_xeshTnPZg
UCZHmQk67mSJgfCCTn7xBfew
Perceiver: General Perception with Iterative Attention (Google DeepMind Research Paper Explained)
null
2021-03-22T17:26:39Z
null
perceiver #deepmind #transformer Inspired by the fact that biological creatures attend to multiple modalities at the same time, ...
null
null
null
rl4nUngiR2k
UCZHmQk67mSJgfCCTn7xBfew
BLEURT: Learning Robust Metrics for Text Generation (Paper Explained)
null
2020-06-07T14:11:39Z
null
Proper evaluation of text generation models, such as machine translation systems, requires expensive and slow human ...
null
null
null
19Q-vMd9bYg
UCZHmQk67mSJgfCCTn7xBfew
Inconsistency in Conference Peer Review: Revisiting the 2014 NeurIPS Experiment (Paper Explained)
null
2021-09-27T13:22:14Z
null
neurips #peerreview #nips The peer-review system at Machine Learning conferences has come under much criticism over the last ...
null
null
null
rHQPBqMULXo
UCZHmQk67mSJgfCCTn7xBfew
Machine Learning PhD Survival Guide 2021 | Advice on Topic Selection, Papers, Conferences & more!
null
2021-03-30T14:09:13Z
null
machinelearning #phd #howto This video is advice for new PhD students in the field of Machine Learning in 2021 and after.
null
null
null
GWt6Fu05voI
UCZHmQk67mSJgfCCTn7xBfew
[Classic] Deep Residual Learning for Image Recognition (Paper Explained)
null
2020-07-14T13:00:04Z
null
ai #research #resnet ResNets are one of the cornerstones of modern Computer Vision. Before their invention, people were not ...
null
null
null
pH2jZun8MoY
UCZHmQk67mSJgfCCTn7xBfew
Involution: Inverting the Inherence of Convolution for Visual Recognition (Research Paper Explained)
null
2021-05-08T10:17:15Z
null
involution #computervision #attention Convolutional Neural Networks (CNNs) have dominated computer vision for almost a ...
null
null
null
mIZLGBD99iU
UCZHmQk67mSJgfCCTn7xBfew
Did Google's LaMDA chatbot just become sentient?
null
2022-06-15T22:16:05Z
null
lamda #google #ai Google engineer Blake Lemoine was put on leave after releasing proprietary information: An interview with the ...
null
null
null
RJwPN4qNi_Y
UCZHmQk67mSJgfCCTn7xBfew
[ML News] Google's 540B PaLM Language Model & OpenAI's DALL-E 2 Text-to-Image Revolution
null
2022-04-10T09:08:24Z
null
mlnews #palm #dalle2 Google releases PaLM and OpenAI releases DALL-E 2 (and more news). Sponsor: Weights & BIases Start ...
null
null
null
YQ2QtKcK2dA
UCZHmQk67mSJgfCCTn7xBfew
The Man behind Stable Diffusion
null
2022-08-13T10:52:40Z
null
stablediffusion #ai #stabilityai An interview with Emad Mostaque, founder of Stability AI. OUTLINE: 0:00 - Intro 1:30 - What is ...
null
null
null
CRlN-cYFxTk
UCZHmQk67mSJgfCCTn7xBfew
NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis (ML Research Paper Explained)
null
2021-04-19T15:21:29Z
null
nerf #neuralrendering #deeplearning View Synthesis is a tricky problem, especially when only given a sparse set of images as an ...
null
null
null
_Z9ZP1eiKsI
UCZHmQk67mSJgfCCTn7xBfew
Curiosity-driven Exploration by Self-supervised Prediction
null
2018-03-18T12:07:52Z
null
https://arxiv.org/abs/1705.05363 Authors: Deepak Pathak, Pulkit Agrawal, Alexei A. Efros, Trevor Darrell Abstract: In many ...
null
null
null
a-VQfQqIMrE
UCZHmQk67mSJgfCCTn7xBfew
mixup: Beyond Empirical Risk Minimization (Paper Explained)
null
2020-05-27T14:13:12Z
null
Neural Networks often draw hard boundaries in high-dimensional space, which makes them very brittle. Mixup is a technique that ...
