What's New in NLP? #7 Cohere Rerank, LivePerson Partnership and More

Greetings, Language AI Enthusiasts

Are you tired of irrelevant search results in your organization's keyword-based systems??Introducing Cohere Rerank , an innovative solution providing accurate, relevant results. Integrate semantic search technology into your existing system with one line of code, improving search quality in 100+ languages without altering the infrastructure. Experience?Rerank's high-performance solution today !

In other news, we're thrilled to share that for the second year in a row, we've made it to?Forbes AI 50 ! Cohere is working hard to make language AI accessible to businesses and developers alike.??

Next, we are?collaborating with conversational AI company LivePerson ?to enhance Cohere's LLM technology. This partnership focuses on addressing bias, increasing explainability, and minimizing AI hallucinations, aiming to improve how enterprises implement LLMs.

• The advantages of Language AI for businesses are becoming increasingly apparent.?A research report by the NBER ?demonstrates that a contact center experienced a 14% productivity increase by employing a language AI model. Interestingly, lower-skilled workers benefited more than their high-skilled counterparts. We anticipate many more examples of successful LLM implementations in the near future, and we’ll share the highlights with you in this newsletter.?
• A key challenge to improving the performance and visibility of LLMs is the use of current, factual data from external sources (the term is retrieval augmented language models). In the paper,?Shall We Pretrain Autoregressive Language Models with Retrieval? A Comprehensive Study , the retrieval-augmented models significantly outperformed GPT models on knowledge-intensive tasks, as well as other metrics. We expect substantial technological progress in the coming months.
• Another adjacent area in retrieval-augmented generation is generative search, which uses generative AI to create fluent and contextual responses based on user queries. Technology media expects this to be a threat to traditional search engines. This paper?Evaluating Verifiability in Generative Search Engines ?examines and finds these young products to generate fluent text, but they often contain unsupported statements and inaccurate citations, which outlines clear areas for future improvements of these systems.
• In?More Design Patterns For Machine Learning Systems ,?Eugene Yan ?shares a number of design patterns for machine learning systems, which are reusable, time-tested solutions to common problems in software engineering, including Process Raw Data Only Once, Human-In-The-Loop, Data Augmentation, Hard Negative Mining, amongst others.
• Also, don't miss out on Cohere For AI?community's top NLP research picks for March . From embodied multimodal language models like PaLM-E that tackle the challenge of grounding language models in real-world robotic applications to the impressive Vid2Seq architecture that leverages narrated videos for dense video captioning, researchers are working to continuously expand the horizons of NLP.

Dive into the latest demos created by the Cohere community:?

1. JeremyKirshbaum#6893 launched?Library of Babel , which “is somewhere between a prompt Wikipedia and an AI Reddit” for collective prompt exploration. Just drop your Cohere API key in the playground and go?create some prompts !?
2. michaelwechner#6901 built of off?Cohere Wikipedia Embeddings Archives ?and launched?Katie Bot , a Wikipedia-Weaviate-Cohere hybrid that you can now test via?Discord ?and Slack.?
3. On a similar note, Kwadwo#8015 took inspiration from our Champion’s?Elle Neal's work on Cofinder , and expanded it to create?Document Cofinder , a demo that let’s you apply semantic search to your own PDF documents.?Try it out ?and share your feedback!??
4. Our Champion Sonam Gupta wrote an article?Cohere Embeddings with Weaviate Vector Database ?summarizing what she learned from our recent webinar hosted by Cohere and Weaviate with Nils Reimers and Etienne Dilocker.
5. Our Champion Elle Neal?worked on Databutton templates for her AI apps that use Cohere API? (like her Cofinder semantic search app), so you can now use them to get started with your own AI projects with ease!

Join the Cohere community on Discord!

Want to explore more demos and collaborate with other NLP practitioners??Join in on the conversation on our Discord

The Embedding Archives: Millions of Wikipedia Article Embeddings in Many Languages

Gain access to nearly 100 million embeddings for Wikipedia articles, covering not only Wikipedia in English but also various other languages.

Unlocking the Power of Cross-Lingual Classification in NLP

Learn how to use Cohere's multilingual embedding model that enables cross-lingual text classification for various applications like sentiment analysis, content moderation, and intent recognition across 100+ languages using a single language's training data.

What Are Transformer Models and How Do They Work?

Learn about the architecture and functioning of Transformer models. This article, with multiple diagrams, explains the fundamentals of transformers in an easily readable way.???

READ MORE

Updates From Cohere For AI?

• ICLR 2023: We’re excited to showcase our latest research in our main track and workshop papers and looking forward to connecting with our community! Attending in person? Stop by the Cohere booth to say hello and pick up limited-edition swag!
• New Research:?On the Challenges of Using Black-Box APIs for Toxicity Evaluation in Research?was published this week. Learn more about the work?here. ?

Join our community! We want to keep growing our?fantastic events and programs ?and provide new opportunities for aspiring researchers to get involved with our Open Science research project.?Complete the community application form ?to join in.

JOIN COHERE FOR AI

??? May 11, 1:00 pm - 2:00 pm EDT

The 2023 Zeitgeist Report - Generative AI and Best Practices for Enterprise Adoption?

Explore the Scale Zeitgeist AI Readiness Report, surveying 1.6k+ execs & ML pros on challenges, LLMs' impact & real business applications.

SIGN UP

??? May 26, Friday - 12 pm ET

co:lab friday 4.0 – Community x Cohere

This colab friday demo showcase features outstanding Language AI demos.

SIGN UP

We want to make NLP useful and accessible to anyone who needs it.

Whether you're a beginner or an expert, we're here to help.?

REGISTER NOW