PaLM-SayCan

Parameters

540B

Published

Aug 16, 2022

PaLM-SayCan is an AI model developed by Google (United States), first published in August 2022. It works in the robotics domain, on tasks such as robotic manipulation.

Epoch AI has no training-compute estimate for this model. The model has 540,000,000,000 parameters. Training ran on 16 Google TPU v3 for about 100 hours.

Access: Unreleased. Its weights are not openly released. It is built on top of PaLM (540B). The reference paper has 3,081 citations. Epoch AI rates the confidence of this record as likely.

Full record

Organization: Google
Country of organization: United States
Domain: Robotics
Task: Robotic manipulation
Parameters: 540,000,000,000
Training hardware: Google TPU v3
Chips used: 16
Training time: 100 h
Training power draw: 14.4 kW
Model accessibility: Unreleased
Open weights: No
Base model: PaLM (540B)
Citations: 3,081
Epoch confidence: Likely