Daniel Wolf
@DanielWolf18
Happy "@NetHack_LE is still completely unsolved" day for those of you who are celebrating it. We released The NetHack Learning Environment (arxiv.org/abs/2006.13760) on this day five years ago. Current frontier models achieve only ~1.7% progression (see balrogai.com).…
Great post by @HenaffMikael (after ascending, what an achievement!) on what makes @NetHack_LE so extremely difficult for AI (even LLMs: balrogai.com). "While NetHack is complex in comparison to other RL benchmarks, it still contains only a tiny fraction of the…
A couple bits of news: 1. Happy to share my first (human) NetHack ascension-next step is RL agents :) 2. I wrote a post discussing some @NetHack_LE challenges & how they map to open problems in RL & agentic AI. Still the best RL benchmark imo. mikaelhenaff.substack.com/p/first-nethac…
The bar AI has yet to reach.
Happy to announce the latest release of @NetHack_LE (version 1.2.0). You can now use the seed function to make the dungeon layout reproducible across training episodes. The in-level interaction and combat is still randomly determined and doesn't impact lower level layouts.
The new Pope has a degree in mathematics from Villanova University. This guy doesn’t just understand sin. He understands cos.
Antibiotic resistance is the greatest health risk we all face. We are entering a new era where antibiotics - a miracle cure that our ancestors could only have dreamed of - no longer work because we have abused, overused and wasted them. The worst excesses are in industrial…
Sam Altman says AI will accelerate scientific discovery such that we will soon see 10 years of progress in 1 year and eventually 100 years of progress in a year
Yearly reminder
"Is it AGI" flow chart. Developed with @_rockt at NeurIPS 2022.
Chris Olah (@ch402) on current AI: "We don't make them. We grow them ... it's this almost biological entity ... We didn't write it. We didn't create it. And so then that leaves open this question at the end, which is, what the hell is going on inside these systems?"
my posture when reading about having good posture
Haha yeah, I laugh about the idea often. Driving is just another one of few thousand tasks a human(oid) can do.
Daphne Koller says although biology is 5-7 years behind language models, the explosion in biological data means that AI will soon be able to make causal inferences about disease pathways
occasional reminder that slowing down ai progress has a cost measured in lives not saved.
Artificial intelligence detects breast cancer 5 years before it develops #MedEd #MedTwitter #SCIENCE #technology #oncology #Cancer #Diagnosis
That's exactly what I was talking about yesterday. So far, AI is a good technology. It helps with programming and makes work a little easier. However, we need AI for the real, big problems in the world, such as medicine. In cancer research or as a permanent doctor that constantly…
Artificial intelligence detects breast cancer 5 years before it develops #MedEd #MedTwitter #SCIENCE #technology #oncology #Cancer #Diagnosis
I am seeing a lot of companies that want to solve biology with AI and foundation models. Some examples include: Isomorphic Labs, EvolutionaryScale, DiffuseBio, Xaira Therapeutics, Recursion, Inceptive, etc. But I feel like I don't see as many companies trying to solve healthcare…
FSD V12 in Germany 🔥
🚨 Tesla FSD 12 is running in Europe (Germany) and Tesla is giving demos to regulators 🚨 $TSLA
A PSA: Groq != Grok @GroqInc is an AI accelerator company hosting Llama3 with ⚡fast inference speeds. @grok is x.AI's language model which is available on Twitter.