curius graph

all pages

showing 80351-80400 of 203087 pages (sorted by popularity)
Universal Jailbreak Backdoors from Poisoned Human Feedback | SPY Lab
Simple distribution approximation: When sampled 100 times, can language models yield 80% A and 20% B? — AI Alignment Forum
How Can We Harness Pre-Training to Develop Robust Models? – gradient science
How to compute Hessian-vector products? | ICLR Blogposts 2024
Prompt Injection Defenses Should Suck Less
FACT SHEET: Vice President Harris Announces OMB Policy to Advance Governance, Innovation, and Risk Management in Federal Agencies’ Use of Artificial Intelligence | The White House
kronfluence/DOCUMENTATION.md at main · pomonam/kronfluence
A Fire Upon the Deep
Making a SOTA Adversarial Attack on LLMs 38x Faster | Haize Labs Blog 🕊️
Andy Jones
Empirical Evidence Against "The Longest Training Run" — LessWrong
Calibrating the Mosaic Evaluation Gauntlet | Databricks Blog
Manifest AI - Compute-Optimal Context Size
epochai.org/files/direct-approach.pdf
Fabien's Shortform — LessWrong
The Limited Benefit of Recycling Foundation Models – Epoch AI
ZebraLogic: Benchmarking the Logical Reasoning Ability of Language Models
Faith and Fate: Transformers as fuzzy pattern matchers – Answer.AI
nvlpubs.nist.gov/nistpubs/ai/NIST.AI.800-1.ipd.pdf
Reason for eigenvalues
III: Scaling to deep learning
There Are No Magic Outcome Variables | Elements of Evolutionary Anthropology
Hybrid Heterogeneous Clusters Can Lower the Energy Consumption of LLM Inference Workloads
Does Robustness Improve with Scale? | FAR AI
Demystifying AI Inference Deployments for Trillion Parameter Large Language Models | NVIDIA Technical Blog
Can You Trust An AI Press Release?—Asterisk
The Decline in Writing About Progress - by Matt Clancy
Behavioral red-teaming is unlikely to produce clear, strong evidence that models aren't scheming
If I wanted to spend WAY more on AI, what would I spend it on? — LessWrong
Scaling Automatic Neuron Description | Transluce AI
Memorandum on Advancing the United States’ Leadership in Artificial Intelligence; Harnessing Artificial Intelligence to Fulfill National Security Objectives; and Fostering the Safety, Security, and Trustworthiness of Artificial Intelligence | The White House
Trendlines in AIxBio evals – Lennart Justen
On LLM prompt optimization and amortization
Critical batch-size and effective dimension in Ordinary Least Squares
Tyler Cowen's Ethnic Dining Guide | All food is ethnic food.
Coalescence: making LLM inference 5x faster
Technical Deep-Dive: Curating Our Way to a State-of-the-Art Text Dataset
Release v1.0.0 · MilesCranmer/PySR
A Causal Framework for AI Regulation and Auditing — Apollo Research
benjaminedelman.com/assets/dissertation.pdf
Min-cut optimal(*) recomputation (i.e. activation checkpointing) with AOTAutograd - compiler - PyTorch Developer Mailing List
Biden’s Final Export Control Salvo Misfires
Deep Causal Transcoding: A Framework for Mechanistically Eliciting Latent Behaviors in Language Models — AI Alignment Forum
What to do to scale up?
How "Scope Sensitive" are o1-preview and GPT4o's forecasts? Not very.
Safe Predictive Agents with Joint Scoring Rules — AI Alignment Forum
Are We on the Brink of AGI? - by Steve Newman
Sharing Reality with Walt Whitman - YouTube
Extending control evaluations to non-scheming threats — AI Alignment Forum
rand.org/content/dam/rand/pubs/perspectives/PEA3700/PEA3776-1/RAND_PEA3776-1.pdf