AI Roundup 02/02 -> 02/09/2023

Last week's most interesting AI news in <2 minutes

1. Google and The Hebrew University of Jerusalem release Dreamix - a new method for video generation and editing (Paper's website)

   

2. Microsoft announces a new ChatGPT-powered Bing search engine (Microsoft Blog)

   

3. Google introduces Bard - an experimental conversational AI as part of Search, powered by Google's LLM LaMDA (Google Blog)

   

4. Researchers from CMU and Adobe propose pix2pix-zero - a new image-to-image translation method that outperforms existing models for real and synthetic image editing (Paper's website)

   

5. Google introduces SPEAR-TTS, a multi-speaker text-to-speech system that can be trained with minimal supervision (Paper)

6. Microsoft presents Synthetic Prompting - a method to generate more efficient prompts from a few handcrafted examples outperforming existing prompting techniques (Paper)

7. Researchers from NYC and the University of Maryland propose a new approach to generate hard text prompts from images (Paper's Github repo)

   

8. Berkeley researchers propose Hindsight Finetuning - a novel technique to significantly improve LLMs performance with a limited amount of human feedback (Paper)

   

9. Microsoft releases BioGPT-Large - a 1.5B parameters model from medical text generation achieving a SOTA performance of 81% accuracy (Hugging Face)

10. LAION announces Open Assistant - an initiative aimed at building an open-source alternative to commercial LLMs such as ChatGPT (Analytics India Magazine)

11. ByteDance introduces Design Booster - a text-guided model for fast image editing (Paper)

    

12. Runway presents a new model that uses language and images to generate new videos (Twitter)

13. Salesforce's recently-released BLIP-2, an LLM capable of zero-shot image-to-text generation, is now live on Hugging Face (Hugging Face)

Last week’s AI Tidbits