Theo FunnyStrip @thecodingtheo

In retrospect it is surprising that OpenAI released o1-preview. As soon as they showed off reasoning, everyone copied it immediately. And if they had held off releasing a reasoning/planning model until o3 (& called that GPT-5) it would have been a startling leap in AI abilities.

Noam Brown @polynoamial

12 months ago

In retrospect it is surprising that OpenAI released o1-preview. As soon as they showed off reasoning, everyone copied it immediately. And if they had held off releasing a reasoning/planning model until o3 (& called that GPT-5) it would have been a startling leap in AI abilities.

43 221 2K 1.3M 500

Download Image

I have been arguing for a long time now that RL training will be key part of building product I cannot emphasize how important it is for the broader startup ecosystem to open up the artificial intelligence supply chain

Guillermo Rauch @rauchg

16 hours ago

I have been arguing for a long time now that RL training will be key part of building product I cannot emphasize how important it is for the broader startup ecosystem to open up the artificial intelligence supply chain

78 93 1K 369K 1K

Download Video

ModelOpt, @nvidia TensorRT Model Optimizer – open-source toolkit to accelerate models • Supports end-to-end optimizations: quantization, pruning, distillation, speculative decoding, sparsity • Works with Hugging Face, PyTorch, ONNX models • Integrates with NeMo, Megatron-LM,…

OpenAI’s latest internal forecast shows total spending soaring to $115 billion through 2029, roughly $80 billion more than it previously anticipated. The spike comes from: 🔹A plan to design in-house server chips and build proprietary data-center capacity to cut cloud-rental…

keyboard_arrow_left Previous keyboard_arrow_right Next

Google Employee, Logan Kilpatrick: 99% plus of all software that will be built in the future is going to be AI-powered in some capacity So we're launching tools to guide the ideation phase, even if imperfect, to help non-experts explore and build

📚🤖 AI Review Article Writer A technical series on building a system that automatically generates academic review articles using LangGraph's multi-agent architecture, handling complex document processing while maintaining rigorous academic standards. Learn more →…

INTERVIEW: Reid Hoffman (@reidhoffman) on AI, kids, and the future of learning. “Forbidding kids from using AI is a mistake. AI literacy will matter more than ever. ChatGPT already has study modes that guide students to answers instead of giving them away — turning it into a…

This is huge Continuing our foundational work to enable anyone to train state of the art AI model, we’re thrilled to release « FinePDFs » 3T tokens of textual data that until now was locked away in PDFs, arguably some of the highest quality publicly available data out there.…

Hynek Kydlíček @HKydlicek

a day ago

This is huge Continuing our foundational work to enable anyone to train state of the art AI model, we’re thrilled to release « FinePDFs » 3T tokens of textual data that until now was locked away in PDFs, arguably some of the highest quality publicly available data out there.…

24 110 681 173K 392

Download Image

LLMs do many things, to different levels of quality, the “jagged frontier” of ability that my coauthors and I discussed in 2023. One weak part of multimodal LLMs has been seeing fine visual details. So this is an interesting benchmark to watch to follow progress in this area.

Alek Safar @alek_safar

2 days ago

LLMs do many things, to different levels of quality, the “jagged frontier” of ability that my coauthors and I discussed in 2023. One weak part of multimodal LLMs has been seeing fine visual details. So this is an interesting benchmark to watch to follow progress in this area.

85 137 1K 398K 470

Download Image

Anthropic CEO, Dario Amodei: AI will soon make original scientific discoveries and, with a small group of humans, build new companies If AI can cure disease, that's the clearest way it makes life better for billions

🗺️🤖 Location-Aware AI Agents DeepAgents and Camino AI launch intelligent agents that combine location data with web search capabilities, powered by LangChain's framework for enhanced spatial research and reasoning. Learn more here 👉 blog.getcamino.ai/deepagents-loc…

It is often said that AI would turn us all into intelligent beings because everyone would have a phd in their pocket. That's not true. The most difficult thing is prompting. And especially asking the right questions. It takes knowledge of the subject area to design the question…

OpenAI published an important paper on: Hallucination why do LLMs hallucinate? "because we train them on tons of facts and grade them only right/wrong — with no credit for 'I don't know', that pushes them to guess" that's huge. this might be the missing piece for making LLMs…

If AI development continues over the next 1-3 years (which Amodei expects), we will see unimaginable discoveries. Everything will change.

Jon Hernandez @JonhernandezIA

2 days ago

If AI development continues over the next 1-3 years (which Amodei expects), we will see unimaginable discoveries. Everything will change. https://t.co/eMboXYWvEW

48 61 417 258K 132

Download Video

We are starting to see some nuanced discussions of what it means to work with advanced AI In this case, GPT-5 Pro was able to do novel math, but only when guided by a math professor (though the paper also noted the speed of advance since GPT-4). The reflection is worth reading.

keyboard_arrow_left Previous keyboard_arrow_right Next

Universal Deep Research NVIDIA recently published another banger tech report! The idea is simple: allow users to build their own custom, model-agnostic deep research agents with little effort. Here is what you need to know:

I wrote about how shockingly effective GPT-5 Thinking is at using its Bing-backed search tool, illustrated with 8 examples from the last 48 hours I've started calling it my Research Goblin

A new way for LLM agents to adapt ➡️ Memento uses memory + reinforcement learning (RL) to perform case-based continual learning for LLM agents. It has a Case Bank – a growing memory of past experiences, that updates during use, to help agents learn on the fly. This setup is…

ARC Prize will present at @OpenAI DevDay this October We'll be sharing ARC-AGI-3 progress, including first results on human performance and how interactive evaluations open a new axis for measuring intelligence

OpenAI @OpenAI

2 months ago

ARC Prize will present at @OpenAI DevDay this October We'll be sharing ARC-AGI-3 progress, including first results on human performance and how interactive evaluations open a new axis for measuring intelligence https://t.co/0DFGUpO3CV

333 576 4K 1.6M 760

Qwen3-Max-Preview (Instruct) biggest model yet from Qwen, with over 1 trillion parameters is now available in anycoder Benchmarks show it beats previous best, Qwen3-235B-A22B-2507 Voxel Pagoda garden, 1 shot