What's New in NLP? #2
Welcome to the November edition of Cohere’s monthly newsletter, where we share the latest updates in the world of LLMs.
News and Announcements
Hola ??,
In this issue, we share the latest of what’s new in NLP, our recent launch of the?Cohere Sandbox open-source models for developers, Cohere For AI highlights of what’s the latest in language research, and the exciting events we’ll be attending to, including NeurIPS 2022!?
As a?platform that can read and write, we've been named one of Fast Company's 2022 Next Big Things in Tech! ?? As one of 5 winners across AI and Data, recognized for "bringing natural language processing to the little guy." Learn more about the?next 9 big things in AI and data.
What's New in NLP?
In a?recent paper?published by Stanford's Center for Research on Foundational Models, Cohere's xlarge model (with 52B parameters) achieved higher accuracy than a number of well-known commercially available models that are 3x larger, including GPT-3, Jurassic-1, Jumbo, and BLOOM. These results show that model size should not be the only—or even the primary—focus for commercial applications. At Cohere, we know our customers need not only accuracy, but also speed, safety, cost, and ease of use—demonstrating that Cohere can deliver world-class accuracy with a model that's a fraction of the size of most top-tier models.
As you may know, NeurIPS (the Neural Information Processing Systems conference), one of the largest machine learning conferences in the world, is happening this week. If you're in New Orleans, come chat with us at booth #615. We're excited to show you some demos, tell you about our research papers, share information about open opportunities and, of course, give out sick swag!
Meta announced an?AI agent called CICERO, the first AI to achieve human-level performance in the popular strategy game Diplomacy. CICERO uses NLP to negotiate, persuade, and work with human players to achieve strategic goals and win the game.
Stanford’s Center for Research on Foundational Models has developed a new benchmarking approach to evaluate Large Language Models.?The `Holistic Evaluation of Language Models?(HELM) aims to bring transparency to the LLM and NLP space.
The?release of Stable Diffusion 2.0?is making the rounds in the AI image generation circles. Armed with a new and larger language model (OpenCLIP), the community is rediscovering how to best prompt-engineer inputs to this new model, as the old prompts are not producing the same results.??
NVIDIA announced a collaboration with Microsoft to build a?"massive" cloud computer?focused on AI. It will reportedly use tens of thousands of high-end Nvidia GPUs for applications like deep learning and large language models. The companies aim to make it one of the most powerful AI supercomputers in the world.
Google proposed a general paradigm that combines reasoning and acting advances to enable language models to solve various language reasoning and decision-making tasks.?ReAct?enables language models to generate both verbal reasoning traces and text actions in an interleaved manner.
Top Demos of the Month
Get building with Sandbox, Cohere open-source libraries:
In this article, we go through the details of an end-to-end application from frontend to backend, including how the Turing team used the Cohere Generate and Cohere Embed endpoints to develop a practical business solution for customer support chatbots.
This tutorial showcases how to use large language models (LLMs) to generate advertising content quickly, specifically product descriptions. Using the artificial intelligence (AI) capabilities of the Cohere Platform and its Generate endpoint, you can create a simple web application that uses keywords in a short phrase to generate multiple product descriptions and automate this repetitive copywriting task.
In this tutorial, we show you how to use Cohere's Python SDK to add NLP functionality to an existing Python application. Then, your app will use Cohere's text summarizer endpoint to automatically summarize user reviews and analyze their overall sentiment and toxicity levels.
领英推荐
Community Contribution of The Month
Shared by Ankit Grover!?Join the Cohere community and share yours. It could be featured in our next issue.
Bidirectional Encoder Representations from Transformers (BERT) is a transformer-based machine learning technique for natural language processing pre-training developed by Google. Following the introduction of BERT, many other language techniques were published including the famous BERT under their model naming conventions.
What's New on the Blog
Last month, co:lab fridays made one heck of a comeback! Not only did we restart our community meetup after a short break, but we added a few brand-new segments to the agenda. Now happening on the last Friday of every month, co:lab fridays is featuring a lineup of demos from our amazing makers around the world.
In our recent NLP Jumpstart Series, “How Startups Can Use NLP to Build a Competitive Moat,” we explore how language AI can fundamentally transform how startups can leverage LLMs to build products and experiences.
In the first episode of the Talking Language AI series, I spoke with Maarten Grootendorst, author and maintainer of the BERTopic open source package (over 3,000 stars on Github). BERTopic is used to explore collections of text to spot trends and identify the topics in these texts. This is an NLP task called Topic Modeling.
Updates From Cohere For AI?
Cohere For AI is a non-profit research lab that seeks to solve complex machine learning problems.?Learn more?about the lab and follow us on?Twitter.
Hosted Events
Conferences & Meetups
Join the Cohere co:mmunity on Discord!
Want to explore more demos and collaborate with other NLP practitioners??Join in on the conversation on our Discord
?Like what you're reading? Share it with your friends and colleagues.