#001 Tech Mahindra Leads The Way In AI, Hugging Face raises Series D of $235 million, Fine-Tune GPT-3.5 with Scale AI, Meta’s New Open-Source AI Tool - Code Llama
Fresh & Hot curated AI happenings in one snack. Never miss a byte 🍔
This snack byte will take approx 5 minutes to consume.
AI BYTE # 1 - How Tech Mahindra is Leading the Way in Generative AI Among Indian IT Giants 🚀
⭐ Generative AI is one of the most exciting and promising fields of artificial intelligence, enabling machines to create novel and diverse content such as text, images, music, and code.
It has immense potential to transform various industries and domains, from entertainment and education to healthcare and e-commerce.
However, developing and deploying generative AI applications is not an easy task. It requires a lot of data, computing power, and expertise.
That’s why many Indian IT companies are partnering with global tech giants like Google, Microsoft, and AWS to leverage their generative AI platforms and solutions.
But there is one company that stands out from the crowd: Tech Mahindra.
Tech Mahindra is the only major Indian IT company that is working on its own proprietary large language model (LLM) called Project Indus.
Project Indus aims to speak over 40 Indic languages in the first phase, including some rare and endangered ones.
It is an open-source initiative that invites people from different regions and dialects to contribute their speech data to train the model using natural language processing (NLP) algorithms.
Project Indus is a remarkable and ambitious project that showcases Tech Mahindra’s vision and leadership in generative AI.
It also reflects the company’s commitment to preserving and promoting India’s linguistic diversity and cultural heritage. Project Indus is not just a technological innovation, but also a civilizational one.
AI BYTE # 2 - Hugging Face Democratizing AI for Everyone with a Series D raise of $235 million, valuing the company at $4.5 billion.
⭐ This is a remarkable achievement for a startup that started as a chatbot app for teenagers and pivoted to become an open platform for AI development.
Hugging Face offers a range of data science tools, such as a hub for AI code, models, and datasets, libraries for dataset processing and model evaluation, and web apps to demo AI-powered applications.
It also provides paid features, such as AutoTrain, Inference API, and Infinity, that help developers train, host, and scale their AI models.
Hugging Face is not only a platform for AI development but also a community for AI research and innovation. It has launched several open-source projects, such as BigScience, Bloom, StarCoder, and SafeCoder, that aim to advance the state-of-the-art in natural language processing, code generation, and AI ethics.
It has also partnered with major cloud providers, such as Google, Amazon, Nvidia, and Microsoft, to extend its products and services to their customers.
Hugging Face is on a mission to democratize AI and make it accessible to everyone. It believes that AI is the new way of building all software and the most important paradigm shift of the decade.
If you are interested in joining this movement, check out Hugging Face’s website and follow them on LinkedIn. You can also join their 10,000 customers and 50,000 organizations on their platform and start building your own AI solutions today.
AI BYTE # 3 - How to Fine-Tune GPT-3.5 for Your Business Needs with Scale AI
⭐ If you are looking for a way to customize the powerful GPT-3.5 large language model (LLM) for your specific use cases, you might be interested in the new partnership between OpenAI and Scale AI.
They have announced that Scale AI is the preferred partner for fine-tuning GPT-3.5, which means you can leverage their expertise and data engine to create state-of-the-art models that suit your business needs.
Fine-tuning is a process of training an existing model on your own data to improve its performance and accuracy.
It can help you achieve better results than just using the stock model with prompts. For example, fintech company Brex has been using fine-tuning to generate high-quality expense memos that ease the compliance burden for employees.
They have reported that the fine-tuned GPT-3.5 model outperformed the stock model 66% of the time while reducing cost and latency.
Scale AI is a leading provider of data labeling and model optimization services. They have been fine-tuning many commercial and open-source models, including GPT-4.
With their partnership with OpenAI, they can now offer fine-tuning for GPT-3.5 using their data engine, which generates prompts and ranks model outputs. This can help you create custom models that are tailored to your domain and task.
AI BYTE # 4 - Code Llama: Meta’s New Open-Source AI Tool For Generating And Explaining Code.
⭐ Meta, the company formerly known as Facebook, has recently released Code Llama, a machine-learning system that can generate and explain code in natural language.
Code Llama is based on Llama 2, a large language model that Meta open-sourced earlier this month. Code Llama can complete code and debug existing code across a range of programming languages, such as Python, C++, Java, PHP, Typescript, C# and Bash.
Code Llama is designed to support software engineers in all sectors, from research and industry to open-source projects and NGOs.
It can also help non-programmers learn coding concepts and syntax by providing natural language explanations.
Code Llama is available in several flavors, including a version optimized for Python and a version fine-tuned to understand instructions.
Meta claims that the 34 billion-parameter model is the best-performing of any code generator open-sourced to date.
However, Code Llama is not without its challenges and limitations.
Meta admits that the model might generate inaccurate or objectionable responses to some prompts and that developers should perform safety testing and tuning before deploying any applications of Code Llama.
Moreover, Code Llama might raise some ethical and legal issues regarding the ownership and quality of the generated code, as well as the potential misuse of the tool for malicious purposes.
Code Llama is an impressive example of how generative AI can be applied to coding tasks. It also demonstrates Meta’s commitment to making its AI models more accessible and transparent to the public.
However, it also raises some important questions about the benefits and risks of using AI tools for software development.
A bored 13-year-old Magnus Carlsen against Gary Kasparov.
The reactions of Magnus at this young age and the pain felt by Gary can never be replicated by any AI Robot
Excellent initiative and superb narration of the latest happenings in the field of AI!!!