Catch up on the latest in tech, business and startups - all in 5 minutes. [Newsletter Banner] Daily News Roundup Welcome to the NextBigWhat Daily News! Today's headlines cover Zomato's platform fee increase, India's first leukaemia drug breakthrough, and Sam Altman's departure from OpenAI. 📰 Help us reach your your friends and colleagues! Share the NextBigWhat newsletter. [Tweet it]( [Whatsapp]( MrBeast declines Elon Musk’s invitation to upload videos on X due to financial constraints - YouTube star MrBeast rejected Elon Musk’s proposal to upload his content on X, citing the high costs of video production.
- MrBeast stated that even a billion views on X wouldn’t cover a fraction of his production costs, but hinted at potential future collaboration.
- Musk’s goal is to transform X into an ‘everything app’, with videos playing a crucial role, similar to China’s WeChat. Meta permanently reduces prices for Quest 2 headset and accessories - Meta has permanently reduced the price of its older Quest 2 headset by $50, with the 128GB version now at $249 and the 256GB version at $299.
- Quest 2 accessories, including the Elite Strap with battery, are also being offered at lower prices.
- The price drop may encourage consumers to try VR with the more affordable Quest 2 instead of the $500 Quest 3. Transform your 2024 : Build better habits ! [NBW Newsletter Ad]( 50% Launch Discount ! - Email based delivery - AI-Personalized Learning Reset and Build Better Habits for 2024​ - with our email based course. Get it Zomato increases mandatory platform fee to Rs 4 per order - Zomato has raised its platform fee from Rs 3 to Rs 4, effective from January 1, to improve profitability.
- The fee hike applies to all customers and is effective in 33% of the cities where Zomato operates.
- The platform fee helps Zomato improve its take rate on each food delivery order, making its economics more viable in the long run. Google unveils VideoPoet, a versatile large language model for video generation - Google’s new large language model, VideoPoet, can process various inputs such as text, images, video, and audio to generate videos, marking a significant advancement in video generation.
- VideoPoet’s decoder-only architecture allows it to produce content for tasks it hasn’t been trained on, and it has shown success in various video-related tasks, including text-to-video and video-to-audio generation.
- Despite its current limitation in producing longer videos, VideoPoet demonstrates potential for creative applications, such as creating short films and manipulating object movements in existing videos. Get NBW News: The short news app for busy professionals. [App Store]( [Google Play]( Fidelity slashes valuation of Elon Musk’s X holdings by 71.5% - Fidelity has reduced its valuation of X holdings, Elon Musk’s parent company of the former Twitter, by 71.5% from the original share valuation.
- The company has faced challenges in retaining advertisers after Musk’s controversial comments, leading to a boycott by major firms.
- Despite this, X plans to focus on attracting small and medium businesses, contesting claims of a $75 million loss due to the boycott. Taiwan’s TSMC, Apple’s chip supplier, caught in political crossfire - Taiwan Semiconductor Manufacturing Company (TSMC), Apple’s main chip supplier, is embroiled in a political dispute amid Taiwan’s upcoming presidential elections.
- Vice presidential candidates are debating TSMC’s overseas investments and the impact of China-Taiwan tensions on the company’s future.
- While one candidate warns of potential investment loss due to tensions, the other argues TSMC should not be used for political competition. India’s first oral chemotherapy drug for Leukaemia developed by Tata Memorial Hospital - Physicians from Tata Memorial Hospital, ACTREC, and IDRS Labs have developed India’s first oral chemotherapy drug, PREVALL, for treating Acute Lymphoblastic Leukaemia.
- The drug, which is easy to prepare and administer, offers a more precise and child-friendly alternative to traditional tablets, ensuring accurate dosage and reducing the chance of spills.
- Following its official launch and introduction at Tata Memorial Hospital, PREVALL is set to become accessible in major hospital pharmacies across India, potentially benefiting around 10,000 children diagnosed with ALL each year. Krutrim AI: 5 Questions Ola’s Bhavish has announced the launch of Krutrim AI – claiming it as India’s own AI. Well, we at NextBigWhat have 5 ‘fundamental’ questions to Bhavish and his team. - Considering that the demo showed programming capabilities, and this was not referenced as part of the specialized training, one would assume that an open-source model or a mix of them were used as the base for Krutrim – which ones and why?
- What is the breakdown of the Indic data used for training by language and source? Was copyrighted Indic data used in the training? If so, will the creators be compensated? And if not, what is Ola’s philosophical take on the subject?
