David R.
@drbn.bsky.social
Creative tech enthusiast turning ideas into reality. Building apps, exploring new projects, and always learning.
This comparison between 𝗠𝗲𝗹𝗼𝗱𝗶𝗼𝘂𝘀 𝗪𝗮𝗿𝗯𝗹𝗲𝗿 and 𝗜𝗰𝘁𝗲𝗿𝗶𝗻𝗲 𝗪𝗮𝗿𝗯𝗹𝗲𝗿 was generated entirely using OpenAI’s new image creation tool.
While not perfect, it’s getting seriously good, the illustrations are more detailed, and it now manages real text. The progress is evident. This is moving fast.
While not perfect, it’s getting seriously good, the illustrations are more detailed, and it now manages real text. The progress is evident. This is moving fast.
March 29, 2025 at 1:05 PM
This comparison between 𝗠𝗲𝗹𝗼𝗱𝗶𝗼𝘂𝘀 𝗪𝗮𝗿𝗯𝗹𝗲𝗿 and 𝗜𝗰𝘁𝗲𝗿𝗶𝗻𝗲 𝗪𝗮𝗿𝗯𝗹𝗲𝗿 was generated entirely using OpenAI’s new image creation tool.
While not perfect, it’s getting seriously good, the illustrations are more detailed, and it now manages real text. The progress is evident. This is moving fast.
While not perfect, it’s getting seriously good, the illustrations are more detailed, and it now manages real text. The progress is evident. This is moving fast.
Received two books this Christmas: How to Make the World Add Up by Tim Harford and AI Needs You by Verity Harding.
Harford’s book focuses on understanding data in a world of misinformation, while Harding’s explores the ethical impact of SI.
Looking forward to applying their insights.
Harford’s book focuses on understanding data in a world of misinformation, while Harding’s explores the ethical impact of SI.
Looking forward to applying their insights.
January 6, 2025 at 5:16 PM
Received two books this Christmas: How to Make the World Add Up by Tim Harford and AI Needs You by Verity Harding.
Harford’s book focuses on understanding data in a world of misinformation, while Harding’s explores the ethical impact of SI.
Looking forward to applying their insights.
Harford’s book focuses on understanding data in a world of misinformation, while Harding’s explores the ethical impact of SI.
Looking forward to applying their insights.
The evolution of AI agents feels like a quiet yet profound tide sweeping the tech sector.
What Is an AI Agent? A Computer Scientist Explains the Next Wave of AI Tools
The AI agents big tech companies are now developing are significant advances over chatbots because they possess the ability to take actions on your behalf.
singularityhub.com
January 5, 2025 at 8:00 AM
The evolution of AI agents feels like a quiet yet profound tide sweeping the tech sector.
Nvidia's firm hold on AI chips may wane, showing the volatility and unpredictability of tech market trends.
Here's How Nvidia’s Vice-Like Grip on AI Chips Could Slip
A recent change in priorities at the biggest AI developers could shake up the industry.
singularityhub.com
January 4, 2025 at 8:00 AM
Nvidia's firm hold on AI chips may wane, showing the volatility and unpredictability of tech market trends.
I’ve been exploring AI orchestration frameworks for my next project, and honestly, it’s not easy to find one that feels right.
There’s a lot to consider: stability, flexibility, ease of integration and documentation.
Right now, I’m leaning toward Haystack.
haystack.deepset.ai
There’s a lot to consider: stability, flexibility, ease of integration and documentation.
Right now, I’m leaning toward Haystack.
haystack.deepset.ai
Haystack | Haystack
Haystack, the composable open-source AI framework
haystack.deepset.ai
December 31, 2024 at 11:24 AM
I’ve been exploring AI orchestration frameworks for my next project, and honestly, it’s not easy to find one that feels right.
There’s a lot to consider: stability, flexibility, ease of integration and documentation.
Right now, I’m leaning toward Haystack.
haystack.deepset.ai
There’s a lot to consider: stability, flexibility, ease of integration and documentation.
Right now, I’m leaning toward Haystack.
haystack.deepset.ai
The prospect of AI agents ruling 2025? Quite an ambitious bet, don't you think?
Is 2025 the Year AI Agents Take Over? Industry Bets Billions on AI's Killer App
Industry experts say AI will begin to live up to the hype next year as AI agents force their way into all aspects of our lives.
singularityhub.com
December 27, 2024 at 7:30 PM
The prospect of AI agents ruling 2025? Quite an ambitious bet, don't you think?
Agentic AI is becoming more popular, and GeAR offers a fresh take on Retrieval-Augmented Generation (RAG).
It combines graph expansion with agent-based methods.