null
null
null
af6WPqvzjjk
UCZHmQk67mSJgfCCTn7xBfew
[ML News] Text-to-Image models are taking over! (Imagen, DALL-E 2, Midjourney, CogView 2 & more)
null
2022-08-07T12:54:12Z
null
mlnews #dalle #imagen All things text-to-image models like DALL-E and Imagen! OUTLINE: 0:00 - Intro 0:30 - Imagen: Google's ...
null
null
null
vLTmnaMpQCs
UCZHmQk67mSJgfCCTn7xBfew
Learning to summarize from human feedback (Paper Explained)
null
2020-09-07T11:56:46Z
null
summarization #gpt3 #openai Text Summarization is a hard task, both in training and evaluation. Training is usually done ...
null
null
null
We20YSAJZSE
UCZHmQk67mSJgfCCTn7xBfew
MuZero: Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model
null
2019-11-21T12:23:05Z
null
MuZero harnesses the power of AlphaZero, but without relying on an accurate environment model. This opens up planning-based ...
null
null
null
6dvcYx9hcbE
UCZHmQk67mSJgfCCTn7xBfew
Spurious normativity enhances learning of compliance and enforcement behavior in artificial agents
null
2022-03-08T16:24:37Z
null
deepmind #rl #society This is an in-depth paper review, followed by an interview with the papers' authors! Society is ruled by ...
null
null
null
PuOASKpiThY
UCZHmQk67mSJgfCCTn7xBfew
I'm taking a break
null
2021-07-11T12:51:06Z
null
I'll be back, don't worry :) Links: TabNine Code Completion (Referral): http://bit.ly/tabnine-yannick YouTube: ...
null
null
null
OioFONrSETc
UCZHmQk67mSJgfCCTn7xBfew
Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
null
2019-02-02T15:16:38Z
null
https://arxiv.org/abs/1502.03167 Abstract: Training Deep Neural Networks is complicated by the fact that the distribution of each ...
null
null
null
M2-BE5JotjA
UCZHmQk67mSJgfCCTn7xBfew
PAIR AI Explorables | Is the problem in the data? Examples on Fairness, Diversity, and Bias.
null
2021-04-07T12:38:57Z
null
In the recurring debate about bias in Machine Learning models, there is a growing argument saying that "the problem is not in the ...
null
null
null
7K4Z8RqjWIk
UCZHmQk67mSJgfCCTn7xBfew
MLP-Mixer: An all-MLP Architecture for Vision (Machine Learning Research Paper Explained)
null
2021-05-06T11:53:49Z
null
mixer #google #imagenet Convolutional Neural Networks have dominated computer vision for nearly 10 years, and that might ...
null
null
null
dND-7llwrpw
UCZHmQk67mSJgfCCTn7xBfew
Grokking: Generalization beyond Overfitting on small algorithmic datasets (Paper Explained)
null
2021-10-06T20:12:52Z
null
grokking #openai #deeplearning Grokking is a phenomenon when a neural network suddenly learns a pattern in the dataset and ...
null
null
null
T35ba_VXkMY
UCZHmQk67mSJgfCCTn7xBfew
DETR: End-to-End Object Detection with Transformers (Paper Explained)
null
2020-05-28T15:09:01Z
null
Object detection in images is a notoriously hard task! Objects can be of a wide variety of classes, can be numerous or absent, they ...
null
null
null
_8KNb5iqblE
UCZHmQk67mSJgfCCTn7xBfew
Longformer: The Long-Document Transformer
null
2020-04-20T14:07:56Z
null
The Longformer extends the Transformer by introducing sliding window attention and sparse global attention. This allows for the ...
null
null
null
T9XSU0pKX2E
UCZHmQk67mSJgfCCTn7xBfew
OpenAI CLIP: ConnectingText and Images (Paper Explained)
null
2021-01-12T14:52:03Z
null
ai #openai #technology Paper Title: Learning Transferable Visual Models From Natural Language Supervision CLIP trains on 400 ...