- What are the genuinely ‘Indic’ pain points that ‘Indic’ LLMs can solve that go beyond low-hanging generative fruit?
- What were the actual prompts/tests used to benchmark on Indic performance against state of the art models? This was not clear in the presentation.
- Google has the most sophisticated Indic language related technologies on the planet, and they are likely to integrate the capabilities into their new models and be able to optimize at a gigantic scale – how do you intend to compete with that? Is there a holistic thrust towards building language-related technologies that complement these efforts? (By Neeraj, product engg at [AtomicIdeas.AI]( / image [credit]() Top 10 open source OpenAI alternatives Here’s an introduction to the top 10 open-source OpenAI alternatives: 1. LLaMA: LLaMA is a collection of open-source foundation language models developed by Meta Research. It is trained on trillions of tokens and can be used for a variety of tasks, including text generation, translation, question answering, and code generation. LLaMA is available in four sizes: 7B, 13B, 33B, and 65B parameters. 2. OPT: Open Pretrained Transformer (OPT) is a suite of open-sourced large language models developed by Meta AI. It is based on the decoder-only Transformer architecture and consists of eight models ranging in size from 125 million to 175 billion parameters. 3. MPT-7B: An English text and code pretrained from scratch on 1T tokens, MPT-7B is a transformer in the style of a decoder. This model was trained by MosaicML. The Mosaic Pretrained Transformer (MPT) model family, of which MPT-7B is a member, employs a modified transformer design that has been tuned for effective inference and training. 4.  Dolly 2.0: Dolly 2.0 is a 12B parameter language model based on the EleutherAI pythia model family and fine-tuned exclusively on a new, high-quality human generated instruction following dataset, crowdsourced among Databricks employees. 5. GPT-J: GPT-J is another open-source large language model developed by EleutherAI. It is trained on a dataset of 6 billion words and has 6 billion parameters. GPT-J is known for its ability to generate human-quality text. 6. GPT-Neo: GPT-Neo is a family of open-source large language models developed by EleutherAI. It is based on the same architecture as GPT-3, but it is trained on a different dataset and has a different parameter count. GPT-Neo is available in three sizes: 120 million parameters, 1.3 billion parameters, and 2.7 billion parameters. 7. LaMDA: LaMDA is a conversational AI developed by Google AI. It is trained on a dataset of 1.56 trillion words and is known for its ability to engage in open-ended, informative, and comprehensive conversations. LaMDA is not currently available as an open-source model, but it is considered a state-of-the-art conversational AI. 8. Jurassic-1 Jumbo: Jurassic-1 Jumbo is a family of open-source large language models developed by AI21 Labs. It is based on the Jurassic-1 architecture and is available in three sizes: 175 million parameters, 345 million parameters, and 1.75 billion parameters. Jurassic-1 Jumbo is known for its ability to generate creative text formats of text content, like poems, code, scripts, musical pieces, email, letters, etc. 9. AllenNLP: AllenNLP is an open-source natural language processing library built on top of PyTorch. It provides a wide range of pre-trained models and tools for tasks like text classification, language modeling, and more. 10. XLNet: XLNet is a generalized autoregressive pretraining method for natural language understanding (NLU) tasks. XLNet is based on the Transformer architecture, but it differs from other Transformer-based models in that it does not use a fixed factorization order. This allows XLNet to capture bidirectional context, which means that it can take into account both the words that come before a particular word and the words that come after it. What are the other open source OpenAI alternatives you would recommend us to add to the list? Sam Altman, OpenAI co-founder and CEO, fired from his position - Sam Altman, the CEO and co-founder of OpenAI, has been removed from his position by the board, citing a lack of consistent candor in communications. The board stated that Altman’s alleged lack of consistent communication hindered its ability to fulfill its responsibilities.
- Mira Murati, OpenAI’s CTO, will serve as interim CEO. Greg Brockman will step down as the board chair but retain the position of OpenAI president. MakeMyTrip founder Deep Kalra discusses company’s early struggles and importance of unit economics - Deep Kalra, founder of MakeMyTrip, discussed the company’s initial struggles, emphasizing the importance of metrics and unit economics.
- Kalra advised young entrepreneurs to commit to their ventures for at least 4-5 years, highlighting that many successful companies were not built in just 2 years.
- Despite early challenges, including the dotcom bust and SARS outbreak, MakeMyTrip managed to list on NASDAQ in 2010, raising $70 million. Gen Z prefers UPI to credit cards © 2023 Zakti Techmedia Pvt. Ltd. All rights reserved. | [Unsubscribe](