With interest in Agentic AI growing, this is an interesting direction to watch.
arxiv.org/abs/2412.18431
It combines graph expansion with agent-based methods.
With interest in Agentic AI growing, this is an interesting direction to watch.
arxiv.org/abs/2412.18431
GeAR: Graph-enhanced Agent for Retrieval-augmented Generation
Retrieval-augmented generation systems rely on effective document retrieval capabilities. By design, conventional sparse or dense retrievers face challenges in multi-hop retrieval scenarios. In this p...
arxiv.org
December 27, 2024 at 11:42 AM
Agentic AI is becoming more popular, and GeAR offers a fresh take on Retrieval-Augmented Generation (RAG).
It combines graph expansion with agent-based methods.
With interest in Agentic AI growing, this is an interesting direction to watch.
arxiv.org/abs/2412.18431
It combines graph expansion with agent-based methods.
With interest in Agentic AI growing, this is an interesting direction to watch.
arxiv.org/abs/2412.18431
This AI development is quite impressive, isn't it? The potential for future innovation feels immense.
AI That Can Design Life's Machinery From Scratch Had a Big Year. Here's What Happens Next.
It seems imagination is the only limit to AI-based protein design. But there's room for even more progress in 2025.
singularityhub.com
December 23, 2024 at 6:30 PM
This AI development is quite impressive, isn't it? The potential for future innovation feels immense.
Reposted by David R.
We are getting into a Catalan Christmas tradition today.
My daughter “found” Caga Tió (literally, “sh!tting log”) in the mountains.
We are now keeping him warm and feeding him until Christmas Eve.
en.m.wikipedia.org/wiki/Ti%C3%B...
My daughter “found” Caga Tió (literally, “sh!tting log”) in the mountains.
We are now keeping him warm and feeding him until Christmas Eve.
en.m.wikipedia.org/wiki/Ti%C3%B...
December 21, 2024 at 9:49 AM
We are getting into a Catalan Christmas tradition today.
My daughter “found” Caga Tió (literally, “sh!tting log”) in the mountains.
We are now keeping him warm and feeding him until Christmas Eve.
en.m.wikipedia.org/wiki/Ti%C3%B...
My daughter “found” Caga Tió (literally, “sh!tting log”) in the mountains.
We are now keeping him warm and feeding him until Christmas Eve.
en.m.wikipedia.org/wiki/Ti%C3%B...
Well. We have OpenAI's o3 model. The AI landscape just got even more competitive. With Google unveiling Gemini 2 Flash and now OpenAI launching GPT-o3, it’s clear we’re in the middle of an AI arms race.
x.com/OpenAI/statu...
x.com/OpenAI/statu...
December 20, 2024 at 8:08 PM
Well. We have OpenAI's o3 model. The AI landscape just got even more competitive. With Google unveiling Gemini 2 Flash and now OpenAI launching GPT-o3, it’s clear we’re in the middle of an AI arms race.
x.com/OpenAI/statu...
x.com/OpenAI/statu...
Google just introduced Gemini 2, its new generative AI model designed to handle reasoning and context more effectively. Like OpenAI's o1 model, they're taking a smarter approach by refining how the model thinks and processes task.
aistudio.google.com/prompts/new_...
aistudio.google.com/prompts/new_...
December 20, 2024 at 10:56 AM
Google just introduced Gemini 2, its new generative AI model designed to handle reasoning and context more effectively. Like OpenAI's o1 model, they're taking a smarter approach by refining how the model thinks and processes task.
aistudio.google.com/prompts/new_...
aistudio.google.com/prompts/new_...
GitHub has introduced a free tier for Copilot in Visual Studio Code, offering developers up to 2,000 code completions and 50 chat messages per month. This plan provides access to AI models like Anthropic's Claude 3.5 Sonnet and OpenAI's GPT-4o.
github.blog/news-insight...
github.blog/news-insight...
Announcing 150M developers and a new free tier for GitHub Copilot in VS Code
Come and join 150M developers on GitHub that can now code with Copilot for free in VS Code.
github.blog
December 18, 2024 at 11:27 PM
GitHub has introduced a free tier for Copilot in Visual Studio Code, offering developers up to 2,000 code completions and 50 chat messages per month. This plan provides access to AI models like Anthropic's Claude 3.5 Sonnet and OpenAI's GPT-4o.
github.blog/news-insight...
github.blog/news-insight...
With the NVIDIA Jetson Orin Nano Developer Kit, you could build a robot that navigates on its own, a smart home system that understands voice commands, or a device that detects objects in real-time. It’s a powerful tool for your AI ideas.
www.nvidia.com/en-us/autono...
www.nvidia.com/en-us/autono...
