Piotr Skalski

Recently, Hugging Face - the largest AI community and hub for open-source models - released their mobile app. HuggingChat - an open-source alternative to ChatGPT. It is currently available only on iOS and the web, but I believe that Android version is coming soon. This is a bold move for HF, as it signifies a shift from the developer community to the consumer market, where they will compete with such top-of-mind chars like ChatGPT and Perplexity. It's interesting that HuggingChat wasn't widely discussed in the news. Well, let's see how HF presents itself on the broader consumer market. #AI #Huggingface #ChatGPT #Perplexity #OpenAI #LLM #opensource

7

5 komentarzy

Pay attention to these two guys at the top. They are fresh releases from MistralAI. The first one is Mamba-based (that's Selective State Space Model), which is quite unconventional and is indeed MistralAI's very first model based on that architecture. I haven't tested it yet, but I anticipate that it excels at coding, given its benchmarks. Even more intriguing would be to examine its memory consumption and performance with long inputs, as these are the advantages of the SSM architecture. The second is just another 7B model from the Math domain. Little Santa's Helper for the homework on math ;) By the way, I've just rolled out a huge update for LLM Explorer, which now employs a dynamic scoring and ranking for LLMs. I've named it the "LLM Explorer Rank." As a result, all 34,000 models are adequately ranked, regardless of their presence on the leaderboards. Check it out at https://llm.extractum.io.

9

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

Except they aren't the same. Despite me trying to convince myself they are. I just finished reading Aurimas' newsletter where he talked about memory for agents. Side note, I'm organizing the AI Agents in production virtual conference on November 13th. We will go in-depth on AI agents. Register here https://mlops-l.ink/0srs Some key non-AI generated thoughts and takeaways from my side: 1. Simplified definition and diagram of agents - basic recipe is knowledge, LLM, short-term and long-term memory and tools. (check out the diagram if you are a visual learner. 2. Goldfish agents are worthless - I think lots of early agent frameworks ran into problems because they couldn't properly manage short-term memory. ie continuously updating the context in the system prompt is a recipe for failure. 3. Long-term memory to save the day - lots of focus is being shifted to long-term memory to remedy these short-term memory problems. 4. Hive mind memory - interesting thought experiment is to consider that multiple instances of the same agent interacting with different humans could tap into this memory independently creating a sort of hive mind. 5. Could episodic memory give us collaborative mode - how can I make my agents available for other folks to use? how can I give you an agent workflow that you can tweak to your needs? 6. Semantic memory can get you far - mainly for grounding and reinforcing something that you hold important. (it also resembles a rag system a bit more) 7. Procedural memory for reproducibility - this is where the agentic frameworks value props shine. anything that will help codify the process so that you can track changes and better reproduce successful results should be tracked. Diving into the AI agents world makes me so stoked for the Agents in production virtual conference. https://mlops-l.ink/0srs

172

25 komentarzy

Insightful post by Heather Couture, PhD on dataset size and quality of the dataset. Thanks for sharing Heather 🙂 #RemoteSensing #EarthObservation #MachineLearning #DeepLearning #ComputerVision

2

Many people need to realize that the output of Generative AI models can now be fully controlled—2023 was like ten years ago! With GPT-4-turbo, you can retrieve and augment user input with up to 128,000 tokens! The output will be transparent and reliable for everybody. I'll be sure to break this down for you. 1. The old days (2023) are over in generative AI. The attached notebook open-source, free notebook, shows you how to do it. You can find it in the Chapter07 free access directory of the GitHub repository of Transformers for Natural Language Processing and Computer Vision, 3rd Edition on this page: on my resource page: https://lnkd.in/egKueDhE 2. The code proves the point: you can augment a user input with retrieved documents and control the generative AI Model. Of course, you need to engineer the retrieval properly. 3. Get ready to embark on an exciting journey into the new era of Generative AI! Run the code and explore the endless possibilities that these advancements offer. If you have any questions, you can ask them right away and here in the comment section. #ai #genai #rag

