Piotr Wiercinski

Thank you, Sion Smith, for having us! 🎙 In this talk, Yingjun Wu introduces RisingWave, a scalable streaming database that utilizes #S3 as its primary storage. From underlying technology to real-world scenarios, Yingjun illustrates how RisingWave supports heavy traffic workloads in enterprise-grade production environments. #risingwave #streamingdatabase

3

Delivering personalised recommendations is key to engaging our customers, but traditional models can miss the complexity of user-content interactions. By integrating graph neural networks (GNNs), we’re exploring a cutting-edge approach to better predict clicks and enhance the shopping experience. In her post, Senior Applied Scientist Mariia Bulycheva explains how GNN features were integrated into the models powering the Zalando Homepage: https://bit.ly/4fNnQZC #ZalandoTech #InsideZalando #ZalandoScience #RecSys

43

1 Comment

We recently held the BCI and WBE for Safe AI Workshop. Diogo Di Lucena's (AE Studio) talk reviewed 'Self Modeling in Neural Networks.' Di Lucena discussed how he and his collaborators implement self-modeling in neural networks, inspired by Graziano's attention schema theory, to reduce model complexity and foster prosocial behavior in multi-agent systems. By creating simplified models of one's attention and the exact mechanisms used to model others' attention, the team demonstrates that self-modeling reduces complexity, making it easier for agents to model the system and potentially enhancing prosocial behavior. Evaluating their method on various tasks and models, the authors find consistent results in complexity reduction, suggesting that their approach, grounded in neuroscience, is distinct from other regularization methods and applicable across diverse contexts to improve model complexity and prosocial behavior in multi-agent settings. Learn more: https://lnkd.in/gFtdec-m

2

Our episode on the The Data Exchange Podcast with our fantastic host Ben Lorica 罗瑞卡 , and featuring my great colleague Hima Patel , on data preparation for LLM applications (RAG, fine-tuning, instruct-tuning, and pre-training) using the Data Prep Kit is out! Data Prep Kit is the open-source (Apache 2.0 license) framework and transforms for language and code, which we used in IBM Research to prepare the data for the training of #IBM 's #Granite models ( https://lnkd.in/eUvZVK2t ). Running at any-scale, from a laptop to a datacenter and handling from a few pages of text to hundreds of billions of documents, it is available in pure-#Python , #Ray , #Spark and with #KubeFlow pipelines. Feel free to check it out here: https://lnkd.in/ehGjpbzG #IBMResearch #AI #GenAI #LLM

42

2 Comments

Back to basics: From normalization to dimensional data modeling. Check out my latest blog post covering the essentials of data modeling. In the Konstellation blog. https://lnkd.in/etCdwwwm #DataEngineering #DataManagement #DataArchitecture

18

1 Comment

"...embeddings enable machine learning models to find similar objects." Read more of Gregory Hodum's article on embeddings https://lnkd.in/eFs6hhAP #ai #softwaredevelopment #embeddings #machinelearning

9

The days of using machine learning for conversational data are numbered. In the latest Dashbot blog, NLP engineer Seamus Healy-Singh dives headfirst into the wild world of large language models (LLMs), those unruly beasts that are tearing apart the old ways and bringing in a new era of data science. Forget topic modeling; it’s as outdated as a rotary phone. LLMs are here, and they’re not just more flexible—they’re a revelation, a revolution in scale and capability. Strap in and join us on this ride through the seismic shift in data analysis. We break down why LLMs are outpacing traditional ML methods and how they’re reshaping the landscape. If you're clinging to the old ways, it’s time to wake up and smell the future. Dive into our latest post and see why LLMs are the new kings of data. #ML #MachineLearning #LLMs

5

In context of debate if large context #LLM models will eventually negate the need for #RAG, an article from Databricks has some good insights - • Longer contexts boost RAG accuracy... to a point • Performance often peaks at 32k-64k tokens, then declines • Top models (GPT-4, Claude 3.5) maintain consistency at scale • Others show unique failure modes (e.g., copyright concerns, summarization instead of Q&A) • Optimal context size varies by model and task • Lack of long-context training data may explain some issues Key takeaway: Long context and RAG are synergistic, but LLMs need refinement to fully leverage extended contexts. https://lnkd.in/eehdKtCm

17

I've been beating the drum for some time now on language models' limitations with respect to numerical analysis. Not that there's not a place for them, but their contributions should be carefully tailored to account for their inherent strengths and weaknesses. Mingtian Tan and coauthors released a paper recently that represents another helpful contribution towards the quantification of LLMs' limitations in performing analytical tasks. Specifically, they examine language model performance with time series forecasting. TL;DR? Let's just say they find it lacking. If you are focused on the overlap between language models and numerical analysis, this is worth a read both for the value in the conclusions they reach and some of the intuition embedded in their experiment design.

