Shows
How AI Is Built#022 The Limits of Embeddings, Out-of-Domain Data, Long Context, Finetuning (and How We're Fixing It)Text embeddings have limitations when it comes to handling long documents and out-of-domain data.Today, we are talking to Nils Reimers. He is one of the researchers who kickstarted the field of dense embeddings, developed sentence transformers, started HuggingFace’s Neural Search team and now leads the development of search foundational models at Cohere. Tbh, he has too many accolades to count off here.We talk about the main limitations of embeddings:Failing out of domainStruggling with long documentsVery hard to debugHard to find formalize what actually is similarAre you still not su...2024-09-1946 min
How AI Is BuiltSeason 2 Trailer: Mastering SearchToday we are launching the season 2 of How AI Is Built.The last few weeks, we spoke to a lot of regular listeners and past guests and collected feedback. Analyzed our episode data. And we will be applying the learnings to season 2.This season will be all about search.We are trying to make it better, more actionable, and more in-depth. The goal is that at the end of this season, you have a full-fleshed course on search in podcast form, which mini-courses on specific elements like RAG.We will be...2024-08-0804 min
Weaviate PodcastThe Future of Search with Nils Reimers and Erika Cardenas - Weaviate Podcast #97!Hey everyone! I am SUPER excited to publish our 97th Weaviate Podcast on the state of AI-powered Search technology featuring Nils Reimers and Erika Cardenas! Erika and I have been super excited about Cohere's latest works to advance RAG and Search and it was amazing getting to pick Nils' brain about all these topics!
We began with the development of Compass! Nils explains the current problem with embeddings as a soup!! For example, imagine embedding this video description, the first part is about the launch of a podcast, whereas this part is about an...2024-06-1159 min
PetropoddenNils Reimers, TomaxNils Reimers, Tomax by Petro.no2024-03-1227 min
Learning from Machine LearningLewis Tunstall: Hugging Face, SetFit and Reinforcement Learning | Learning from Machine Learning #6This episode features Lewis Tunstall, machine learning engineer at Hugging Face and author of the best selling book Natural Language Processing with Transformers. He currently focuses on one of the hottest topic in NLP right now reinforcement learning from human feedback (RLHF). Lewis holds a PhD in quantum physics and his research has taken him around the world and into some of the most impactful projects including the Large Hadron Collider, the world's largest and most powerful particle accelerator. Lewis shares his unique story from Quantum Physicist to Data Scientist to Machine Learning Engineer. Resources to learn...2023-10-031h 18
The Alldus Podcast - AI in ActionE477 Nils Reimers, Director of Machine Learning at CohereToday's guest is Nils Reimers, Director of Machine Learning at Cohere. Founded in 2019, Cohere empowers every developer and enterprise to build amazing products and capture true business value with language AI. The company believes that the union of research and product will realize a world where technology commands language in a way that’s as compelling and coherent as ourselves. They live at the forefront of ML/AI research to bring the latest advancements in language AI to their platform.
Nils is a NLP and Deep Learning researcher with extensive experiences on representing text in dense vector sp...2023-09-1819 minWeaviate PodcastNils Reimers on Cohere Search AI - Weaviate Podcast #63!Hey everyone! Thank you so much for watching the 63rd Weaviate Podcast, I couldn't be more excited to welcome Nils Reimers back to the podcast!! Similar to our debut episode together, we began by describing the latest collaboration of Weaviate and Cohere (episode 1, new multilingual embedding models; episode 2, rerankers!), and then continued into some of the key questions around search technology. In this one, we discussed the importance of temporal queries and metadata extraction, long document representation, and future directions for Retrieval-Augmented Generation! I hope you enjoy the podcast, as always I am more than happy to answer any...2023-08-171h 05
The New Stack PodcastWhy Developers Need Vector SearchIn this episode of The New Stack Makers podcast, the focus is on the challenges of handling unstructured data in today's data-rich world and the potential solutions offered by vector databases and vector searches. The use of relational databases is limited when dealing with text, images, and voice data, which makes it difficult to uncover meaningful relationships between different data points.Vector databases, which facilitate vector searches, have become increasingly popular for addressing this issue. They allow organizations to store, search, and index data that would be challenging to manage in traditional databases. Semantic search and Large...2023-07-1827 min
Let's Talk AI#31 - Deep Learning, Research, NLP, Career and Vector Databases with Nils Reimers🎙️ Who is Nils Reimers?
