Other

Openai Opco, LLC

Raul Gonzalez

Ilge Akkaya

Jie Tang

Peter Zhokhov

Bowen Baker

Jeffrey Clune

Brandon Houghton

Adrien Ecoffet

Young Paul Yee

Connie Gonzalez

Joost Huizanga

Joost Huizinga

Disclosed herein are methods, systems, and computer-readable media for training a machine learning model to label unlabeled data and/or perform automated actions. In an embodiment, a method comprises receiving unlabeled digital video data, generating pseudo-labels for the unlabeled digital video data, the generating comprising receiving labeled digital video data, training an inverse dynamics model (IDM) using the labeled digital video data, and generating at least one pseudo-label for the unlabeled digital video data, wherein the at least one pseudo-label is based on a prediction, generated by the IDM, of one or more actions that mimic at least one timestep of the unlabeled digital video data. In some embodiments, the method further comprises adding the at least one pseudo-label to the unlabeled digital video data and further training the IDM or a machine learning model using the pseudo-labeled digital video data.

Using machine learning to train and use a model to perform automatic interface actions based on video and input datasets

Comb

A debris recovery device for drainage systems that includes a shaft with a peripheral wall. A basket made out of a meshed material has a spring loaded hingedly mounted bottom wall that is partially opened when it carries a load of a predetermined weight. A peripheral flange rests on the shoulder of the peripheral wall. An oil trap assembly can be optionally added to catch oil spills before reading the drainage system. The oil trap includes a sheet of a porous oil-absorbing material.

Debris recovery for drainage system

This apparatus is an add-on computer component peripheral for existing computer systems. The mouse/keyboard capture recorder (MCR) is capable of recording keyboard keystrokes, mouse coordinate movements, and mouse actions such as left, middle and right mouse clicks and/or mouse drag movements. The MCR has external cables that connect inline to a host computer's external keyboard and mouse without modification on the host computer system and its peripherals. The MCR apparatus operates independently of the hardware and software that is installed on a host computer system. A microprocessor monitors and controls all operations and stores mouse actions and keyboard actions in non-volatile memory during the MCR's record mode. In the MCR play mode, the apparatus will transmit the stored keyboard/mouse data by sending electrical signals representing the stored data to a host computer system via the mouse port and the keyboard port. The apparatus may be installed permanently or temporarily in the keyboard and mouse line without taking any precautions to save the recorded commands or actions when the computer is powered down or when the apparatus is removed from the user's computer system. The apparatus is compatible with all computers that utilize external mouse devices and/or external keyboards and is transparent to the user in that the computer system will operate the same with or without the apparatus attached. It is compatible with all computer programs and needs no support software or additional circuit boards installed in the (main) computer system to operate.

Mouse/keyboard capture recorder (MCR) apparatus and methodology

Growing community of inventors

Miami, FL, United States of America

Raul Gonzalez