Reaction
Anthropic's 2021 roadmap sparks a look back at AI tool use
One poster calls Anthropic early on reinforcement learning and tool use; a reply points to a LaMDA paper.
TLDR
A poster looking at Anthropic's 2021 roadmap says the company was early on reinforcement learning and tool use. They suggest two pre-training terms, “periodic batched backprop” and “skeletokenization,” might correspond to papers they link. In a reply, another commenter points to a LaMDA paper mentioning tool use in January 2022 as an example of how old the concept is.
Combined views
187
1 Source, first seen ago
2 likes1 comments1 saves