Jetson Orin Nano Super Developer Kit
The most affordable generative AI supercomputer.
www.nvidia.com
December 18, 2024 at 8:40 AM
With the NVIDIA Jetson Orin Nano Developer Kit, you could build a robot that navigates on its own, a smart home system that understands voice commands, or a device that detects objects in real-time. It’s a powerful tool for your AI ideas.
www.nvidia.com/en-us/autono...
www.nvidia.com/en-us/autono...
STAR (Synthesis of Tailored Architectures) is an innovative method for automating the design of neural network architectures using evolutionary algorithms.
Designing neural networks is manual and slow. This method can automate it, speeding up development.
www.liquid.ai/research/aut...
Designing neural networks is manual and slow. This method can automate it, speeding up development.
www.liquid.ai/research/aut...
Automated Architecture Synthesis via Targeted Evolution
Today, we report advances in automated neural network architecture design and customization. We developed algorithms for the synthesis of tailored architectures (STAR), based on evolutionary algorithm...
www.liquid.ai
December 16, 2024 at 11:12 PM
STAR (Synthesis of Tailored Architectures) is an innovative method for automating the design of neural network architectures using evolutionary algorithms.
Designing neural networks is manual and slow. This method can automate it, speeding up development.
www.liquid.ai/research/aut...
Designing neural networks is manual and slow. This method can automate it, speeding up development.
www.liquid.ai/research/aut...
AnnotateAI: A tool for annotating research papers
AnnotateAI extracts key concepts, highlights important sections, and creates concise annotations. It works well with papers from arXiv, PubMed, and other scientific sources, making it easier to review and focus on key ideas. Worth a try.
AnnotateAI extracts key concepts, highlights important sections, and creates concise annotations. It works well with papers from arXiv, PubMed, and other scientific sources, making it easier to review and focus on key ideas. Worth a try.
GitHub - neuml/annotateai: 📝 Automatically annotate papers using LLMs
📝 Automatically annotate papers using LLMs . Contribute to neuml/annotateai development by creating an account on GitHub.
github.com
December 15, 2024 at 7:49 AM
AnnotateAI: A tool for annotating research papers
AnnotateAI extracts key concepts, highlights important sections, and creates concise annotations. It works well with papers from arXiv, PubMed, and other scientific sources, making it easier to review and focus on key ideas. Worth a try.
AnnotateAI extracts key concepts, highlights important sections, and creates concise annotations. It works well with papers from arXiv, PubMed, and other scientific sources, making it easier to review and focus on key ideas. Worth a try.
"We've achieved peak data and there'll be no more."
"Next-generation models are going to be agentic in real ways."
"The more a system reasons, the more unpredictable it becomes."
These insights from OpenAi cofounder ilya Sutskever shed light on the challenges as AI becomes more autonomous.
"Next-generation models are going to be agentic in real ways."
"The more a system reasons, the more unpredictable it becomes."
These insights from OpenAi cofounder ilya Sutskever shed light on the challenges as AI becomes more autonomous.
OpenAI cofounder Ilya Sutskever says the way AI is built is about to change
“We’ve achieved peak data and there’ll be no more.”
www.theverge.com
December 14, 2024 at 6:50 PM
"We've achieved peak data and there'll be no more."
"Next-generation models are going to be agentic in real ways."
"The more a system reasons, the more unpredictable it becomes."
These insights from OpenAi cofounder ilya Sutskever shed light on the challenges as AI becomes more autonomous.
"Next-generation models are going to be agentic in real ways."
"The more a system reasons, the more unpredictable it becomes."
These insights from OpenAi cofounder ilya Sutskever shed light on the challenges as AI becomes more autonomous.
This Forbes article talks about two AI-powered search engines, Consensus and OpenEvidence, that are changing how doctors access medical knowledge.
Both tools look like a great help for healthcare professionals to stay up-to-date and make more informed choices when diagnosing or treating patients.
Both tools look like a great help for healthcare professionals to stay up-to-date and make more informed choices when diagnosing or treating patients.
How AI Is Changing The Way Doctors Access Medical Knowledge
Throughout history, doctors have shifted how they access medical knowledge from their minds to paper to digital tools. AI is the next step in this evolution.
www.forbes.com
December 13, 2024 at 8:04 PM
This Forbes article talks about two AI-powered search engines, Consensus and OpenEvidence, that are changing how doctors access medical knowledge.
Both tools look like a great help for healthcare professionals to stay up-to-date and make more informed choices when diagnosing or treating patients.
Both tools look like a great help for healthcare professionals to stay up-to-date and make more informed choices when diagnosing or treating patients.
Quantum computing is inching closer to reality with Google's latest breakthrough, making practical machines seem less elusive.