null
null
null
h3ij3F3cPIk
UCZHmQk67mSJgfCCTn7xBfew
DINO: Emerging Properties in Self-Supervised Vision Transformers (Facebook AI Research Explained)
null
2021-05-01T19:53:03Z
null
dino #facebook #selfsupervised Self-Supervised Learning is the final frontier in Representation Learning: Getting useful features ...
null
null
null
dmH1ZpcROMk
UCZHmQk67mSJgfCCTn7xBfew
Reward Is Enough (Machine Learning Research Paper Explained)
null
2021-05-31T13:27:21Z
null
reinforcementlearning #deepmind #agi What's the most promising path to creating Artificial General Intelligence (AGI)? This paper ...
null
null
null
rFwQDDbYTm4
UCZHmQk67mSJgfCCTn7xBfew
[Classic] Playing Atari with Deep Reinforcement Learning (Paper Explained)
null
2020-07-26T13:00:23Z
null
ai #dqn #deepmind After the initial success of deep neural networks, especially convolutional neural networks on supervised ...
null
null
null
aX8phGhG8VQ
UCZHmQk67mSJgfCCTn7xBfew
Does GPT-3 lie? - Misinformation and fear-mongering around the TruthfulQA dataset
null
2021-09-21T13:43:28Z
null
gpt-3 #truth #conspiracy A new benchmark paper has created quite an uproar in the community. TruthfulQA is a dataset of 817 ...
null
null
null
Elxn8rS88bI
UCZHmQk67mSJgfCCTn7xBfew
Pretrained Transformers as Universal Computation Engines (Machine Learning Research Paper Explained)
null
2021-03-16T18:15:38Z
null
universalcomputation #pretrainedtransformers #finetuning Large-scale pre-training and subsequent fine-tuning is a common ...
null
null
null
jltgNGt8Lpg
UCZHmQk67mSJgfCCTn7xBfew
Neural Ordinary Differential Equations
null
2019-02-19T05:12:20Z
null
https://arxiv.org/abs/1806.07366 Abstract: We introduce a new family of deep neural network models. Instead of specifying a ...
null
null
null
eYgPJ_7BkEw
UCZHmQk67mSJgfCCTn7xBfew
FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence
null
2020-04-15T18:01:56Z
null
FixMatch is a simple, yet surprisingly effective approach to semi-supervised learning. It combines two previous methods in a ...
null
null
null
DEh1GR0t29k
UCZHmQk67mSJgfCCTn7xBfew
Peer Review is still BROKEN! The NeurIPS 2021 Review Experiment (results are in)
null
2021-11-25T18:26:52Z
null
neurips #peerreview #machinelearning A look at the results of the 2021 NeurIPS peer review experiment.
null
null
null
hsOMCwvFv80
UCZHmQk67mSJgfCCTn7xBfew
I'm out of Academia
null
2021-05-04T13:25:48Z
null
machinelearning #ai #phd Done with my PhD in Machine Learning at ETH Zurich. On to new lands! Links: TabNine Code ...
null
null
null
VgqHitvEbR0
UCZHmQk67mSJgfCCTn7xBfew
[Rant] REVIEWER #2: How Peer Review is FAILING in Machine Learning
null
2020-08-18T11:16:07Z
null
ai #research #peerreview Machine Learning research is in dire straits as more people flood into the field and competent reviewers ...
null
null
null
kOy49NqZeqI
UCZHmQk67mSJgfCCTn7xBfew
IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
null
2019-11-01T13:08:17Z
null
Policy Gradient RL on a massively distributed scale with theoretical guarantees! Abstract: In this work we aim to solve a large ...
null
null
null
xbxe-x6wvRw
UCZHmQk67mSJgfCCTn7xBfew
[ML News] Stable Diffusion Takes Over! (Open Source AI Art)
null
2022-09-19T00:06:24Z
null
stablediffusion #aiart #mlnews Stable Diffusion has been released and is riding a wave of creativity and collaboration. But not ...
