In 1989, 32-year-old Yann LeCun introduced the first Convolutional Neural Network (CNN) for text recognition. This breakthrough in AI and machine learning, which accurately read handwritten text, led to the development of today's image and text recognition technologies.
@ylecun
---
Microsoft released Phi-3.5 model today.
A multimodal (vision) and MoE version is added.
It is claimed that the mini version is equivalent/outperforms Llama 3.1 8b but remember the recent occurrences of LLMs overfitting test benchmarks.
I came across this text promptable image segmentation demo:
Combination of the zero-shot object detection model OWLv2 and mask generation model SAM2
Give it a try in this HuggingFace space:
Here's my conversation with Pieter Levels (
@levelsio
), self-taught developer and entrepreneur who designed, programmed, shipped, and ran over 40 startups, many of which are hugely successful. In most cases, he did it all by himself, while living the digital nomad life in over 40
Have you heard about GitHub models?
It's a new experimental feature that allows developers to prototype products that use AI models for free.
Join the waitlist to get early access.
Read more:
Also, let's connect on GitHub:
Grok-2 from
@xai
is out.
Predictions said that it might beat GPT-4o. It did not but is still one of the top models in Chatbot Arena.
Read more:
#GPT4O
#GROK2AI
#Chatbots
Llama-3.1
-Enhancing the 8B and 70B versions
-Introducing the powerful 405B model
Llama-3.1-405B is potentially the strongest open-source language model, surpassing GPT-4 in benchmarks and offering a 128k token context length.
@karpathy
Such a deterministic software design and still these events happen. Do you still think AI-based (blackbox) OS is possible in near future? Isn't that too optimistic?