20

📝 The release of Meta's Llama 2 model last year prompted a wave of excitement over 'open' AI models, with a host of major companies having launched their own since, including Mistral and Databricks. 🔎 But when it comes to AI models, there's a stark difference between 'open' and 'open source', writes ITPro's George Fitzmaurice. 🧠 George caught up with Amanda Brock, CEO at OpenUK, who said that while open AI development should be welcomed, it's about time firms started making things clear for users. ❌ Many 'open' LLMs, including Meta's new Llama 3 model, come with restrictions that are shrouded amidst the hype of a new launch. 👀 Read more here #AI | #OpenSource | #LLMs

1

🚀 Want to employ the power of #PyTorch for your scientific models without a steep learning curve? 👀 Check out our latest blog, where Travis Oliphant discusses how PyTorch's integration with the Array API makes it easy to use with familiar libraries like NumPy and scikit-learn.🔥 https://lnkd.in/dDRkEY3v #Innovation

10

📣 Speaker Announcement: MLOps London 📣 We're thrilled to welcome Suyash J., Senior Developer Advocate at InfluxData, to teach us about Harnessing Time Series Data for MLOps: Anomaly Detection, Forecasting, and LLM Applications ✨ RSVP now to save your spot at Meetup: https://lnkd.in/edQDgJWQ 📅 When: January 28th 📍 Where: Rise London, Shoreditch (in person & online) In this talk, we will cover topics like using time series databases like InfluxDB for monitoring, anomaly detection, and forecasting, implementing Facebook’s Prophet and Forest, and live code demo showcasing real-time anomaly detection, forecasting trends, and analyzing time series metrics with InfluxDB. Whether you're a data scientist, engineer, or just curious about the latest innovations in MLOps, don't miss this deep dive into how time series data can transform your MLOps practices. #MLOpsLondon #TimeSeriesData #AnomalyDetection #Forecasting #LLM #InfluxDB #MLOps

9

🚀 Looking to streamline fine-tuning for models like Llama 3, Mistral, and Mixtral? Our latest blog shows how Anyscale simplifies LLMs fine-tuning with Ray: 🔑 Key takeaways: • Scalable fine-tuning with LLM-Forge • Model Deployment with RayLLM Read more: https://lnkd.in/gQwD4A2U

35

💡 GPU? Lazy eval? No problem! 💻 Write dataframe code that works beautifully across pandas, Polars, cuDF & more with Narwhals. 🔥 Check out our in-depth blog, by Marco Gorelli on making data science more interoperable: https://lnkd.in/dedkSPPy #Polars #DataFrame #Pandas #Narwhals

14

Working with large language models has become a critical part of any data scientist’s or ML engineer’s job, and fine-tuning the large language models can lead to powerful improvements in the language models’ capabilities. Eivind Kjosbakken's article shows you how you can fine-tune Llama3.2 to improve its performance within a targeted domain. #LLM #Llama

13

Fine-tuning LLMs with Reinforcement Learning from Human Feedback This blog post explores how Reinforcement Learning from Human Feedback (RLHF) can be used to fine-tune Large Language Models (LLMs) to align with human preferences. It delves into the concept of Proximal Policy Optimization (PPO) and how it is used in this context. The key takeaway is that RLHF offers a powerful framework to train LLMs that can produce human-aligned text, while still maintaining creativity. This paves the way for more natural and nuanced language generation in the future of AI and natural language understanding. Read the article: https://lnkd.in/d9EGf579 #AI #machinelearning #goML

8

🔢 Exciting new blog from Olly Britton, a #QuansightLabs summer intern! He's added support for multidimensional numerical integration of array-valued functions to SciPy. This fills an important gap in SciPy's integration toolkit! Read his about it here: 👉 https://lnkd.in/dNmHTX3G #OpenSource #Python #SciPy

12