Claude 2 is a new language model from Anthropic, with a 100K token context window currently available to users (a 200K token context window may become available later). This video describes the details provided by Anthropic in their model card.
Timestamps:
00:00 - Claude 2
01:11 - The larger context window
02:33 - Translation abilities
03:44 - Standard benchmarks (HumanEval, MMLU etc.)
05:31 - Multistate Bar Examination results
05:34 - United States Medical Licensing Examination
07:05 - Areas for Improvement
08:43 - Model details
09:02 - Confabulations
10:23 - Claude 2's constitution
10:46 - Red teaming
11:31 - Combined helpful, honest and harmless evaluations
Topics: ##ai #anthropic #claude2 #contextwindow
Link to the paper: https://www-files.anthropic.com/produ...
For related content:
- Twitter: / samuelalbanie
- Research lab: https://caml-lab.com/
- personal webpage: https://samuelalbanie.com/
- YouTube: / @samuelalbanie1
- TikTok: / samuelalbanie
- Instagram: / samuelalbanie
- LinkedIn: / samuel-albanie
- Threads: https://www.threads.net/@samuelalbanie
- Discord server for filtir: / discord
(Optional) if you'd like to support the channel:
- https://www.buymeacoffee.com/samuelal...
- / samuel_albanie