An AI startup founded by a former engineer of OpenAI announces the second generation chat AI 'Claude 2', and a demonstration video to write code in chat is also being released
In March 2023,
Introducing Claude 2! Our latest model has improved performance in coding, math and reasoning. It can produce longer responses, and is available in a new public-facing beta website at https://t.co/uLbS2JNczH in the US and UK. pic.twitter.com/jSkvbXnqLd
—Anthropic (@AnthropicAI) July 11, 2023
Anthropic Claude 2
https://www.anthropic.com/index/claude-2
Anthropic releases Claude 2, its second-gen AI chatbot | TechCrunch
https://techcrunch.com/2023/07/11/anthropic-releases-claude-2-the-second-generation-of-its-ai-chatbot/
Claude is a chat AI that can access the chat interface of the developer console via an API and perform various conversational and text processing tasks. In-the-wild users rate Claude as being less likely to generate harmful responses than other chat AIs and remembering past conversations longer.
In the chat AI benchmark ranking released in May by the Large Model Systems Org (LMSYS Org), a research organization established by the University of California and Carnegie Mellon University, Claude ranked second after GPT-4-based ChatGPT. bottom. Claude users can enter up to 100,000 tokens of data at the prompt, so it is also possible to handle hundreds of pages of literature and books at once.
ChatGPT based on GPT-4 is ranked first in the interactive chat AI benchmark ranking, Claude-v1 is second, Google's PaLM 2 is also ranked in the top 10 - GIGAZINE
And on July 11th, Anthropic announced a new model of Claude, 'Claude 2'. Anthropic said, 'Claude 2 has improved performance and allows for longer responses.' 'Claude 2 enhances our chat experience and is now generally available in the US and UK. We are working to make it available in more countries in the coming months.'
Cloude 2 scored in the top 10% of examinees on the Graduate Record Examination (GRE) reading and writing test required for entry into graduate school in North America, and also scored multiple-choice questions on the USMLE Medical Exam. I was able to get a passing score. Also, in the multiple-choice question of the Multistate Bar Examination (national common bar examination), the correct answer rate reached 76.5%, while the previous generation Cloude had a correct answer rate of 73%.
In addition, Claude 2 has significantly improved coding skills over its predecessor, scoring 71.2% on the Python coding test,
In fact, Anthropic has released a demo movie on Twitter that lets Claude 2 write the code for the interactive map.
Claude 2 has improved from our previous models on evaluations including Codex HumanEval, GSM8K, and MMLU. You can see the full suite of evaluations in our model card: https://t.co/fJ210d9utd pic.twitter.com/LLOuUNfOFV
—Anthropic (@AnthropicAI) July 11, 2023
If you want to see data when you hover over a country name on the map, you can tell Claude 2 to code it.
When you tell Claude 2, 'I want you to help me visualize the data,' it asks what kind of data you want to process.
Then Claude 2 will automatically analyze the data.
Claude 2 wrote the code when I told him, 'I want you to write a code that displays a numerical value when you hover the mouse cursor over it.'
It also summarizes the main points of the code.
Copy and paste this code and run it.
Then, when you move the cursor to the country on the map, the numerical value is now displayed.
Claude 2 is available for both web and paid APIs, but at the time of writing, it has been deployed only to US and UK users. 'We think it's important to bring these systems to market and understand how people are actually using them,' Sandy Banerjee, Anthropic's head of market development, told tech outlet TechCrunch. 'We will continue to strive to improve performance,' he said.
Related Posts:
in Software, Web Service, Video, Posted by log1h_ik