?? AI Takes Another Leap Forward: GPT-4 Unveiled, Google Suite Gets Smarter, Stanford's Alpaca Offers Low-Cost LLM Training ??????
?? Welcome Readers!
The sphere of AI is expected to become even more dynamic this year, with half driven by innovation and half by competition.
From the astonishing release of GPT-4 to Claude getting deeper into the race of Large Language Models, it's all about "Language" this week!
Instead of integrating AI into their search engine, Google introduces Language Models directly to their Suite, powering Word, Sheets, and more! ??
Sit aside; GPT -4 is here! ??
With the world still starstruck with ChatGPT, OpenAI launches GPT-4 this week, changing the standard of Large Language Models in spheres like Conversational AI, Image-to-text, Generative AI, and Translating AI!
Completely overtaking the status quo set by GPT 3.5 powered ChatGPT, GPT-4 comes powered with even more parameters allowing the model to talk and see, with better context grasping from Images.
The model can compete on even more difficult human examinations and bags an astonishing 90 percentile at the Bar, does that mean Lawyers may be the following jobs threatened by AI? Only time will tell, and let us look at this journey together!
Read More about GPT-4 and its capabilities here!
Claude: Please talk to me, too. ???
Did you catch the mental breakdown ChatGPT experienced when put in a position on Bing? Conversational AI works on learning from conversations and given they are exposed to the entire internet, they tend to get a bit "edgy" (just like you or me). However, to improve this experience, Anthropic's Claude focuses on training helpful, honest, and harmless AI systems.
Claude is capable of a wide variety of conversational and text-processing tasks while maintaining high reliability and predictability. With the model focusing on all the features expected by a nice LLM to perform, the team has taken extra care to make the conversations more steerable.
The latest?public release?focuses on user testimonials, so how about checking this out for your next healthcare or wellness application? Also, if you feel like not spending money (come on, we know you are developers), how about reading their take on?AI Safety?to inspire your policies?
?? Want to stay up-to-date on the latest ML news and trends??Check out our blog?for in-depth analysis and expert insights.
Google Suite powered by AI ?
Thomas Kurian, the CEO of Google Cloud, comes with the announcement about a deeper integration of AI on the entire Google Cloud. Going all the way from Google Vertext to Google Docs, the effort is aimed at increasing Human-AI Collaboration in a more streamlined way.
However, one peculiar thing that shows up in this announcement is the use of Google's legacy Large Language Model PaLM as a model instead of the latest Bard. This begs the question were all the departments of Google escalated to work on AI integration in their services independently following the ChatGPT hype, or was Bard just not good enough to power Docs and Sheets yet?
领英推荐
All that aside, this may come as a boon to all those systems brought to the knees with Grammarly, ChatGPT, Quillbot, etc., working side by side for that school essay you are working on!
Stanford trains their answer to ChatGPT for less than $600 ??
Stanford's Alpaca may be the cheaper alternative to training your in-house large language model, coming closer to the results generated by ChatGPT.??Alpaca 7B, a model fine-tuned from the LLaMA 7B model on 52K instruction-following demonstrations.
However, the model is strictly available for academic research and does not qualify for commercial use because of how it has been developed. The instruction data for the model is based on OpenAI's text-DaVinci-003, whose terms of use prohibit developing models that compete with OpenAI.
Alpaca also exhibits several common deficiencies of language models, including hallucination, toxicity, and stereotypes which it shares with its inspirations, the LLaMA 7B and Text-DaVinci-003, so go for it at your own risk.
?? AI-Fueled Finds
?? We invite you to join us to connect with other like-minded professionals in the AI and MLOps industry.?Join Our Discord Community
?? AI-Related Reads
That concludes this edition of the newsletter. Thanks for reading, and we can't wait to share more with you in our next one!
Follow our?LinkedIn?&?Twitter?for more updates and insights. Until then, stay curious, and we'll see you soon.
We want to give a huge shoutout to our contributors for their tireless efforts.
Content support from:?Aryan Kargwal?&?Krish Katyal
Formatting support from:?Priya S
Lots of ?? from,
NimbleBox.ai?Team
Thanks for Sharing! ?? NimbleBox.ai