Vaiva Augustinaitė

It may seem strange to new guys, but a long long time ago, long long before Python there was C++. And it's still very much alive in libs like quantlib and platforms like Murex flex, as well as being one of the best ways if you're really interested in learning how to think like a computer... kkkkk. Before you ask me why someone would have such a unique desire, let's say it has a lot to do with fashionable expressions like "low latency trading"... Today I just woke up with these peculiar inspirations and as I really like optimization problems (shame on me), I just shared here a small C++ exercise piece: an "exchange-like" order pair matcher, that's actually a multithreading version of the "unsorted greedy" algorithm discussed by Jonsou and Steen (2023) with O(n) [ https://lnkd.in/dTcgWNHn ]. Have fun! https://lnkd.in/dUGM4Vpb

13

Peter Gostev 🔥 Wow, what an intriguing battle of the giants! It's fascinating to see the Elo ratings for the top 5 players in the AI arena over the past year. The inclusion of fresh data for Google models adds even more depth to the analysis. Your approach of using Gemini 1.5 Pro for writing down the lyrics and Suno for generating the voiceover shows innovative thinking and dedication to the project. 🎤 Looking forward to seeing more updates on this exciting competition! #AI #EloRatings #Innovation 💡

1

The video of my talk at C++ on Sea 2024 has been published on YouTube today. It concerns algorithms to convert floating-point numbers to string. How you would go about solving this problem? At first sight, it might look trivial: just a bunch of divisions and modulus by 10. Then, you might realised that for floating-point numbers things are not that easy. You might become hopeless and find it very complicated. That was my journey until I had my Aha! moment. I found a novel algorithm which is simpler and faster than everything I came across before. Curious? If so, watch my talk and if you find worthwhile, please give it a "like" on YouTube and share the video with your friends and colleagues. Note for non C++ programmers: the talk has little to do with C++. It's really about the algorithm which can be implemented in many programming languages and I'm hopeful it will. https://lnkd.in/eP6Nwnyv #cpp #algorithms #mathematics

75

4 Comments

A new paper of mine was published in the Journal of Open Source Software! The paper introduces MarSwitching.jl - The first package for estimating Markov Switching models in The Julia Language. This class of models (and a package) may come in handy if you want to model time-variation or a regime change in the model parameters. I also encourage you to check-out a github repository with the documentation of the package: https://lnkd.in/e3TeNKs5 or a blogpost by Bogumił Kamiński about the package: https://lnkd.in/efd5DCqz

26

Inferring "AI” fast and slow We have so much to gain from AI that takes its time to think - the kind that can handle complex reasoning and sophisticated tooling. Kahneman told us this ages ago, right? System 1 is instinct, blazing fast, while System 2 needs active reasoning and "time"! Turns out, when you let AI spend more time on inference, you get better results. It's called 'scaling test-time computation' - this is how we got OpenAI's o1, and now other similar models are emerging. I surmise that organisations stand to gain much more by working with these new generation of "slow AI" systems. #ai

82

4 Comments

I recommend reading this post if you are interested in use LLM with time series Thanks Pere Martra for your insights about this paper.

5

1 Comment

We found that random projections can cause especially adverse effects on similarities between nodes embeddings, depending on their degree. This has several consequences for applications of graph learning models, including ranking. Fortunately, those pathologies can be avoided by using appropriate similarity functions. Check out the asymptotic and finite-sample analyses on our new pre-print: "Node Similarities under Random Projections: Limits and Pathological Cases".

31

1 Comment

Mom! I'm on YouTube again. This time it's not my own uploaded 😅 PyCon Lithuania uploaded my session (https://lnkd.in/dY2szb_U you can also check out other talks here: https://lnkd.in/dWwxdcWc) Happy watching! ---- Stay in the loop for the latest insights from Uncle Data! * Click the notification bell icon 🔔 to receive updates on Linkedin * Subscribe to my Newsletter on Substack ✏ * Listen to Uncle Data Podcast ⌨️ Listen. Learn. Connect ---- #dataengineering #patterns #python #spark #iceberg

44

I'm starting a new series where I break down new upcoming CV models and explain why they're special/different! More detailed explanation in the comments on how Florence-2 actually works thanks to Piotr Skalski and Roboflow!

