🔥 LLMs on Fire 🔥

Can't read or see images? View this email in a browser

 THE WEEKLY NEWSLETTER OF AIM.  Monday, Apr 15, 2024 | Was this email forwarded to you? Sign up here  By Amit Naik   A few days ago, OpenAI's GPT-4 Turboreclaimed the top spot on the LMSys Chatbot Arena Leaderboard after briefly trailing behind its competitors Anthropic’s Claude 3 Opus and Google’s Gemini Pro. Cohere, aka the secret superstar of LLMs, also made it to the top spot with Command R+. Surprisingly, Databricks’ open model, DBRX, which claims to rival advanced models such as GPT-4, has yet to reach the top of the leaderboard.   This is just a glimpse of what’s to come as Llama 3 and GPT-5 are poised to be launched this summer, heating up the LLM race to the next level.    Among all the LLMs that made it to the leaderboard, Cohere’s Command R+ and Mistral Large clearly stand out, catching up in performance to more established, proprietary models, which is a significant step forward for open-source AI development, alongside setting the bar for Llama 3 and GPT-5 going forward.   Besides Cohere and Mistral, open models by Alibaba, Nexusflow, and Microsoft also serve as potential alternatives for many developers and companies experimenting with building LLM-based applications. Open models by Databricks, Google, Meta, and DeepSeek are also emerging as promising alternatives to many enterprises.   While the effectiveness of benchmarks like LMSys might not be accurate as they are likely inherently limited by human evaluation capabilities, the platform helps understand the current advancements in the LLM landscape at a global scale.    Indic LLM Leaderboard?  The wait for the Indic LLM leaderboard is over. Recently, a 20-year-old founder and AI researcher at CognitiveLab, Adithya S Kolavi, has built an Indic LLM Leaderboard to assess open-source LLMs. Read more here >> Unsurprisingly, India’s open-source AI revolution has already started, and 20-year-old developers are leading it. Read the full story here. Register for our upcoming Masterclass: Automating Data Pipelines with Snowflake: Accelerating Data Integration and ETL Workflows >>  Hindi Mein LLM In the latest episode of Analytics Guru, AIM explores LLMs, detailing their function, training, and applications, and examines the future impact of AI technology on our daily interactions.  AI Backbone  Today, hundreds of companies are advancing generative AI by developing foundational models, providing infrastructure and sourcing talent alongside collectively driving rapid innovation in the field.  Here are some of the notable AI companies: (See image below) Also, check out the list of top LMS platforms that are perfect for enterprises looking to enhance their AI capabilities here >>  Bhasha Techathon  Join India’s most exciting hackathon, Bhasha Techathon, a collaborative hackathon organised by Machine Hack, Digital India Bhashini Division and Google Cloud to innovate technology solutions for Indian languages, focusing on challenges in NLP.   What are you waiting for? Register now.  Turing Times Avi Wigderson, a prominent computer scientist, won the prestigious 2023 Turing Award for his transformative contributions to understanding randomness in computation and his leadership in theoretical computer science. Check out his journey here >>  Data Dialogues In an exclusive interaction with AIM, Karthikeyan Viswanathan, the chief of DataSwitch, takes us through the journey of integrating data from various sources, and enhancing data-driven decision-making, alongside discussing the challenges posed by fragmented data systems that hinder business analytics. Read the full story here.  Binge on AI Chips  Google recently introduced a new Arm-based processor, Axion, which offers a 30% improvement in performance compared to general-purpose Arm chips and a 50% increase over the current generation x86 chips made by Intel and AMD.  Google also introduced the TPU v5p chip, configured to operate in large-scale pods containing up to 8,960 chips, providing double the raw performance of the previous generation of TPUs.  Meta recently unveiled its next-generation AI chip, the Meta Training and Inference Accelerator (MTIA), which more than doubles previous solutions' compute and memory bandwidth, specifically enhancing the efficiency of ranking and recommendation models in Meta's apps and services. Intel's new Gaudi 3 AI accelerator chip, revealed at the Vision 2024 event, claimed a 50% faster performance on AI language models than NVIDIA’s H100.  AI Nuggets  TCS records AI and generative AI pipeline worth $900 million in the recent quarter of FY24 Q4.  xAI recently launched Grok-1.5V, which outperformed GPT-4V in various benchmarks, with capabilities in processing and translating complex visual information into executable code.  Ola Krutrim creates history by running its AI operations on its own cloud infrastructure instead of using providers like AWS or Azure. Know more here. WhatsApp is testing the Meta AI chatbot in India, allowing select users to interact with an AI assistant to answer queries. NaMo App introduces NaMo AI, a unique chatbot that provides information on PM Modi and government schemes.

                                                           

This email was sent by support@analyticsindiamag.com to alexvarboffin.abbb@blogger.com Not interested? Unsubscribe | Manage Preference | Update profile Analytics India Magazine | 280, 2nd floor, 5th Main, 15 A cross, Sector 6, HSR layout Bengaluru, Karnataka 560102