null
null
null
0PAiQ1jTN5k
UCZHmQk67mSJgfCCTn7xBfew
How to make your CPU as fast as a GPU - Advances in Sparsity w/ Nir Shavit
null
2022-09-17T12:19:00Z
null
ai #sparsity #gpu Sparsity is awesome, but only recently has it become possible to properly handle sparse models at good ...
null
null
null
rNkHjZtH0RQ
UCZHmQk67mSJgfCCTn7xBfew
NFNets: High-Performance Large-Scale Image Recognition Without Normalization (ML Paper Explained)
null
2021-02-14T16:51:10Z
null
nfnets #deepmind #machinelearning Batch Normalization is a core component of modern deep learning. It enables training at ...
null
null
null
dWGjoInRaAs
UCZHmQk67mSJgfCCTn7xBfew
[ML News] DeepMind fails to get independence from Google
null
2021-05-26T20:24:08Z
null
deepmind #google #mlnews DeepMind has reportedly failed to negotiate for greater independence from Google/Alphabet.
null
null
null
iAR8LkkMMIM
UCZHmQk67mSJgfCCTn7xBfew
Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity
null
2021-01-22T19:38:10Z
null
ai #technology #switchtransformer Scale is the next frontier for AI. Google Brain uses sparsity and hard routing to massively ...
null
null
null
_9aN1-0T8hg
UCZHmQk67mSJgfCCTn7xBfew
[ML News] AI models that write code (Copilot, CodeWhisperer, Pangu-Coder, etc.)
null
2022-08-10T20:55:14Z
null
mlnews #ai #copilot OUTLINE: 0:00 - Intro 0:20 - Copilot Now Generally Available 3:20 - FOSS Org leaves GitHub 6:45 - Google's ...
null
null
null
pwSnC8jlh50
UCZHmQk67mSJgfCCTn7xBfew
[ML News] Meta's OPT 175B language model | DALL-E Mega is training | TorToiSe TTS fakes my voice
null
2022-05-10T12:54:25Z
null
mlnews #dalle #gpt3 An inside look of what's happening in the ML world! Sponsor: Weights & Biases https://wandb.me/yannic ...
null
null
null
-9evrZnBorM
UCZHmQk67mSJgfCCTn7xBfew
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
null
2019-01-30T08:51:15Z
null
https://arxiv.org/abs/1810.04805 Abstract: We introduce a new language representation model called BERT, which stands for ...
null
null
null
awyuuJoHawo
UCZHmQk67mSJgfCCTn7xBfew
Dream to Control: Learning Behaviors by Latent Imagination
null
2020-04-03T12:46:59Z
null
Dreamer is a new RL agent by DeepMind that learns a continuous control task through forward-imagination in latent space.
null
null
null
FNDVy_BR8aA
UCZHmQk67mSJgfCCTn7xBfew
Can Wikipedia Help Offline Reinforcement Learning? (Author Interview)
null
2022-02-28T15:24:13Z
null
wikipedia #reinforcementlearning #languagemodels Original paper review here: https://youtu.be/XHGh19Hbx48 Machel Reid and ...
null
null
null
DEh1GR0t29k
UCZHmQk67mSJgfCCTn7xBfew
Peer Review is still BROKEN! The NeurIPS 2021 Review Experiment (results are in)
null
2021-11-25T18:26:52Z
null
neurips #peerreview #machinelearning A look at the results of the 2021 NeurIPS peer review experiment.
null
null
null
yPjuAo53uNI
UCZHmQk67mSJgfCCTn7xBfew
[Rant] The Male Only History of Deep Learning
null
2020-04-22T11:58:28Z
null
This casting of our field in terms of ideological narrow-sighted group-think is disgusting. Keep Science about ideas!
null
null
null
a4VvcmqnkhY
UCZHmQk67mSJgfCCTn7xBfew
What Matters In On-Policy Reinforcement Learning? A Large-Scale Empirical Study (Paper Explained)
null
2020-08-20T09:46:35Z
null
ai #research #machinelearning Online Reinforcement Learning is a flourishing field with countless methods for practitioners to ...