84

5 Comments

Dear colleagues and friends, Hereby, I announce that a little open source package, called 𝚜𝚝𝚊𝚝𝚐𝚊𝚖𝚎𝚜 is now publicly available on GitHub: https://lnkd.in/dmzp6HHU by which anybody can easily calculate equilibrium quantities for Fisher, Bayesian and Statistical games, introduced in the arXiv paper: Statistical Games (arXiv:2402.15892) https://lnkd.in/dTkgGnxx Besides the pip installable Python package, there is an interactive Jupyter notebook available on Google Colab: https://lnkd.in/dqWXwr7k This contain the reproduction of many figures from the paper (originally produced by Wolfram Mathematica) with the help of open source plotting libraries and the 𝚜𝚝𝚊𝚝𝚐𝚊𝚖𝚎𝚜 package. (There is also a downloadable static version of the notebook on GitHub: https://lnkd.in/dr4z54Bb ) Feel free to try out the package and the notebook, use them to your advantage, and let me know if you have any questions, comments, or feedback about the project. This is one further step in delivering the promised background materials related to the project. Stay tuned for a Wolfram Mathematica package and notebook, a Blog post and an Essay. I am deeply grateful to Anita L Verő for her technical help and encouragement in both the Python package and notebook. Previous related posts on LinkedIn: https://lnkd.in/dfqEYjht https://lnkd.in/d8pP9pXF https://lnkd.in/dpbeEDkx Recorded AI for QFT seminar talk on Queen Mary University of London uploaded to YouTube: https://lnkd.in/dCVvq2qT Slides for the talk: https://lnkd.in/d2aYcYU3 (The cover image shows Roman (or medieval) dice from York. Their imperfection symbolises that not all uncertain possibilities can be associated with easy-to-deduce probabilities.)

15

1 Comment

𝗧𝗶𝗺𝗲 𝗦𝗲𝗿𝗶𝗲𝘀 𝗙𝗼𝗿𝗲𝗰𝗮𝘀𝘁𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗧𝗶𝗺𝗲𝗚𝗣𝗧 📈 Nixtla has introduced the first foundation transformer model specifically designed for time series forecasting! TimeGPT was trained on over 100 billion data points, including time series from domains like finance, energy, web traffic, sales and more. As a foundation model, TimeGPT can generate accurate predictions for datasets not seen in training, making it a valuable tool. Furthermore, TimeGPT can be accessed via a Python library or the associated API created by Nixtla. Check the links for more information, and make sure to follow me for regular content! 𝗧𝗶𝗺𝗲𝗚𝗣𝗧 𝗤𝘂𝗶𝗰𝗸𝘀𝘁𝗮𝗿𝘁: https://lnkd.in/d6-3qJV6 𝗟𝗲𝗮𝗿𝗻 𝗠𝗟 𝗮𝗻𝗱 𝗧𝗦 𝗙𝗼𝗿𝗲𝗰𝗮𝘀𝘁𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗣𝘆𝗖𝗮𝗿𝗲𝘁📚: https://lnkd.in/dyByK4F #datascience #python #machinelearning #deeplearning #linkedin

4,233

145 Comments

Live Workshop: Build Algorithmic Trading Strategies with Python (that actually work). Here's what's coming: Inside the LIVE EVENT, we will demystify: 1. How to trade for yourself 2. Which trading strategies work (and which don't) 3. What it takes to manage risk (and prevent large losses) 4. Why backtests fail (and how to improve them) 5. How to trade responsibly so your wins stack up (and you can compete with the pros) What's your next step? Register here (1788 registered): https://lnkd.in/d4XgzYaw

