Abstract: A system for training and deploying a sanitizer model of a large language model (LLM) to sanitize the harmful language text. The system generates paired data sets for training using LLM techniques. Training includes a first training phase using a paired dataset to predict the benign language text in the paired dataset based on loss of information computed between the benign language text in the paired dataset and the sanitized language text. Training also includes a second training phase using a harm model that computes a harm score of the LLM prediction by adjusting the trained sanitizing step to reduce the harm score for subsequent LLM responses.

Abstract: Certain aspects of the disclosure relate to profanity detection and mitigation. A method generally includes training a machine learning (ML) model using labeled training data instances by, for each training data instance: providing the tokens of the respective training data instance to an input layer of the ML model; receiving a first output for each token of the respective training data instance classifying the respective token as a profanity-containing or a non-profanity-containing token; receiving a second output for the respective training data instance classifying the respective training data instance as a profanity-containing or a non-profanity-containing instance; determining a loss value based on the first output for each token and the second output using a loss function comprising a regularization term configured to increase loss based on disagreement between the first output for each token and the second output; and modifying parameter(s) of the ML model based on the loss value.

Abstract: A system for training and deploying a sanitizer model of a large language model (LLM) to sanitize the harmful language text. The system generates paired data sets for training using LLM techniques. Training includes a first training phase using a paired dataset to predict the benign language text in the paired dataset based on loss of information computed between the benign language text in the paired dataset and the sanitized language text. Training also includes a second training phase using a harm model that computes a harm score of the LLM prediction by adjusting the trained sanitizing step to reduce the harm score for subsequent LLM responses.

Abstract: A vehicle mud flap system is presented. This system provides the ability to allow a user a single person to replace a mud flap on a large truck such as a mining truck in approximately an hour. Currently, 3-4 people and about 3 hours are needed. The vehicle mud flap system comprises a first elongated member, a second elongated member and a third elongated member. The first elongated member is designed to install onto a vehicle. The second elongated member is connected to a mud flap. The third elongated member is connected to a base member and includes an elongated slit. The second elongated member is designed to slide into the third elongated member with the mud flap passing through the elongated slit. The third elongated member is designed to slide into the first elongated member.

Abstract: A flange headed screw. The flanged headed screw has a threaded shank portion with a shank diameter, a distal tip, and a proximal end. A stop flange with a diameter that is larger than the shank diameter of the threaded shank located at the proximal end of the threaded shank portion. A projection portion extends from the stop flange, and the projection portion has a diameter smaller than the diameter of the stop flange. A driving head with a driving engagement located at a proximal end of the projection portion.