Nils Reimers, a renowned NLP and Deep Learning researcher with extensive experience in representing text in dense vector spaces. He is the creator of SBERT.net and currently serves as the Director of Machine Learning at cohere.ai, where he is responsible for creating the world's best language models for search and content aggregation. Nils also into many others achievements created the science team for Neural Search at HuggingFace from scratch and much more.
💡 In this episode...
Nils discusses his journey into AI and his work on real-time research and search planning...2023-07-0555 min
MLOps.communityLarge Language Models at Cohere // Nils Reimers // MLOps Podcast #158MLOps Coffee Sessions #158 with Nils Reimer, MLOps Build or Buy, Large Language Model at Scale co-hosted by Abi Aryan.
// Abstract
Large Language Models with billions of parameters have the possibility to change how we work with textual data. However, running them on scale at potentially hundred millions of texts a day is a massive challenge.
Nils talks about finding the right model size for respective tasks, model distillation, and promising new ways on transferring knowledge from large to smaller models.
// Bio
Nils Reimers is highly recognized throughout the AI community for creating and maintaining the now-famous Sentence Transformers library...2023-05-161h 14
Das Niederdeutsche HörspielHeinke Hannig: De StubenreisDie Agentur "Minimales Reisen" bietet einen besonders klimaneutralen Service an: Zwei Stunden ungestörtes Reisen durch eine Privatwohnung. Neugierig geworden, buchen Lukas und Sandra dieses Angebot und finden sich in der jeweils anderen Wohnung wieder.
Was eigentlich als harmloser, zeitvertreibender Rundgang geplant war, entpuppt sich nach und nach als eine ganz andere, sehr viel überraschendere und für beide wegweisende Reise.
Mitwirkende
Erkki Hopf: Erzähler
Gesa Retzlaff: Sandra Möller
Nils Owe Krack: Lukas Deichmann
Sonja Stein: Martje Vollmer, Reiseagentin
Marco Reimers: Lasse, Sandras Freund
Birgit Bockmann: Sandr...2023-04-2846 minLearning from Machine LearningNils Reimers: Sentence Transformers, Search, Future of NLP | Learning from Machine Learning #3This episode welcomes Nils Reimers, Director of Machine Learning at Cohere and former research at Hugging Face, to discuss Natural Language Processing, Sentence Transformers and the future of Machine Learning. Nils is best known as the creator of Sentence Transformers, a powerful framework for generating high-quality sentence embeddings that has become increasingly popular in the ML community with over 9K stars on Github. With Sentence Transformers, Nils has enabled researchers and developers (including me) to train state-of-the-art models for a wide range of NLP tasks, including text classification, semantic similarity, and question-answering. His contributions have been recognized by numerous...2023-02-241h 02
Weaviate PodcastNils Reimers on Cohere Embedding ModelsWeaviate podcast #33.