19

1 Comment

ML projects remind me of Hofstadter's law, which states “It always takes longer than you expect, even when you take into account Hofstadter's Law.” We see so much variability in project length, for a few main reasons. - Data quality - the number one make or break element of any ML project. Garbage in = garbage out. Creating a robust data practice takes time, and that's required before starting any ML project. - Improving performance - ML projects typically go from 0% to ~70% efficiency quickly, and start to slow down in improvement rate after that. Some researchers spend years trying to improve an algorithm by 2%. Years! - Lack of experience - this ML project might be someone's first rodeo. That's fine, but there's a lot to learn that differs from an engineering sprint or a UX study. We usually "take over" the project from an in-experienced ML engineer or data engineer. ML engineering is not something your "internal team can try." It is like asking a baseball player to play basketball. My number one tip....plan for unpredictability. Be conservative in estimates, and account for delays in your timeline.

5

Apparently, someone (maybe KPMG) is running a supply chain due diligence on pydantic on behalf of NASA. The thread is interesting. I link in the comment, but I summarise in a couple of points: - "This is so dumb. Do they know what OSS is?" Dumbness is not an issue in regulations. There was literally a supply chain attack recently on Linux, and if successful, a state actor would have gained access to half the internet (which makes you wonder what else is happening out there). You can read more about this if you search "hacker news xz," which is a fascinating story; I might write about it. - "KPMG ripping of NASA and the taxpayers" Maybe the astronauts and the rocket scientists don't want to spend their time on this? Outsourcing these are pretty standard to those poor people whose entire job is just this. (Shed a tear for their souls) - "Can you do it for free for NASA?" Why would he? They have an enormous budget. Pydantic must have a good place for that money. - "Did not OSS maintainers suffer enough?" They can ignore the request, and NASA will have no other choice but to either fork pydantic or recreate it from scratch. They have no other choice. They are not doing this because they have nothing better to do. - "AI will replace these people, can't wait for it." That's just mean. - "This is so dumb." They plan to send people to the Moon, not exactly a "move fast and break things" mission. Look at what happened when this principle was ignored at Boeing. NASA wants to ensure that if there is a bug or issue with pydantic, someone will be there to fix it. They also want to ensure that no nefarious actors have a chance to take down anything serious or at least have a contained blast radius. This is a negotiation. How much the maintainer can charge for this? What is the alternative? Recreate pydantic from scratch? How many lines of code? How many engineers and how much time to recreate? How much time does it take to review a fork and maintain compatibility? This will provide a BATNA ("Best Alternative to Negotiated Agreement) that you can use as a baseline. On the other hand, you have your costs to actually fulfil the above agreement. Some quoted lowball numbers, which indicate that the entire deal can be done in 1-2 weeks. It won't, and you will regret accepting the deal. There is a huge crisis in OSS, especially with single-person maintainers. Huge companies rely on these softwares, but because there is no business behind them, the individual probably can't or doesn't want to deal with the commercial side. They started doing it in their spare time and now things go out of hand. [continued in comments]

89

40 Comments

That's awesome! Mistral's new features for building custom LLMs sound incredibly developer-friendly. The ability to fine-tune and create custom agents makes it a powerful tool for personalized AI solutions. Thanks for sharing the video link—can't wait to check it out! 🚀🔍 #CustomLLMs #AIInnovation #DeveloperTools #MachineLearning #MistralLLM

1

This AI-powered Jupyter implementation is a game-changer! The addition of Mistral's Codestral and GPT4-o for inline AI Copilot features truly enhances the coding experience by providing auto-complete, code generation, editing, error fixing, and sidebar chat. It feels more like a collaborative tool, making coding more efficient and enjoyable. Can't wait to see its impact on the AI and developer community! 💻🚀 #AI #Coding #MachineLearning #TechInnovation #Jupyter #DeveloperTools

1