This Week I Learned - Week #46 2023
This Week I Learned -
* PartyRock is a fun way to learn the fundamentals of foundation models and then collaborate with other developers. Anyone can access PartyRock through the website, and you don’t need an Amazon Web Services (AWS) account. PartyRock uses foundation models from Amazon Bedrock to turn your ideas into working PartyRock apps. You don't need to be a developer or ML engineer to start creating your own apps. Just write a text-based prompt to describe what you want your app to do in the PartyRock app builder.
* Amazon Bedrock is a fully managed service that makes foundation models (FMs) from Amazon and leading AI companies available on AWS available through an API. It is the easiest way to build and scale generative AI applications with foundation models. PartyRock provides builders with access to FMs from Amazon Bedrock in a code-free app-building playground to learn the fundamentals of prompt engineering and generative AI.
* Virtual Assistants (VA) are built by stringing together multiple kinds of models: speech recognition (speech-to-text), natural language understanding, and speech synthesis (text-to-speech). The errors can be in any/several of these parts. UX plays a critical role in the success of a VA in two important aspects:
1. Graceful handling of errors (understanding from user interaction that it has made an error)
2. Easy way to collect user feedback smoothly weaved into UX design.
* How do Large Language Models (LLMs) like ChatGPT work on the inside? How to introduce your own custom data into LLMs using OCI GPUs and open source technologies? That's explained in this YouTube video from the Oracle Developers channel
* A low-code approach to building a ChatGPT app with custom data
* Gradio is an open-source Python frontend framework that makes it easy to build quick interfaces like chatbots, voice-activated bots, and even full-fledged web applications to share your machine learning model, API or data science workflow with clients or collaborators. Chatbots are a popular application of Large Language Models (LLMs). Because the interaction with LLMs feels natural and intuitive, businesses are turning to conversational interfaces such as voice-activated chatbots or voice bots. Gradio can be used to build chatbots that can answer user questions using a variety of data sources.
* AI21 Studio Playground allows developers to create and test their models through a web-based interface. The code editor and model training output are both available in the main window, where programmers can write Python code and use the AI21 Studio APIs. Access to data sets, training material, and documentation are available on the sidebar. The Playground has an interactive debugger that aids developers in finding and fixing issues. AI21 Studio provides a range of capabilities, including automatic model versioning, model management, data exploration, and monitoring, to assist users in managing their models and data. You can choose from AI21's 4 models with varying parameters and customize Jurassic-1 (J1) model:
- j1-large (7.5B)
- j1-grande (17B)
- j1-grande-instruct (beta) (17B)
- j1-jumbo (178B)
* Oracle has invested in generative AI company Cohere, in a deal that valued the business at around $2.2 billion. Oracle's investment is said to loosely mirror that of Microsoft with OpenAI and Google with Anthropic, as tech giants look for promising startups in the burgeoning generative AI space. While Microsoft has exclusive rights to resell OpenAI products on Azure, Cohere has publicly said that it is "cloud agnostic." The company aims for its models to also work on private clouds or with on-prem deployments. Cohere's models are already available on Google Cloud and Amazon Web Services. - DCD
* Since around 2009, GPUs have been essential for deep learning, driving innovative breakthroughs like Alex-Net in 2013, which was built on the CUDA API for NVIDIA GPUs. Hardware is growing more heterogeneous, with GPUs firmly established for graphics processing first and then becoming necessary for the performance demands of data science, new CPU architectures like ARM, and even new kinds of silicon specifically for deep learning. In our current era of AI enthusiasm, the co-evolution of hardware and software has become crucial. - Hardware > Software > Process report
* Free Manning Publications books sponsored by its partners.
* WhatsApp messages can now be edited for up to fifteen minutes after you've posted.
* FTX, which had previously held a prominent position among cryptocurrency exchanges in terms of trading volume, experienced a significant setback, particularly in the realm of Non-Fungible Tokens (NFTs). The repercussions of the FTX crash served as a wake-up call for governments globally, highlighting the necessity for more stringent regulations in the cryptocurrency space. This incident contributed to increased scrutiny by regulators on Binance, the largest Bitcoin exchange, leading to its withdrawal from certain countries like Canada and the Netherlands due to regulatory pressures. Notably, Binance continues to operate without a formal headquarters, adding a layer of complexity to regulatory oversight. Meanwhile, the founder of FTX, Bankman-Fried, faced legal challenges, culminating in his conviction on seven counts of fraud and conspiracy after a month-long trial. These events underscore the evolving landscape of cryptocurrency exchanges and the growing attention from regulatory authorities seeking to establish a more secure and accountable crypto ecosystem.
* Bankman-Fried graduated from the Massachusetts Institute of Technology in 2014 with a bachelor's degree in physics and a minor in mathematics. As an MIT student he lived in a coeducational group house called Epsilon Theta. In January 2018, Bankman-Fried organized an arbitrage trade, moving up to $25 million per day to take advantage of the higher price of bitcoin in Japan compared to the United States. After attending a cryptocurrency conference in Macau in late 2018, he moved to Hong Kong. Bankman-Fried founded the FTX cryptocurrency derivatives exchange in 2019. In September 2021, Bankman-Fried and the entire senior staff of FTX moved from Hong Kong to the Bahamas due to its friendly regulatory environment. In November 2022, Bankman-Fried, who had been listed the year before among Forbes 30 Under 30, resigned as CEO of FTX and FTX declared bankruptcy. In December 2022, Bankman-Fried was charged with "wire fraud, wire fraud conspiracy, securities fraud, securities fraud conspiracy, and money laundering". He was released from prison on a $250 million bond, the largest such bond ever set in an American criminal proceeding. In July 2023, allegations emerged that Bankman-Fried had considered "purchasing" the island country of Nauru to use as a bunker in the event of an apocalyptic event, in what has been described as a "misguided and sometimes dystopian" project. His bail was revoked in August 2023 due to witness tampering. The 31 year old Bankman-Fried's counsel averred that his client was not being provided a vegan diet in jail and that his medications for ADHD and depression were running low.
*The average attrition rate for Indian IT companies stands at 20%. A factor that contributes to the high attrition rate in Indian IT companies is the lack of opportunities for growth and development, alongside stagnant work culture.
* Informal definition of Ultra-Processed Food: if it’s wrapped in plastic and has at least one ingredient that you wouldn’t find in your kitchen, it’s UPF. - Ultra-Processed People
* Happilo makes ₹500 Cr revenue selling dry fruits & nuts. Their goal is to capture the ₹9,000Cr dry fruits category & build a monopoly.
Comments
Post a Comment