Thank you so much for watching the 33rd Weaviate Podcast! This episode features one of the heroes of Deep Learning for Search, Nils Reimers! Nils' work on SentenceBERT is one of the foundational works for applying Deep Representation Learning to text search. This is the idea that personally inspired me to work in this field. Having seen the successes of Contrastive Representation Learning for Computer Vision, I was mind-blown by the possibility of this for NLP and text search. In addition to the scientific foundation, the software development of the Sentence Transformers library and BEIR...2023-01-1155 min
Vector PodcastMalte Pietsch - CTO, Deepset - Passion in NLP and bridging the academia-industry gap with HaystackTopics:00:00 Introduction01:12 Malte’s background07:58 NLP crossing paths with Search11:20 Product discovery: early stage repetitive use cases pre-dating Haystack16:25 Acyclic directed graph for modeling a complex search pipeline18:22 Early integrations with Vector Databases20:09 Aha!-use case in Haystack23:23 Capabilities of Haystack today30:11 Deepset Cloud: end-to-end deployment, experiment tracking, observability, evaluation, debugging and communicating with stakeholders39:00 Examples of value for the end-users of Deepset Cloud46:00 Success metrics50:35 Where Haystack is taking us beyond MLOps for search ex...2022-08-301h 26
Neural Search Talks — Zeta AlphaOpen Pre-Trained Transformer Language Models (OPT): What does it take to train GPT-3?Andrew Yates (Assistant Professor at the University of Amsterdam) and Sergi Castella i Sapé discuss the recent "Open Pre-trained Transformer (OPT) Language Models" from Meta AI (formerly Facebook). In this replication work, Meta developed and trained a 175 Billion parameter Transformer very similar to GPT-3 from OpenAI, documenting the process in detail to share their findings with the community. The code, pretrained weights, and logbook are available on their Github repository (links below).
Links
❓Feedback Form: https://scastella.typeform.com/to/rg7a5GfJ
📄 OPT paper: https://arxiv.org/abs/2205.01068
👾 Code: https://github.com...2022-06-1647 min
Yannic Kilcher Videos (Audio Only)OpenAI Embeddings (and Controversy?!)#mlnews #openai #embeddings
COMMENTS DIRECTLY FROM THE AUTHOR (thanks a lot for reaching out Arvind :) ):
1. The FIQA results you share also have code to reproduce the results in the paper using the API: https://twitter.com/arvind_io/status/... There's no discrepancy AFAIK.
2. We leave out 6 not 7 BEIR datasets. Results on msmarco, nq and triviaqa are in a separate table (Table 5 in the paper). NQ is part of BEIR too and we didn't want to repeat it. Finally, the 6 datasets we leave out are not readily available and it is...2022-02-1615 minNeural Search Talks — Zeta AlphaThe Curse of Dense Low-Dimensional Information Retrieval for Large Index SizesWe discuss the Information Retrieval publication "The Curse of Dense Low-Dimensional Information Retrieval for Large Index Sizes" by Nils Reimers and Iryna Gurevych, which explores how Dense Passage Retrieval performance degrades as the index size varies and how it compares to traditional sparse or keyword-based methods.
Timestamps:
00:00 Co-host introduction
00:26 Paper introduction
02:18 Dense vs. Sparse retrieval
05:46 Theoretical analysis of false positives(1)
08:17 What is low vs. high dimensional representations
11:49 Theoretical analysis o false positives (2)
20:10 First results: growing the MS-Marco index
28:35...2022-01-2154 min
TechtiefenNLP Update: Attention & Big Science
Natürliche Sprachverarbeitung (NLP) war bereits vor 2 Jahren das Thema einer Techtiefen Podcast-Triologie. Seit dem hat sich aber sehr viel getan, insbesondere Transfer-Learning und die Transformer Technologie haben mächtige Modelle wie Bert oder GPT ermöglicht, es wird also höchste Zeit für ein Update. Wer zuerst noch einmal die Grundlagen auffrischen möchte, dem sei Techtiefen [NLP] Moderne Sprachverarbeitung empfohlen.
Experte dieser Folge ist Nils Reimers, NLP Forscher bei Huggingface und Autor der Sentence-Transformers Library. Er erklärt zunächst die wichtigen Neuheiten der vergangen Jahre wie etwa Transfer Learning, Attention, Transfomer und Neuheiten bei Toke...2021-07-211h 34