Artificial Intelligence #248

Hey, in this issue: using artificial intelligence is easier than you think; PyTorch 2.5.0 release; learning the importance of training data under concept drift; you can now train a 70b language model at home; Google Shopping 10M dataset; Meta's new research, models, and datasets; evaluate LLMs with Hugging Face Lighteval; and more.

The sponsors of this issue are Sequentum and Galileo .

• [Wired] Using artificial intelligence is easier than you think
• PyTorch 2.5.0 release : SDPA CuDNN backend, flex attention
• [Google] Learning the importance of training data under concept drift
• [Sponsored] Sequentum Cloud, the ultimate low code web scraper for trusted data

Sequentum Cloud , the ultimate low code web scraping tool with a point & click editor, text format option with robust customization, built-in tools for data quality control on a pay-as-you-go basis with no software needed. Click here to sign up today .

• [DIY] You can now train a 70b language model at home
• [Meta] Dualformer: Controllable fast and slow thinking by learning with randomized reasoning traces
• [Dataset] Google Shopping 10M dataset for large scale multimodal product retrieval and ranking
• [Sponsored] Join 10,000+ AI professionals to learn how to design an enterprise GenAI stack to get apps into production. Register now !
• [Meta] Sharing new research, models, and datasets
• [Tutorial] Evaluate LLMs with Hugging Face Lighteval on Amazon SageMaker

More than 850,000 subscribers are reading this newsletter. If you are building an AI or a data product or service, you can become a sponsor of one of the future newsletter issues and get your business featured in the newsletter. Feel free to reach out to [email protected] for more details on sponsorships.

Enjoy the newsletter? Please help us make it bigger and better by sharing it with colleagues and friends.