170

27 komentarzy

Is Prime Number - Faster Wojciech Christopher Peter Fiałkiewicz ABSTRACT: There is a Basic Algorithm for determining if a given number is a Prime Number. An enhanced version of this algorithm is presented in this document. CONTENT:     bool IsPrimeNumber(ulong nr)     {       ulong half = nr / 2;       for (ulong i = 2; i <= half; i++)       {         if (nr % i == 0)         {           return false;         }       }       return true;     }     bool IsPrimeNumberFaster(ulong nr)   {       ulong max = nr / 2;      bool can_be = false;      if (nr == 0 || nr == 1 || nr == 2 || nr == 3)       {         return true;   }     for (ulong i = 2; i <= max; i++)      {        if (can_be && IsPrimeNumber(i))       {         max = nr / i;        can_be = false;     }    else    {         can_be = true;       }     if (nr % i == 0)    {       return false;        }      }   return true; }

1

💥💥💥 Discrete Latent Structure in Neural Networks Vlad Niculae, Caio F. Corro, Nikita Nangia, Tsvetomila Mihaylova, André F. T. Martins Abstract Many types of data from fields including natural language processing, computer vision, and bioinformatics, are well represented by discrete, compositional structures such as trees, sequences, or matchings. Latent structure models are a powerful tool for learning to extract such representations, offering a way to incorporate structural bias, discover insight about the data, and interpret decisions. However, effective training is challenging, as neural networks are typically designed for continuous computation. This text explores three broad strategies for learning with discrete latent structure: continuous relaxation, surrogate gradients, and probabilistic estimation. Our presentation relies on consistent notations for a wide range of models. As such, we reveal many new connections between latent structure learning strategies, showing how most consist of the same small set of fundamental building blocks, but use them differently, leading to substantially different applicability and properties. 👉 https://lnkd.in/g3uNut6S #machinelearning

4

🎁🎁A guide for modern Computer Vision🎁🎁 👉In the last 18 months we received 1,100+ applications for research roles in ARGO Vision. We are really grateful for so much love💙 Unfortunately the majority part of the applicants doesn't deeply know a few milestones in neural CV necessary to deal with our interviews. This is a short collection of mostly-free resources could help you to join us or other cool companies in #AI world (or just spending a bit of good time in the summer). The full list upon request. 𝐁𝐨𝐨𝐤𝐬 (recommended): ✅DL with Python (chapters 1-5) https://t.ly/VjaVx ✅Python Object Programming https://t.ly/pTQRm 𝐁𝐨𝐨𝐤𝐬 (good to read): ✅CV - Algorithm & Apps: https://szeliski.org/Book ✅Deep Learning Book: https://lnkd.in/dmXGHun2 ✅UDL https://lnkd.in/dbwx4FWS 𝐎𝐧𝐥𝐢𝐧𝐞 V𝐢𝐝𝐞𝐨 𝐂𝐨𝐮𝐫𝐬𝐞𝐬 (recommended): ✅Berkeley - Modern CV (2023) https://t.ly/AU7S3 ✅Stanford - ML (2022) https://t.ly/LSYqp ✅MIT - Intro to DL (2023) https://t.ly/7eDx6 𝐋𝐢𝐛𝐫𝐚𝐫𝐢𝐞𝐬: ✅PyTorch https://lnkd.in/dTvJbjAx ✅PyTorchLighting https://lnkd.in/dAruPA6T ✅Albumentations https://albumentations.ai/ 𝐏𝐚𝐩𝐞𝐫𝐬 (Image Classification): ✅ResNet https://lnkd.in/dk9aQHjd ✅EfficientNet https://lnkd.in/dTsT44ae ✅ViT https://lnkd.in/dB5yKdaW 𝐏𝐚𝐩𝐞𝐫𝐬 (Semantic Segmentation): ✅UNet https://lnkd.in/dnpKVa6T ✅DeepLabV3+ https://lnkd.in/dVvqkmPk ✅SAM https://lnkd.in/dkKSbgAF ✅[EDIT] SAM 2 https://t.ly/oovJZ 𝐏𝐚𝐩𝐞𝐫𝐬 (Object Detection): ✅YOLOv1: https://lnkd.in/dQ9rs53B ✅YOLOv2: https://lnkd.in/dNRqXG47 ✅YOLOX: https://lnkd.in/d9ZtsF7g 𝐏𝐚𝐩𝐞𝐫𝐬 (Misc): ✅Adam: https://lnkd.in/d8PrjTid ✅Attention is all you need https://lnkd.in/d3QXm36m ✅Patches are all you need: https://lnkd.in/duaUU-d5 👉Interested in receiving the full list of our recommended resources? Comment the post (maybe with other milestones/resources in CV) and I will send it to you in DM🔥 #AI #artificialintelligence #deeplearning #computervision #machinelearning #ml #AIwithPapers #metaverse