null
null
null
dPsXxLyqpfs
UCZHmQk67mSJgfCCTn7xBfew
World Models
null
2018-04-07T13:36:31Z
null
Authors: David Ha, Jรผrgen Schmidhuber Abstract: We explore building generative neural network models of popular ...
null
null
null
q6Kyvy1zLwQ
UCZHmQk67mSJgfCCTn7xBfew
BERTology Meets Biology: Interpreting Attention in Protein Language Models (Paper Explained)
null
2020-07-02T13:27:53Z
null
Proteins are the workhorses of almost all cellular functions and a core component of life. But despite their versatility, all proteins ...
null
null
null
iAR8LkkMMIM
UCZHmQk67mSJgfCCTn7xBfew
Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity
null
2021-01-22T19:38:10Z
null
ai #technology #switchtransformer Scale is the next frontier for AI. Google Brain uses sparsity and hard routing to massively ...
null
null
null
W3mrgqtm5R4
UCZHmQk67mSJgfCCTn7xBfew
[ML News] BLOOM: 176B Open-Source | Chinese Brain-Scale Computer | Meta AI: No Language Left Behind
null
2022-07-27T20:22:22Z
null
mlnews #bloom #ai Today we look at all the recent giant language models in the AI world! OUTLINE: 0:00 - Intro 0:55 - BLOOM: ...
null
null
null
5skIqoO3ku0
UCZHmQk67mSJgfCCTn7xBfew
OpenAI Embeddings (and Controversy?!)
null
2022-02-07T20:10:42Z
null
mlnews #openai #embeddings COMMENTS DIRECTLY FROM THE AUTHOR (thanks a lot for reaching out Arvind :) ): 1.
null
null
null
af6WPqvzjjk
UCZHmQk67mSJgfCCTn7xBfew
[ML News] Text-to-Image models are taking over! (Imagen, DALL-E 2, Midjourney, CogView 2 & more)
null
2022-08-07T12:54:12Z
null
mlnews #dalle #imagen All things text-to-image models like DALL-E and Imagen! OUTLINE: 0:00 - Intro 0:30 - Imagen: Google's ...
null
null
null
dND-7llwrpw
UCZHmQk67mSJgfCCTn7xBfew
Grokking: Generalization beyond Overfitting on small algorithmic datasets (Paper Explained)
null
2021-10-06T20:12:52Z
null
grokking #openai #deeplearning Grokking is a phenomenon when a neural network suddenly learns a pattern in the dataset and ...
null
null
null
FC-R4MlIqrc
UCZHmQk67mSJgfCCTn7xBfew
[ML News] Cedille French Language Model | YOU Search Engine | AI Finds Profitable MEME TOKENS
null
2021-11-18T14:54:52Z
null
mlnews #cedille #wmt Only the greatest of news from the world of Machine Learning. OUTLINE: 0:00 - Sponsor: Weights & Biases ...
null
null
null
56GW1IlWgMg
UCZHmQk67mSJgfCCTn7xBfew
Learning model-based planning from scratch
null
2017-08-09T06:02:41Z
null
https://arxiv.org/abs/1707.06170 Abstract: Conventional wisdom holds that model-based planning is a powerful approach to ...
null
null
null
T35ba_VXkMY
UCZHmQk67mSJgfCCTn7xBfew
DETR: End-to-End Object Detection with Transformers (Paper Explained)
null
2020-05-28T15:09:01Z
null
Object detection in images is a notoriously hard task! Objects can be of a wide variety of classes, can be numerous or absent, they ...
null
null
null
AJwnbSP_rq8
UCZHmQk67mSJgfCCTn7xBfew
GPT-NeoX-20B - Open-Source huge language model by EleutherAI (Interview w/ co-founder Connor Leahy)
null
2022-02-04T16:49:40Z
null
eleuther #gptneo #gptj EleutherAI announces GPT-NeoX-20B, a 20 billion parameter open-source language model, inspired by ...
