Anthropic Claude LLM
Jump to navigation
Jump to search
An Anthropic Claude LLM is an Anthropic LLM.
- Example(s):
- Claude 1.
- Claude 2 [1].
- Claude 2.1 [2].
- ...
- Counter-Example(s):
- OpenAI LLM, such as: GPT-4.
- See: [[]].
References
2023
- https://www.anthropic.com/index/claude-2-1
- QUOTE:
- 200K token context window - Claude can now process documents up to 200,000 tokens long, over 500 pages of text. This allows analyzing entire codebases, long documents.
- 2x decrease in hallucination rates - Claude 2.1 is significantly less likely to provide false information compared to the previous version. This improves reliability and trust.
- New tool use beta feature - Allows Claude to integrate with users' processes, products, APIs to orchestrate workflows, retrieve info from databases/knowledge bases.
- Upgrades to developer experience - New Workbench product for iterating on prompts, system prompts for customizing Claude's behavior, generating code snippets.
- Claude 2.1 now powers free and paid chat interface at claude.ai. The 200k token context window is for paid users.
- Overall, Claude 2.1 focuses on enterprise capabilities - accuracy, honesty, customizability. Key goals are building trust in AI systems and enabling business use cases.
- QUOTE: