Big Language Models Can Train Small and Cheap Language Models

Large language models can train smaller language models and uplevel them quickly. Stanford researchers trained Alpaca 7B, a model fine-tuned from the LLaMA 7B model on 52K instruction-following demonstrations. they used GPT 3.5 to train it. On our preliminary evaluation of single-turn instruction following, Alpaca behaves qualitatively similarly to OpenAI’s text-davinci-003, while being surprisingly small …

Read more

500 Wh/kg Silicon Anode Amprius Battery for Drones and Electric Passenger Planes

Amprius Technologies’ batteries deliver up to double the energy density over standard lithium-ion batteries. This means the battery cells provide more energy and power with much less weight and volume. Amprius will have a quarterly investor call at 5 pm EST, 2pm PST today. TODAY, Amprius Technologies, Inc. (“Amprius” or the “Company”) (NYSE: AMPX), a …

Read more

Gen 2 AI Text to Video

Gen-2: The Next Step Forward for Generative AI. It is a multi-modal AI system that can generate novel videos with text, images, or video clips. Arxiv- Structure and Content-Guided Video Synthesis with Diffusion Models Text-guided generative diffusion models unlock powerful image creation and editing tools. While these have been extended to video generation, current approaches …

Read more

Artificial General Intelligence is Happening, Just a Question of How Much

Stephen McAleese, at Less Wrong, used scaling rules and formulas to try to predict the performance of GPT-4. He underpredicted the capabilities of GPT-4. However, he was not widely wrong. GPT-4 came out earlier than he expected and had better performance than expected. Multi-task Language Understanding on MMLU was better than expected. GPT-4 can work …

Read more

Google Open Bard Which is AI Competition to ChatGPT

Google has opened access to Bard which is an early experiment that lets you collaborate with generative AI. This is part of Google’s answer to Chatgpt. Bard is powered by a research large language model (LLM), specifically a lightweight and optimized version of LaMDA, and will be updated with newer, more capable models over time. …

Read more

GPT4 Can Replace Jobs By Making the Best More Productive

A senior developer tested GPT4 for programming. GPT4 gave the Terraform script code for a single instance of the Fargate API. GPT4 knows that the code will not scale to 10,000 requests per second. It then describes how to create an auto-scaling group and make the modifications to scale the code with AWS and configure …

Read more

Microsoft 365 CoPilot AI to Increase Work Productivity

Microsoft Satya Nadella and Jared Spataro focused on how AI will power a whole new way of working for everyone. Word, Excel, Powerpoint and other microsoft software will create first drafts of documents which can be based upon the users previous style. CoPilot can convert Word documents into Powerpoint and Powerpoint into Word. CoPilot can …

Read more

Midjourney Image Generating AI Improves

MidJourney Version 5 launched yesterday and they did a 3 1/2 hour livestream for the launch. The improvements for generating images from prompts are: * doubled the resolution * more realistic skin textures. Blemishes, freckles and moles. * teeth and hands are more realistic * it takes grammatically correct sentences for prompts Brian WangBrian Wang …

Read more

Bing AI and GPT4 Has an IQ of 114 and is Smarter Than the Average Human

GPT4 and Google’s unreleased Palm system are getting IQ scores above the average human in IQ tests and tests that correlate to IQ. GMAT scores and other tests have been mapped to IQ and the high IQ societies accept a range of tests that GPT4 has taken. Brian WangBrian Wang is a Futurist Thought Leader …

Read more