null
null
null
fvctpYph8Pc
UCZHmQk67mSJgfCCTn7xBfew
Do ImageNet Classifiers Generalize to ImageNet? (Paper Explained)
null
2020-04-27T12:42:47Z
null
Has the world overfitted to ImageNet? What if we collect another dataset in exactly the same fashion? This paper gives a ...
null
null
null
ZVVnvZdUMUk
UCZHmQk67mSJgfCCTn7xBfew
The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks
null
2020-04-13T14:51:36Z
null
Stunning evidence for the hypothesis that neural networks work so well because their random initialization almost certainly ...
null
null
null
rFwQDDbYTm4
UCZHmQk67mSJgfCCTn7xBfew
[Classic] Playing Atari with Deep Reinforcement Learning (Paper Explained)
null
2020-07-26T13:00:23Z
null
ai #dqn #deepmind After the initial success of deep neural networks, especially convolutional neural networks on supervised ...
null
null
null
xnChXNUNS2A
UCZHmQk67mSJgfCCTn7xBfew
[ML News] This AI completes Wikipedia! Meta AI Sphere | Google Minerva | GPT-3 writes a paper
null
2022-07-31T13:33:13Z
null
mlnews #ai #minerva This episode is all about models that reason. OUTLINE: 0:00 - Intro 0:35 - Meta AI learns Wikipedia citations ...
null
null
null
ZfDZRX3WiJg
UCZHmQk67mSJgfCCTn7xBfew
VirTex: Learning Visual Representations from Textual Annotations (Paper Explained)
null
2020-06-12T17:28:15Z
null
Pre-training a CNN backbone for visual transfer learning has recently seen a big push into the direction of incorporating more data ...
null
null
null
CA8JPbJ75tY
UCZHmQk67mSJgfCCTn7xBfew
CornerNet: Detecting Objects as Paired Keypoints (Paper Explained)
null
2020-06-05T13:31:46Z
null
Many object detectors focus on locating the center of the object they want to find. However, this leaves them with the secondary ...
null
null
null
wTzvKB6D_34
UCZHmQk67mSJgfCCTn7xBfew
How far can we scale up? Deep Learning's Diminishing Returns (Article Review)
null
2021-10-02T14:24:54Z
null
deeplearning #co2 #cost Deep Learning has achieved impressive results in the last years, not least due to the massive increases ...
null
null
null
pZyxlf6l0N8
UCZHmQk67mSJgfCCTn7xBfew
Thinking While Moving: Deep Reinforcement Learning with Concurrent Control
null
2020-04-23T13:26:07Z
null
Classic RL "stops" the world whenever the Agent computes a new action. This paper considers a more realistic scenario where ...
null
null
null
hMO6rbMAPew
UCZHmQk67mSJgfCCTn7xBfew
Adversarial Examples Are Not Bugs, They Are Features
null
2019-05-14T13:45:57Z
null
Abstract: Adversarial examples have attracted significant attention in machine learning, but the reasons for their existence and ...
null
null
null
WknN4E-y44E
UCZHmQk67mSJgfCCTn7xBfew
Research Conference ICML drops their acceptance rate | Area Chairs instructed to be more picky
null
2021-05-11T17:47:17Z
null
icml #machinelearning #conference In a controversial move, ICML Area Chairs were instructed to raise the bar on acceptance to ...
null
null
null
rNkHjZtH0RQ
UCZHmQk67mSJgfCCTn7xBfew
NFNets: High-Performance Large-Scale Image Recognition Without Normalization (ML Paper Explained)
null
2021-02-14T16:51:10Z
null
nfnets #deepmind #machinelearning Batch Normalization is a core component of modern deep learning. It enables training at ...
null
null
null
o75ybZ-6Uu8
UCZHmQk67mSJgfCCTn7xBfew
Dreamer v2: Mastering Atari with Discrete World Models (Machine Learning Research Paper Explained)
null
2021-02-19T16:11:18Z
null
dreamer #deeprl #reinforcementlearning Model-Based Reinforcement Learning has been lagging behind Model-Free RL on Atari ...
null
null
null