Kamron M retweeted
دو دستاورد اصلی انقلاب ملی تا اينجا 1. ملت قدرت خیزش میلیونی خود را دید و باور کرد( نیروی اصلی انقلاب از اینجا میآید)۲. رهبری شاهزاده با مقبولیت و مشروعیت بالا و با رای اعتماد میلیونها ایرانی در خیابان تثبیت شد. میخواهند این دو دستاورد بزرگ ملت را زیر سوال ببرند. اجازه ندهیم.
🔍Unraveling the Evolution of Self-Attention in Vision! Dive into my latest video discussing key papers & techniques reshaping image processing with self-attention mechanisms. youtu.be/bavfa_Rr2f4 #ImageProcessing #SelfAttention
1
421
Reading @rasbt's insightful book. Absolutely loving his compelling description of self-supervised pre-training vs transfer-learning.
#MachineLearning A definite must-read! 📚#CurrentReads
leanpub.com/machine-learning…
Kamron M retweeted
MMS: Massively Multilingual Speech.
- Can do speech2text and text speech in 1100 languages.
- Can recognize 4000 spoken languages.
- Code and models available under the CC-BY-NC 4.0 license.
- half the word error rate of Whisper.
Code+Models: github.com/facebookresearch/…
Paper: scontent-lga3-2.xx.fbcdn.net…
Blog: ai.facebook.com/blog/multili…
Made a video covering Self-Attention using Scaled-Dot-Product approach , explains the steps with visual representation #NLP #MachineLearning #DeepLearning
youtu.be/1IKrHh2X0F0
4
635
Lightning 2.0 is released , with the Fabric library allowing scaling up regular #pytorch codes easily
One of the Lightning 2.0 highlights is the open source Fabric library (lightning.ai/docs/fabric/sta…)!
Fabric let's you scale you PyTorch code with only a few lines of code.
Used the pre-release for my research & it's amazing!
Here's a quick demo + code:
github.com/rasbt/pytorch-fab…
Surprised to see that ChatGPT also provided link to the documentation for my query:
#ChatGPT
211
Just built a Voice Assistant WebApp using ChatGPT and Whisper API! Exciting project with some cool tech involved. Check out my latest blog post for the details! #MachineLearning #chatgpt vahidmirjalili.com/blog/post…
1
178
After #ChatGPT, the next big game-changer in the world of AI is combining modalities.
Meet Kosmos-1 by Microsoft, the Multimodal Large Language Model that takes it to the next level, perceiving information from image data and more. #Kosmos1 #LLMs #MultimodalLearning #AI
1
2
404
MusicLM 🎶 🎶 a model by Google Research for generating music from text description
google-research.github.io/se…
Paper: arxiv.org/abs/2301.11325
1
87
Kamron M retweeted
I was very skeptical of GPTZero that claims to detect human vs ChatGPT output. Now I’m more convinced after seeing this LLM watermark algorithm @tomgoldsteincs. If OpenAI deploys something like this, then GPTZero will do a *much* better job.
Brace yourself, high-schoolers!
1/
Using universal-architectures like Mask2Former from the @huggingface 🤗 transformers library for panoptic image segmentation huggingface.co/blog/mask2for…
2
202
Building a GPT model from scratch (video by Andrej Karpathy) youtu.be/kCc8FmEb1nY
#DeepLearning #GPT #ChatGPT
2
180
Kamron M retweeted
Deep embedding and alignment of protein sequences | Nature Methods nature.com/articles/s41592-0…
Q&A session on PyTorch 2.0
#PyTorch #DeepLearning
📣 Our 5th PyTorch 2.0 Q&A Session will be about TorchRec and FSDP in Production. Ask the team about updates on PyTorch FSDP for production-ready large model training.
🗓️ Weds Dec 21 @ 1pm PT
📺 Watch Dennis’ #PyTorchConference talk from before the Q&A: youtu.be/XolK2iMHc4g
234
An improved text embedding model by OpenAI:
#NLP #DeepLearning
Our new embedding model is significantly more capable at language processing and code tasks, cost effective, and simpler to use. openai.com/blog/new-and-impr…
4
324
A few useful tasks to try with ChatGPT assistant: summarizing an article, writing basic codes, task scheduling, and motivating phrases #ChatGPT
ChatGPT is changing the game, and I want to share real things you can do with this AI system today.
Please save this thread and start testing this technology NOW so you’re ahead of the curve.
Kamron M retweeted
Web Apps
With Mercury, you can build an interactive web app.
Define the interactive widgets and deploy the app to any server. When users interact with the app, they can save the results as PDF or HTML.
github.com/mljar/mercury



















