TLDR AI 2023-11-02

3D asset generation in the .obj format is the next big innovation for creatives working with AI. Stability AI is working to make that a possibility. This article is light on details, but it is exciting to see what is around the corner.

Microsoft Research has enhanced its smaller language model, Phi 1.5, with multimodal capabilities, enabling it to interpret images similarly to OpenAI's larger GPT-4 model but at a lower computational cost.

Biden has gone all in on an executive order that, while lacking teeth, gives a good sense of where the administration is coming from when it comes to AI, and from that, we can glean potential impacts on the startup ecosystem.

MIT and NVIDIA researchers developed techniques to accelerate sparse tensor processing, enhancing performance and energy efficiency in generative AI tasks.

The study introduces 'Learning from Mistakes' (LeMa), a method that teaches large language models to solve math problems by learning from errors, similar to a human student improving through correcting their mistakes.

Researchers have developed a neural network that can generalize language similarly to humans, outperforming models like ChatGPT.

The easiest and fastest way to build a production-ready LLM application. These templates serve as a set of reference architectures for a wide variety of popular LLM use cases.

This guide provides information and resources to help you set up Llama including how to access the model, hosting, and how-to and integration guides

Jupyter notebook example of how to fine-tune Mistral's 7B model on your own dataset.

Open vocabulary segmentation is a somewhat new and exciting topic in computer vision. This work combines powerful base models with strong language capabilities together to segment based on text input.

Paul Adam, a long-term CP at Intercom, discusses the ways AI is going to impact the world (and your tech stack). There’s quite a bit of fear-setting in this conversation, so if this doesn’t inspire you to look into how you should be using AI, I don’t know what will.

Global stakeholders discussed collaborative efforts to address AI risks through the Bletchley Declaration at the AI Safety Summit in Bletchley Park, England, aiming for a human-centric, trustworthy AI deployment. U.S. Secretary of Commerce Gina Raimondo announced a new AI safety institute within the Department of Commerce, emphasizing global policy alignment on AI safety, while political leaders from various countries stressed inclusivity and responsibility in navigating AI advancements.

Join AI leaders from LlamaIndex, Cohere, MosaicML, Airbnb, and more to learn how to overcome LLMOps challenges and deliver real business value with LLMs. Register now for AI Forward!

Use AI to craft better LinkedIn posts.

Your ability to aggregate and understand customer data is improving, allowing you to have a better understanding of your customers, predict their future behavior more accurately, offer scaled customer support, and curate messaging that resonates more accurately.

Use AI to generate keywords for SEO.

AWS introduced Amazon Elastic Compute Cloud (EC2) Capacity Blocks for ML to facilitate access to GPU resources for running large language models, allowing customers to reserve GPU instances for defined time periods.