stp2y - CybAI news

25 May

Generative AI to digital twins: Powering the AI revolution

stp2y0 CommentsAI

This article is based on Santosh Radha’s brilliant talk at the AI Accelerator Summit in San Jose. As an AIAI member, you can enjoy the complete recording here. For more exclusive content, head to your membership dashboard.Generative AI is revolutionizing how we interact with technology. From chatbots that converse like humans to image generators producing stunning visuals, this incredible tech is transforming our world. But beneath these mind-blowing capabilities lies a massive computing infrastructure packed with technical complexities that often go unnoticed.In this article, we'll dive into the realm of high-performance computing (HPC) and the challenges involved in productionizing generative AI…

25 May

Google ushers in the “Gemini era” with AI advancements

stp2y0 CommentsChat-GPT

Google has unveiled a series of updates to its AI offerings, including the introduction of Gemini 1.5 Flash, enhancements to Gemini 1.5 Pro, and progress on Project Astra, its vision for the future of AI assistants. Gemini 1.5 Flash is a new addition to Google’s family of models, designed to be faster and more efficient to serve at scale. While lighter-weight than the 1.5 Pro, it retains the ability for multimodal reasoning across vast amounts of information and features the breakthrough long context window of one million tokens. “1.5 Flash excels at summarisation, chat applications, image and video captioning, data…

24 May

Tech Layoffs: US Companies With Job Cuts In and 2023 and 2024

stp2y0 CommentsRAG models

US Tech Layoffs Continue With More Total Closures Reported, Yet Another Google Cut, And Even AI Shedding Workers The past couple of weeks were hard on U.S. tech workers — as well as their employers — with shutdowns reported in the healthcare, consumer goods and fintech sectors. Even retail behemoth Walmart joined this layoffs tracker this week with a report that the Bentonville, Arkansas-based company is requiring workers in its Dallas and Atlanta tech hubs to either relocate or face layoffs. It’s not clear how many workers will ultimately be affected by the decision. It was the gaming and human…

24 May

RAGE Against the Machine: Retrieval-Augmented LLM Explanations

stp2y0 CommentsViral News

arXiv:2405.13000v1 Announce Type: new Abstract: This paper demonstrates RAGE, an interactive tool for explaining Large Language Models (LLMs) augmented with retrieval capabilities; i.e., able to query external sources and pull relevant information into their input context. Our explanations are counterfactual in the sense that they identify parts of the input context that, when removed, change the answer to the question posed to the LLM. RAGE includes pruning methods to navigate the vast space of possible explanations, allowing users to view the provenance of the produced answers. Source link lol

24 May

The AI Revolution Will Not Be Monopolized: How open-source beats economies of scale, even for LLMs

stp2y0 CommentsNews

With the latest advancements in Natural Language Processing and Large Language Models (LLMs), and big companies like OpenAI dominating the space, many people wonder: Are we heading further into a black box era with larger and larger models, obscured behind APIs controlled by big tech monopolies? I don’t think so, and in this talk, I’ll show you why. I’ll dive deeper into the open-source model ecosystem, some common misconceptions about use cases for LLMs in industry, practical real-world examples and how basic principles of software development such as modularity, testability and flexibility still apply. LLMs are a great new tool…

24 May

Research Papers in Nov 2023: Tackling Hallucinations, Boosting Reasoning Abilities, and New Insights into the Transformer Architecture

stp2y0 CommentsAI

This month, I want to focus on three papers that address three distinct problem categories of Large Language Models (LLMs): Reducing hallucinations.Enhancing the reasoning capabilities of small, openly available models.Deepening our understanding of, and potentially simplifying, the transformer architecture.Reducing hallucinations is important because, while LLMs like GPT-4 are widely used for knowledge generation, they can still produce plausible yet inaccurate information.Improving the reasoning capabilities of smaller models is also important. Right now, ChatGPT & GPT-4 (vs. private or personal LLMs) are still our go-to when it comes to many tasks. Enhancing the reasoning abilities of these smaller models is one…

24 May

The Onion’s Take on OpenAI’s Scarlett Johansson Disaster Is Pretty Much Perfect

stp2y0 CommentsChat-GPT

Bravo.Stupid SimulacrumThe accusations that Scarlett Johansson has leveled against OpenAI have launched a storm of controversy, turning the ChatGPT creator into a near-pariah overnight for its alleged copying of the actress's voice without her permission.There's already plenty of good writing out there on how the incident encapsulates the AI industry's arrogance and its astounding lack of integrity. But leave it to The Onion, of course, to perfectly sum up the ridiculousness of this whole thing in just one satirical headline."Jerky, 7-Fingered Scarlett Johansson Appears In Video To Express Full-Fledged Approval Of OpenAI," it reads."'It is me, Scar Johnson, to express to…

24 May

The third New England RLHF Hackers Hackathon

stp2y0 CommentsGenAI

At the third New England RLHF Hackathon, several interesting projects were showcased, each focusing on different aspects of machine learning and reinforcement learning. Participants and those interested in future events are encouraged to join the Discord community for more information and updates. Join the discord community The highlighted projects include: Pink Elephants Pt 3 (Authors: Sid Verma, Louis Castricato): This project aimed to train a pink elephant model via ILQL (Inverse Learning from Q-learning), using the standard trlX implementation. The team faced challenges in finding optimal hyperparameters and proposed future research that includes more nuanced reward shaping and combining different…

24 May

FasterViT for Image Classification

stp2y0 CommentsViral News

FasterViT is a family of Vision Transformer models that is both fast and provides better accuracy than other ViT models. It combines the local representation learning of CNNs and the global learning properties of ViTs. In this article, we will cover the FasterViT model for image classification. Figure 1. FasterViT architecture, throughput, and benchmark on ImageNet1K. We will go through image inference using the pretrained network along with a brief of its architectural components. Furthermore, we will also fine-tune a FasterViT model for image classification. We will cover the following topics in this article We will start with a discussion…

24 May

Language Modeling Reading List (to Start Your Paper Club)

stp2y0 CommentsNews

Some friends and I started a weekly paper club to read and discuss fundamental papers in language modeling. By pooling together our shared knowledge, experience, and questions, we learned more as a group than we could have individually. To encourage others to do the same, here’s the list of papers we covered, and a one-sentence summary for each. I’ll update this list with new papers as we discuss them. (Also, why and how to read papers .) Attention Is All You Need: Query, Key, and Value are all you need* (*Also position embeddings, multiple heads, feed-forward layers, skip-connections, etc.) GPT:…