Nils Reimers
@Nils_Reimers
VP AI Search @Cohere | ex-huggingface | Creator of SBERT (http://sbert.net)
𝐕𝐢𝐬𝐢𝐨𝐧 𝐑𝐀𝐆𝐑𝐀𝐆 𝐨𝐧 𝐂𝐨𝐦𝐩𝐥𝐞𝐱 𝐆𝐫𝐚𝐩𝐡𝐢𝐜𝐬🖼️ RAG is mostly text-only, even though we have so much data available as charts/figures. Combine @cohere lastest Embed v4 embedding model with a vision-LLM like @GoogleDeepMind Gemini to get 𝐕𝐢𝐬𝐢𝐨𝐧 𝐑𝐀𝐆
🗽 NYC tech community, come learn how to build agentic AI applications with experts from #AWS, #Cohere, and Pinecone! 🤖 RSVP at: go.aws/4eed3IP Event Details: 📍 Amazon JFK14. 7 W 34th St., New York, NY 10001, USA 📅 Date: Monday, June 30 2:30-6:30 PM
Join us at @Amazon's JFK14 office in Midtown, NYC for "Building Agentic AI Applications with #AWS, Cohere & Pinecone" - a hands-on workshop with experts from @AWScloud, @Cohere, and @Pinecone 📅 Monday, June 30 lu.ma/8rm6zryw?utm_s…
Or, in other words, Gemini2.5 Pro succeeds at 30% of real world office tasks. That's pretty good, considering this is the worst Gemini Pro will ever be.
"Out of thousands of AI agents said to be deployed in businesses throughout the globe, Gartner estimated that "only about 130" are real. The report notes an epidemic of "agent washing," where existing products are rebranded as AI agents to cash in on the tech hype."
As a fan of Encoder models, I really like this type of work.
🚨Should We Still Pretrain Encoders with Masked Language Modeling? We have recently seen massively trained causal decoders take the lead in embedding benchmarks, surpassing encoders w/ bidirectional attention. We revisit whether Bert-style encoders are a thing of the past? (1/N)
‼️Sentence Transformers v5.0 is out! The biggest update yet introduces Sparse Embedding models, encode methods improvements, Router module for asymmetric models & much more. Sparse + Dense = 🔥 hybrid search performance! Details in 🧵
‼️Sentence Transformers v5.0 is out! The biggest update yet introduces Sparse Embedding models, encode methods improvements, Router module for asymmetric models & much more. Sparse + Dense = 🔥 hybrid search performance! Details in 🧵
Microsoft claims their new AI framework diagnoses 4x better than doctors. I'm a medical doctor and I actually read the paper. Here's my perspective on why this is both impressive AND misleading ... 🧵
I love these visualizations of embedding spaces!
Explore Wikipedia through a data map. Pages are grouped by semantic similarity, for topic clusters. Hover to see details, zoom to explore fine-grained topics, click to go to a page. Search page names to find interesting starting points for exploration.🧵 lmcinnes.github.io/datamapplot_ex…
Explore Wikipedia through a data map. Pages are grouped by semantic similarity, for topic clusters. Hover to see details, zoom to explore fine-grained topics, click to go to a page. Search page names to find interesting starting points for exploration.🧵 lmcinnes.github.io/datamapplot_ex…
youtube shorts is the most dangerous brain rot platform with tiktok you know what you’re getting yourself into the second you click the app But shorts shares the same app as youtube so it’s, very easy to want to watch an educational video and get “accidentally” black holed by…
LLMs excel at finding surprising “needles” in very long documents, but can they detect when information is conspicuously missing? 🫥AbsenceBench🫥 shows that even SoTA LLMs struggle on this task, suggesting that LLMs have trouble perceiving “negative space” in documents. paper:…
Exciting to see our advanced retrieval models within Azure AI Foundry. Test Embed 4, to be able to search on your data, in any language, in any format.
Cohere models just leveled up in Azure AI Foundry! You can now deploy Command A, Embed 4, and Rerank 3.5 models on managed compute, bringing scalable performance and more to your GenAI apps. Get started here: msft.it/6011SQO2S
🔹Announcing a new sub-program dedicated to decoding artificial neural networks through the lens of mechanistic interpretability 🔹 Join ML Understanding at @Cohere_Labs to explore how features and circuits emerge, how probing, SAEs, and causal analysis decode model behavior,…
A Canadian success story, @Cohere, is expanding its business in the United Kingdom and working with our government to ensure AI is developed safely and responsibly. A pleasure to welcome its CEO, Aidan Gomez, and Prime Minister Starmer to Parliament Hill today. Together we’re…
We’re proud to partner with the governments of Canada and the UK to accelerate adoption of secure AI solutions in the public sector. Today, our CEO and co-founder @AidanGomez met with @CanadianPM and @10DowningStreet to discuss the strategic importance of AI for national…
Proud to announce agreements with the UK and Canadian governments to strengthen the public sector in both countries with secure AI. We’re excited to grow Cohere’s presence in the UK, promote deeper research ties with researchers across Canada, and help shape the future of AI in…
We’re proud to partner with the governments of Canada and the UK to accelerate adoption of secure AI solutions in the public sector. Today, our CEO and co-founder @AidanGomez met with @CanadianPM and @10DowningStreet to discuss the strategic importance of AI for national…
Exciting to partner with Ensemble, who manage $40 billion in annual net patient revenue, to bring secure AI agents into the healthcare industry.
We’re excited to bring Cohere North to the healthcare industry through our partnership with @EnsembleHP! Together, we’re working to reduce administrative friction and elevate patient experience in hospitals and health systems with our secure AI agents platform.
Every year, Jensen drops new math rules that adds confusion to the whole industry. This year is no different, we now observe a third new Jensen math rule. The first Jensen math rule is that NVIDIA headline FLOPs are quoted with 2:4 sparsity (which no one uses out of China)…
You’re telling me my deleted chatgpt chats are actually not deleted and is being saved to be investigated by a judge?