Dexterous In-Hand Manipulation [control policy]

Training compute

2.2×10²⁰ FLOP

Parameters

3.2M

Published

Aug 1, 2018

Dexterous In-Hand Manipulation [control policy] is an AI model developed by OpenAI (United States), first published in August 2018. It works in the robotics domain, on tasks such as robotic manipulation.

Training it took an estimated 2.2×10²⁰ FLOP of compute (estimation method: hardware). The model has 3,181,588 parameters. Training ran on 8 NVIDIA V100 for about 50 hours.

Access: Unreleased. Its weights are not openly released. The reference paper has 2,156 citations. Epoch AI rates the confidence of this record as confident.

Full record

Organization: OpenAI
Country of organization: United States
Domain: Robotics
Task: Robotic manipulation
Training compute: 2.2×10²⁰ FLOP
Compute estimation method: Hardware
Parameters: 3,181,588
Training hardware: NVIDIA V100
Chips used: 8
Training time: 50 h
Training power draw: 5.0 kW
Model accessibility: Unreleased
Open weights: No
Citations: 2,156
Epoch confidence: Confident