Google's Latest Quantum Computing Breakthrough Shows Practical Machines Are Within Reach
Google’s new Willow quantum processor has crossed a critical threshold in quantum computing.
singularityhub.com
December 12, 2024 at 10:58 PM
Quantum computing is inching closer to reality with Google's latest breakthrough, making practical machines seem less elusive.
See the latest leaderboard of AI providers at artificialanalysis.ai. Who’s leading the race in innovation and performance?
AI Model & API Providers Analysis | Artificial Analysis
Comparison and analysis of AI models and API hosting providers. Independent benchmarks across key performance metrics including quality, price, output speed & latency.
artificialanalysis.ai
December 12, 2024 at 8:31 PM
See the latest leaderboard of AI providers at artificialanalysis.ai. Who’s leading the race in innovation and performance?
This is interesting: a NVIDIA RAG course to learn about retrieval-augmented generation and its practical applications with LLMs. It’s a solid resource for understanding how AI integrates retrieval with generation. #AI #NVIDIA #RAG
learn.nvidia.com/courses/cour...
learn.nvidia.com/courses/cour...
Course Detail | NVIDIA
learn.nvidia.com
December 11, 2024 at 7:52 PM
This is interesting: a NVIDIA RAG course to learn about retrieval-augmented generation and its practical applications with LLMs. It’s a solid resource for understanding how AI integrates retrieval with generation. #AI #NVIDIA #RAG
learn.nvidia.com/courses/cour...
learn.nvidia.com/courses/cour...
Here are some practical lessons for managing and refining prompts in production LLM systems. Very useful.
www.zenml.io/blog/prompt-...
www.zenml.io/blog/prompt-...
Prompt Engineering & Management in Production: Practical Lessons from the LLMOps Database - ZenML Blog
Practical lessons on prompt engineering in production settings, drawn from real LLMOps case studies. It covers key aspects like designing structured prompts (demonstrated by Canva's incident review sy...
www.zenml.io
December 11, 2024 at 1:41 PM
Here are some practical lessons for managing and refining prompts in production LLM systems. Very useful.
www.zenml.io/blog/prompt-...
www.zenml.io/blog/prompt-...
An insightful paper advocating for open-source LLMs in medicine.
pmc.ncbi.nlm.nih.gov/articles/PMC...
pmc.ncbi.nlm.nih.gov/articles/PMC...
The path forward for large language models in medicine is open
Large language models (LLMs) are increasingly applied in medical documentation and have been proposed for clinical decision support. We argue that the future for LLMs in medicine must be based on tran...
pmc.ncbi.nlm.nih.gov
December 11, 2024 at 7:12 AM
An insightful paper advocating for open-source LLMs in medicine.
pmc.ncbi.nlm.nih.gov/articles/PMC...
pmc.ncbi.nlm.nih.gov/articles/PMC...
A very necessary systematic review of the use of large language models in healthcare.
jamanetwork.com/journals/jam...
jamanetwork.com/journals/jam...
Testing and Evaluation of Health Care Applications of Large Language Models
This systematic review characterizes the current performance of large language models in evaluating clinical health care settings, including uniformity, thoroughness, and robustness and proposes a fra...
jamanetwork.com
December 10, 2024 at 9:38 PM
A very necessary systematic review of the use of large language models in healthcare.
jamanetwork.com/journals/jam...
jamanetwork.com/journals/jam...
Yann LeCun's I-JEPA (Joint Embedding Predictive Architecture) represents a shift in how AI could evolve, with a focus on efficiency and reasoning rather than brute force scaling.
ai.meta.com/blog/yann-le...
ai.meta.com/blog/yann-le...
I-JEPA: The first AI model based on Yann LeCun’s vision for more human-like AI
I-JEPA learns by creating an internal model of the outside world, which compares abstract representations of images (rather than comparing the pixels themselves).
ai.meta.com
December 9, 2024 at 9:35 PM
Yann LeCun's I-JEPA (Joint Embedding Predictive Architecture) represents a shift in how AI could evolve, with a focus on efficiency and reasoning rather than brute force scaling.
ai.meta.com/blog/yann-le...
ai.meta.com/blog/yann-le...
An insightful post. Perhaps scaling for its own sake distracts us from addressing the foundational inefficiencies and limitations of LLMs.
The End of Scaling Laws or a New Beginning? - Deep Reflections from an Open-Source Practitioner
Doone Song Recent discussions surrounding large model scaling and low-precision training have intensified, with the latest paper, "Scaling Laws for Precision," igniting widespread debate within the AI...
www.linkedin.com
December 8, 2024 at 12:38 PM
An insightful post. Perhaps scaling for its own sake distracts us from addressing the foundational inefficiencies and limitations of LLMs.