Transformer-Based Large Language Models Are Not General Learners: A Universal Circuit Perspective (openreview.net)
cross-posted from: slrpnk.net/post/5501378...
Understanding GPU Memory 2: Finding and Removing Reference Cycles (pytorch.org)
I hired a pirate to take orders for my entertainment business - Circus Scientist (www.circusscientist.com)
Ahoy there, matey! Welcome aboard Big Top Entertainment, the finest entertainment company on the seven seas!
The Paradox of AI Consciousness: Navigating the Boundaries of Machine Learning (www.ultra-unlimited.com)
An Analysis of DeepMind's 'Language Modeling Is Compression' Paper (codeconfessions.substack.com)
The Physical Process That Powers a New Type of Generative AI (www.quantamagazine.org)
Pretty cool thinking and promising early results.
Machine-learning system based on light could yield more powerful, efficient large language models (news.mit.edu)
Risky Giant Steps Can Solve Optimization Problems Faster (www.quantamagazine.org)
This is about Benjamin Grimmer’s paper arxiv.org/abs/2307.06324 where he proves under certain conditions that large steps lead to faster convergence.
“AI” Hurts Consumers and Workers -- and Isn’t Intelligent (techpolicy.press)
cross-posted from: lemmy.ml/post/2811405...
New AI systems collide with copyright law (www.bbc.co.uk)
3D-LLM Injecting the 3D World into Large Language Models (vis-www.cs.umass.edu)
PaLM-E: An embodied multimodal language model (ai.googleblog.com)
Almost All Research on the Mind is in English. That May Be a Problem (www.wired.com)
Large language models encode clinical knowledge (www.nature.com)
An update on Google’s efforts at LLMs in the medical field.
Google’s language model “NotebookLM” app hits public testing (arstechnica.com)
Generative AI Goes 'MAD' When Trained on AI-Created Data Over Five Times (www.tomshardware.com)
New ChatGPT rival, Claude 2, launches for open beta testing (arstechnica.com)
GPT-4 API general availability and deprecation of older models in the Completions API (openai.com)
Training AI on other AI causes models to collapse (original title : The AI is eating itself) (www.platformer.news)
Hi lemmings, what do you think about this and do you see a parallel with the human mind ?...
New ROCm™ 5.6 Release Brings Enhancements and Optimizations for AI and HPC Workloads (community.amd.com)
cross-posted from: https://lemmy.world/post/811496...
Full DragGAN source code is now released: Interactive Point-Based Manipulation of Images (github.com)
MPT-30B: Raising the bar for open-source foundation models (www.mosaicml.com)
and another commercially viable open-source LLM!
MIT researchers make language models scalable self-learners (news.mit.edu)
TLDR Summary:...
Accelerating Drug Discovery With the AI Behind ChatGPT – Screening 100 Million Compounds a Day (scitechdaily.com)
TLDR summary:...