Welcome to the latest Marvik Digest 🚀  

This month we have some interesting stories involving large language models, the challenge of managing AI projects, a deep dive into Marvik and how we approach problems as a team, and more.  

➡️ Want us to cover a specific topic? DM or ping us to [email protected] to send us your suggestions. Stay tuned!

Marvik in a Nutshell

At Marvik we are a machine learning company. But there’s more to it than meets the eye 👀  

In our latest blog entry, you will get a glimpse of what Marvik is, what we work at, how we approach problems as a team, and the way in which we grow our partnerships with clients and allies.  

👉 Click here to get to know us and learn how we could add value to your business

Bloom

BigScience’s BLOOM is finally out 🌸  

The result of this collaborative initiative led by Hugging Face is the largest open-access multilingual language model ever.  

Main Takeaways:

• An autoregressive Large Language Model (LLM), trained to continue text from a prompt on extensive text data using industrial-scale computational resources.
• 176 billion parameters → can generate text in 46 natural languages and 13 programming languages.  
• Any individual or institution can use and build upon these large language models on a local machine or on a cloud provider → an attempt to democratize its use
• It is the result of the largest collaboration of AI researchers ever involved in a single research project → +1000 researchers from 70+ countries and 250+ institutions  
• Supported by institutions such as Centre national de la recherche scientifique, NVIDIA, GENCI, Microsoft.  

➡️ Find out more here: https://bit.ly/3axAgtb  

➡️ To access the model, click https://bit.ly/3yWRwl4

Our female team keeps growing

To give visibility is to empower 💪 🎉 This week we hit a milestone that we are thrilled to share with our community.  

Our team is now 26% female 🙌 which is higher than the average in the machine learning field. We have a vibrant team of women coming from different backgrounds that keeps growing and growing.  

Still, there is a long way to go 🛣  Beyond just numbers, at Marvik we are putting consistent efforts into bridging the gender gap. We have amazing career opportunities available to women in this exciting industry 🚀  

Are you interested in becoming part of our team?  

👉 Check out our open positions here

No Language Left Behind

Another major milestone for Large Language Models  

‍Meta AI’s No Language Left Behind (NLLB200) initiative is an effort to develop high-quality machine translation capabilities for most of the world’s languages.  

Highlights:  

• The first of its kind to translate across 200 different languages.  
• Achieves state-of-the-art quality, validated through extensive evaluations for each individual language.  
• Created a new evaluation dataset (FLORES-200) to measure the model’s performance and confirmed NLLB-200 exceeds the previous state of the art by around 44%.  
• Open-sourced NLLB-200 models, datasets and code - everyone can now access to build on their work.  
• Various findings of the projects are already being used to improve and extend translations on Facebook, Instagram, and Wikipedia  👉 More on NLLB-200 https://bit.ly/3uZnbjk  

👉 Access the GitHub repo https://bit.ly/3OebXy5

Uncertainty When Managing AI Projects

In our latest blog, our COO Gaston Arismendi discusses the art of mastering uncertainty when managing AI projects. “Beyond being a newer industry and not having the same level of maturity as traditional software, what differentiates AI as discipline could be pictured with only one word: uncertainty. This concept is present at many levels of an AI project.”  

👉 Visit our blog for the full story here

👉 At Marvik, we have dealt with ai projects across several industries. DM or reach out to [email protected] if you are curious about how we deal with this uncertainty so as to deliver high quality results to our clients.  

‍