Machine Learning
1297 Threads
Topological Deep Learning is an immensely powerful and fast emerging field. Our new literature review https://t.co/MLp4AaV8vf is out and hereβs why Iβm very excited about itπ§΅1/5
I learned AI/ML by spending $0 Online. Here are 5 Free Courses that will teach you ML better than the paid ones, A thread 𧡠π
How the Decision Tree model chooses the best questions? Gini impurity index 𧡠1/9 https://t.co/jhHd20CSk8
AI (π€) update for the past couple of days! π 1) @MetaAI released DINOv2 - v2 of their DINO system that uses self-supervised learning to learn features useful for various vision ta...
How can you develop AI projects at lightning speed? By using serverless GPU workflows. Introducing Banana: Serverless GPUs for ML Inference. Hereβs why you should use it for you...
Following the release of LLaMA, we saw a rapid explosion of open-source research on large language models (LLMs). Here are the three most notable model releases during this timeβ¦ π§΅...
π€Generative Agentsπ€ Last week, Park et all released βGenerative Agentsβ, a paper simulating interactions between tens of agents We gave it a close read, and implemented one of th...
What are some design patterns in machine learning systems? Here are a few I've seen: 1. Cascade: Break a complex problem into simpler problems. Each subsequent model focuses on m...
Computer vision had a lot of quick wins in bootstrapping massive datasets over the last decade: π¨ Image colorization - convert color photos to B&W π¬ Image superresolution - downs...
Stanford, Google, Hardvard and more- are offering free courses on AI. Prompt engineering, Conversational AI, Natural Language Processing & more - drives today's AI applications....
I have been coding in Python for 8 years now. β³ Here's a roadmap to Master Python! π A Thread π§΅π
reading the GPTQ paper, about post-training quantization for GPTs https://t.co/JGpdhWOCtG it can quantize 175B models in 4 GPU hours down to 3/4 bits. https://t.co/spIUrlPbha
Statistics for Machine Learning π Understanding the two broad types: - Descriptive - Inferential A Thread π§΅π https://t.co/OFmj8tiVdB
Building an ML model from zero is hard. But transfer learning can save resources and time. Here I explain how it works. 1/6 https://t.co/AVWMDZNMSE
I started my career in Data Science back in 2016. Here's a list of courses that I took!! π§΅ππ»
Presenting Reflected Diffusion Models w/ @StefanoErmon! Diffusion models should reverse a SDE, but common hacks break this. We provide a fix through a general framework. Arxiv: h...
This is the first MLOps stack I used in my ML-life: β AWS Redshift + SQL to generate features β A cron job to schedule runs β An EC2 instance to run training and inference βAn S3...
Missing data can mess up your Machine Learning models. But pseudocounts can help you out. I will show you how. 1/9 https://t.co/GLaMBCoPA2
A very underrated architecture tweak to GPT is multi-query attention (MQA): sharing value/key across attention heads saves a lot of memory in the kv-cache. Max generation batch si...
Thereβs a pervasive myth that the No Free Lunch Theorem prevents us from building general-purpose learners. Instead, we need to select models on a per-domain basis. Is this really...
Embeddings are the building blocks of powerful models like ChatGPT & GPT-4. Let's talk about them today! π We'll also understand how to harness the power of embeddings using an e...
Day 29 of #100dayswithMachinelearning Topic - Machine Learning (ML) Pipelines A Thread 𧡠https://t.co/N448hICioz
Random matrices are very important in modern statistics and machine learning, not to mention physics A model about which much less is known is uniformly sampled matrices from the...
*8 Data Analysis Projects for beginners that will transform your Portfolio and your Resume.* *With source code in Python π*