599

210 komentarzy

Cerebras and Neural Magic have reached a significant milestone in the realm of large language models (LLMs). Through innovative techniques and purpose-built hardware, we achieved up to 70% parameter reduction in LLMs without sacrificing accuracy. This breakthrough opens doors for more efficient training and deployment, democratizing access to LLMs across various sectors. #AI #MachineLearning #Innovation

9

1 komentarz

Learn how to implement an optimal AI desktop app with Electron and NodeJS 😊 The app and blog by my colleague Mikołaj Roszczyk. #openvino #js #nodejs #electron #edgeai https://lnkd.in/dzhWwj8g

21

Papers are published in open-source repos like Arxiv, and why not the presentations? From SSI Club, we are super excited to organize an AI Paper-fest between 13th Nov and 15th Nov, inviting researchers globally to present the papers. These are the papers that are submitted to conferences like EMNLP etc. Key themes of the AI Paper-fest are - LLM Explainability, Benchmarking, Model compression, and Model fine-tuning. We already have 7 confirmed presentations by researchers from the US, China, South Korea, and India, and more will be confirmed soon. If you are an academician or a researcher working on LLMs, enterprise, or VC investing in AI, this is a free-for-all event. https://papers.ssiclub.ai/

15

CerebrasCoder - Llama 3 Powered App Generator (Free) CerebrasCoder, is an LLM-powered platform that can generate apps from prompts. CerebrasCoder is 100% open-source and free to use. CerebrasCoder is powered by Llama 3.3, one of the most powerful open-source LLMs. **CerebrasCoder details in the comments. #coding #aicoder #llms #llama3

263

22 komentarze

If you have been working with any #Python framework e.g. FastAPI - you probably would have worked with Pydantic - They have just released 𝗣𝘆𝗱𝗮𝗻𝘁𝗶𝗰 𝗔𝗜 - an agent framework that will help use Pydantic with any LLM! This framework, developed by the team behind Pydantic, offers a versatile and type-safe solution for building applications with large language models (LLMs). It supports a variety of models including OpenAI, Gemini, Groq, and soon Anthropic, with an easy interface for adding more. By using standard Python for control flow and agent composition, it allows developers to apply familiar best practices. Responses from these models are validated using Pydantic, ensuring data integrity even in real-time, streamed scenarios. Additionally, its type-safe dependency injection system enhances modularity and ease of testing, while Logfire integration provides robust monitoring and debugging tools. Overall, this framework streamlines AI application development, making it more efficient and reliable. More here -> https://ai.pydantic.dev/ __________________________________ ♻️ Repost if you find this useful! 🔔 Follow me, Naqqash Abbassi for more on Generative AI and my journey as a Founder and CTO in AI product development.

9

Scalable Optimization in the Modular Norm Tim Large, Yang Liu, Minyoung Huh, Hyojin Bahng, Phillip Isola, Jeremy Bernstein Abstract To improve performance in contemporary deep learning, one is interested in scaling up the neural network in terms of both the number and the size of the layers. When ramping up the width of a single layer, graceful scaling of training has been linked to the need to normalize the weights and their updates in the "natural norm" particular to that layer. In this paper, we significantly generalize this idea by defining the modular norm, which is the natural norm on the full weight space of any neural network architecture. The modular norm is defined recursively in tandem with the network architecture itself. We show that the modular norm has several promising applications. On the practical side, the modular norm can be used to normalize the updates of any base optimizer so that the learning rate becomes transferable across width and depth. This means that the user does not need to compute optimizer-specific scale factors in order to scale training. On the theoretical side, we show that for any neural network built from "well-behaved" atomic modules, the gradient of the network is Lipschitz-continuous in the modular norm, with the Lipschitz constant admitting a simple recursive formula. This characterization opens the door to porting standard ideas in optimization theory over to deep learning. We have created a Python package called Modula that automatically normalizes weight updates in the modular norm of the architecture. The package is available via "pip install modula" with source code at https://lnkd.in/dnC_nEcH 👉 https://lnkd.in/d2NNvFqP #machinelearning

2