This work is preliminary. Whereas we show that there are many features that seem *plausibly* relevant to safety applications, much more work is needed to establish that our approach is useful in practice.
@anthropicai
-
Claude’s Secrecy Feature: Information Withholding Mechanism
By
–
One notable example is a "secrecy" feature. We observe that it fires for descriptions of people or characters keeping a secret. Activating this feature results in Claude withholding information from the user when it otherwise would not.
-
Model Safety Features: Vulnerabilities, Deception, and Bias Detection
By
–
Among these millions of features, we find several that are relevant to questions of model safety and reliability. These include features related to code vulnerabilities, deception, bias, sycophancy, power-seeking, and criminal activity.
-
Abstract Features in AI Models Causally Influence Outputs
By
–
These features are remarkably abstract, often representing the same concept across contexts and languages, even generalizing to image inputs. Importantly, they also causally influence the model’s outputs in intuitive ways.
-
Claude 3 Sonnet: Millions of Features Extracted from Deployed Model
By
–
For the first time, we’ve extracted millions of features from a high-performing, deployed model (Claude 3 Sonnet). These features cover specific people and places, programming-related abstractions, scientific topics, emotions, among a vast range of other concepts.
-
Dictionary Learning Makes LLM Neurons More Interpretable
By
–
The problem: most LLM neurons are uninterpretable, stopping us from mechanistically understanding the models. In October, we showed that dictionary learning could decompose a small model into "monosemantic" components we call "features"—making the model more interpretable.
-
AI Geopolitics Experts Discuss Ethics at Hay Festival
By
–
We’ll be at @HayFestival this year supporting the ‘AI needs you’ session—featuring a conversation between @Cambridge_Uni
’s AI & Geopolitics Institute director @verityharding
, and AI Ethics & Society associate professor @jonniepenn
. Join us at the event: https://
hayfestival.com/p-21995-verity
-harding-talks-to-jonnie-penn.aspx
… -
Dynamic Prompt Generator with Variable Insertion for Testing
By
–
Our prompt generator also supports dynamic variable insertion, making it easy to test how your prompts perform across different scenarios. Start generating better prompts today: http://
console.anthropic.com -
Anthropic Console Now Generates Production-Ready AI Prompts
By
–
You can now generate production-ready prompts in the Anthropic Console.
— Anthropic (@AnthropicAI) 10 mai 2024
Describe what you want to achieve, and Claude will use prompt engineering techniques like chain-of-thought reasoning to create more effective, precise and reliable prompts. pic.twitter.com/TqylVRkfP5You can now generate production-ready prompts in the Anthropic Console. Describe what you want to achieve, and Claude will use prompt engineering techniques like chain-of-thought reasoning to create more effective, precise and reliable prompts.
-
Zoominfo Leverages Claude for RAG Application Development
By
–
Go-to-market platform @Zoominfo uses Claude to make actionable recommendations and drive value for their customers. Their use of prompt generation helped significantly reduce the time it took to build an MVP of their RAG application, all while improving output quality.