Have you heard these exciting AI news? - May 10, 2024 - AI Updates Weekly
Вставка
- Опубліковано 31 тра 2024
- Have you heard these exciting AI news? - May 10, 2024 - AI Updates Weekly
Presented by Lev Selector
Slides - github.com/lselector/seminar/...
--------- My websites:
- Enterprise AI Solutions - EAIS.ai
- Linkedin - / levselector
- GitHub - github.com/lselector
--------- Contents of today's video:
Llama3 Continued Success
Apple M4 Chip - Available in iPad Pro on May 15
Microsoft MAI-1
AI-enabled Web Search
DeepSeek-V2, open-source, 236B, ~GPT-4
Large LLMs are better at "generalizing"
Regulatory Capture
Andrew Ng about AI and regulations
Yoshua Bengio - Fears About the Future of AI
My "ai" GitHub repo - LLM & RAG
Most profitable business ideas with AI
Galileo Protect
Information Overload
Selling Information Stopped Working
OpenAI will use Stack Overflow's OverflowAPI
Warren Buffett warns about AI scamming
X.ai Grok + X.com: news + comments
ScrapeGraphAI - web scraping using AI
DocRes - Document Image Restoration
DeepMind AlphaFold 3
OpenAI Public Model Specification
DeepLearning.ai - Agentic RAG Course
Mistral AI valued at $6 Billion
Alibaba Qwen2.5
Gemma with 10 Mln tokens context window
Phi-3 WebGPU runs in browser
Guardian for CIA (by Microsoft)
IBM Granite Code - open-source LLMs
massive prompts can outperform fine-tuning
Ilya Sutskever's list of papers
Crowd-sourced "Arena" Leaderboard
Tech Layoffs were lower than in 2023
This is a great roundup of AI news and updates. You should definitely have more views and likes.
Thanks for this videos Lev. I watch them all
Same 😊
I just made a comment on another channel that in context learning is probably now a better way to provide data and instructions to LLMs (fine-tuning having become a very sensitive process). And now you post research proclaiming the same. Its especially effective now that we are getting 10M context windows.
BTW Best GPTs I have previously created do exactly that.
Its massively hypocritical that OpenAI having trained on public and often copyrighted data, now want to turn around and demand regulations to have their model weights protected.