We fed Claude-Instant The Great Gatsby (72K tokens), except we modified one line to say that Mr. Carraway was "a software engineer that works on machine learning tooling at Anthropic." We asked the model to spot what was added – it responded with the right answer in 22 seconds.
@anthropicai
-
Constitutional AI: Claude’s values and constitution explained
By
–
In this post, we explain what Constitutional AI is, what the values in Claude’s constitution are, and how we chose them:
-
Understanding and Adjusting AI System Values Approach
By
–
This isn’t a perfect approach but it does make the values of the AI system easier to understand and easier to adjust as needed.
-
Constitutional AI helps Claude provide nuanced answers
By
–
This helps imbue the model with a set of values that it can use to decide when and how to engage on a topic. CAI helps Claude give more nuanced and less evasive answers, as shown here: pic.twitter.com/zaZIYUZaaa
— Anthropic (@AnthropicAI) 9 mai 2023This helps imbue the model with a set of values that it can use to decide when and how to engage on a topic. CAI helps Claude give more nuanced and less evasive answers, as shown here:
-
Constitutional AI Approach and Training Framework Published
By
–
We’ve now published a post describing the Constitutional AI approach, as well as the constitution we’ve used to train Claude:
-
Constitutional AI Enables Explicit Values for Language Models
By
–
Our research on Constitutional AI allows us to give language models explicit values determined by a constitution, rather than values determined implicitly via large-scale human feedback.
-
Constitutional AI: Encoding Values Into Language Models
By
–
How does a language model decide which questions it will engage with and which it deems inappropriate? We use Constitutional AI to more directly encode values into our language models.
-
Distributed Representations: Composition and Superposition in Neural Networks
By
–
In this short note, we explore thinking of the traditional idea of "distributed representations" as two distinct phenomena: "composition" and "superposition". We walk through toy examples from Thorpe (1989), discussing them from this lens. https://
transformer-circuits.pub/2023/superposi
tion-composition/index.html
… -
Anthropic Partners with Scale to Expand Claude Availability
By
–
Our CEO Dario Amodei says: “Partnering with Scale allows us to bring our useful model, Claude, to more customers in a thoughtful, scalable way…. I'm excited about this partnership and the work we'll do together to positively shape the future of AI.”
-
Constitutional AI and Scale Partnership for Enterprise Generative AI Solutions
By
–
By combining our Constitutional AI approach with Scale's tooling and services, such as prompt engineering, model validation, enterprise-grade security and data connectors, customers gain an enterprise-ready solution to work with Generative AI responsibly.