AI/ML Digest | Issue 35
Roosh Circle
We are a tech community of founders, engineers, investors, and entrepreneurs across Europe
Hello and welcome to the latest edition of the Innovations in AI/ML Digest, brought to you by Roosh Circle . Join us as we explore the recent updates in the dynamic world of artificial intelligence and machine learning.
Before we explore the latest in AI advancements, we encourage you to join us at our upcoming events:
See you there! Now, back to the digest ↓
1/7 Langflow has revolutionized the creation of RAG applications with its no-code, open-source tool.
Developers appreciate the tool’s simplicity and efficiency, which streamlines workflows without any hassle.
A step-by-step guide: https://cutt.ly/9eq9x1FX
2/7 Task delegation just got a major upgrade
AI planning models can now efficiently decompose large tasks into smaller, more manageable components. This method improves organizational efficiency and demonstrates the strategic capability of scaled-down AI models.
Read more: https://cutt.ly/deq9vm6V
3/7 New Cohere Toolkit Accelerates GenAI Application Development
Cohere Toolkit is an open-source repository of production-ready applications deployable across cloud platforms. It was made available for developers to build AI applications faster.
GitHub: https://cutt.ly/2eq9bIsE
4/7 Struggling with JSON data structures? Here's a tip: just copy, paste, and visualize!
It’s a straightforward way to understand your data's architecture without any complications. Transform your data into interactive graphs: https://cutt.ly/4eq9b5PI
5/7 Introducing AI-powered user interviews
AI is redefining product research through user interviews that begin with surveys and evolve into comprehensive conversations, offering a deeper understanding of consumer needs with unprecedented insight.
More info: https://cutt.ly/Aeq9nI4D
6/7 The latest YC Batch W24 demo day reveals the latest AI trends
The latest YC Batch W24 demo day showcased 247 companies, each providing a glimpse into the future of AI.
Check them all: https://cutt.ly/Neq9mZdW
7/7 Adaptive-RAG: Learning to Adapt Retrieval-Augmented Large Language Models through Question Complexity
In their paper, researchers from Cornell University introduce an innovative adaptive QA framework capable of dynamically choosing the optimal strategy for retrieval-augmented large language models (LLMs). This framework assesses the complexity of a query and selects a response strategy ranging from the simplest to the most sophisticated, ensuring the most effective handling of various inquiries.
GitHub: https://cutt.ly/Deq9WXPO
Access Paper: https://cutt.ly/Req9W5Ei
1/8 The evolution of AI landscape
These maps highlight the latest developments in AI technologies and applications, offering valuable insights for navigating this dynamic field.
Check: https://cutt.ly/3eq9OVpO
2/8 Financial analysts, here's a new tool to transform your queries into actionable data
It transforms complex financial questions into actionable data seamlessly. Ask something like, "How did revenue change between Q4 2023 and the year before?" and receive your answer right away.
Explore more: https://cutt.ly/Seq9PEII
3/8 Entropy is making waves in AI training
By measuring diversity in training processes, entropy is becoming an invaluable tool for AI engineers looking to fine-tune their models.
GitHub: https://cutt.ly/Geq9Aqia
Also, here you can read more about entropy in AI.
4/8 Implementing FrugalGPT: reducing LLM costs & improving performance
FrugalGPT is a framework proposed a 2023 paper "FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performance". The paper outlines strategies for more cost-effective and performant usage of large language model (LLM) APIs.
Read more: https://cutt.ly/Veq9A7Vp
5/8 Mixtral 8x22B is setting new standards for AI performance and efficiency
It sets a new standard for performance and efficiency within the AI community. It is a sparse Mixture-of-Experts (SMoE) model that uses only 39B active parameters out of 141B, offering unparalleled cost efficiency for its size.
Dive deeper: https://cutt.ly/Seq9DEIM
6/8 Meta Llama 3 has arrived, claiming the title of the most capable openly available LLM to date
领英推荐
This release features pre-trained and instruction-fine-tuned language models with 8B and 70B parameters that can support a broad range of use cases.
More details: https://cutt.ly/meq9FwD0
7/8 Introducing the Batch API — your key to cost savings and increased rate limits for tasks like summarization and translation.
Create large batches of API requests for asynchronous processing. The Batch API returns completions within 24 hours for a 50% discount.
Code: https://cutt.ly/heq9FLlk
8/8 Meet Payman, the new AI Agent tool
Payman gives your AI Agents the ability to have access to capital so that they can pay humans for their expertise in completing specialized tasks they need done.
Site: https://cutt.ly/weq9GmGt
1/6 torchtitan
The PyTorch team has just introduced torchtitan, a new library for training AI models. Now, you can even train Llama-3 from scratch! This tool is still in pre-release, but it's available on GitHub for those eager to start exploring.
GitHub: https://cutt.ly/Eeq9Jtbn
2/6 More financial tools
Financial Datasets is an open-source Python library that lets you create question & answer financial datasets using Large Language Models (LLMs). With this library, you can easily generate realistic financial datasets from a 10-K, 10-Q, PDF, and other financial texts.
GitHub: https://cutt.ly/veq9JH51
Example of code: https://cutt.ly/Feq9KkCn
3/6 Dive into the world of Transformers with a new educational blog series
This is a multi-part series on creating a Transformer from scratch in PyTorch. By the end of the series, you will be familiar with the architecture of a standard Transformer and common variants you will find across recent models such as GPT, PaLM, LLaMA, MPT, and Falcon. You will also be able to understand how Transformers are being used in domains other than language.
Part 1: The Attention Mechanism: https://cutt.ly/ceq9Xiak
Part 2: The Rest of the Transformer: https://cutt.ly/feq9XlWb
4/6 Introducing FireCrawl by the Mendable team
"Firecrawl is an API service that takes a URL, crawls it, and converts it into clean markdown. We crawl all accessible subpages and give you clean markdown for each. No sitemap required".
GitHub: https://cutt.ly/Peq91I5G
5/6 The AI Index Report 2024 is out
The AI Index report tracks, collates, distills, and visualizes data related to artificial intelligence. Its mission is to provide unbiased, rigorously vetted, broadly sourced data in order for policymakers, researchers, executives, journalists, and the general public to develop a more thorough and nuanced understanding of the complex field of AI.
Full report: https://cutt.ly/eeq90o1j
6/6 Generate beautiful docs from your transcripts and unstructured information with a single command
Lumentis is a simple way to generate comprehensive, easy-to-skim docs from your meeting transcripts and large documents.
GitHub: https://cutt.ly/Seq99fzY
1/4 A new mind map is here to guide startup founders to SEO mastery — and it's free!
Perfect for those ready to boost their online presence without getting lost in the search engine woods.
Site: https://cutt.ly/Aeq98snC
2/4 From lightbulb moment to cash flow!
Discover the 13 steps that can transform your idea into a revenue-generating reality. Thread: https://cutt.ly/aeq94ajs
3/4 The full deck "Future of Autonomous Agents" from the AI Rabbithole event
Read the deck here: https://cutt.ly/8eq94ZFU
4/4 The race is on for A16Z's SPEEDRUN!
Over 40 companies are headed towards a $750k funding prize, but the challenge is real for AI startups. Who will secure their spot at the finish line?
More: https://cutt.ly/Meq95Exg
Thanks for subscribing and reading! Stay tuned for more updates and new AI/ML Digest editions