Abstract: A system for versioning generative components may implement a generative computational module to provide an output dataset in response to an input dataset, and a versioning module. The versioning module may commit versioned resources to a store on the memory device to generate snapshots of the resources on the memory device in different states, where the versioned resources include a representation of the generative computational module in a particular state and at least one of the input dataset to the generative computational module in the particular state, or outputs of the generative computational module in the particular state in response to the inputs. The versioning module may further check out one or more of the snapshots of any of the resources when prompted for execution by the generative computational module.

Abstract: A system for agent training may receive one or more tasks for a character to complete in one or more training sessions of an environment, where the character is controlled by a player that is selectable between a human user and an agent implementing an active model. The system may dynamically switch the player between the human user and the agent during the training sessions, where at least one condition for switching the player is based on an introspection metric associated with at least one of a confidence of the agent or observed actions when the player is the human user. The system may further capture observations from the character in the environment and corresponding actions of the character in the environment from the player during the training sessions as experience data, and train one or more models for the agent based on the experience data.

Abstract: A system for visualizing telemetry of characters may capture telemetry data of characters within an environment, where the characters are controlled by human or agent players, and where the telemetry data for each of the characters includes a temporal stream of observations provided to the characters and actions taken by the characters. The system may further capture video data of the characters with the telemetry data, embed metadata into the telemetry data or the video data associated with a state of the environment, synchronize the telemetry data and the video data into a unified time stream based at least in part on the metadata, provide a user with an interactive interface allowing the user to select at least one of a portion of the unified time stream for visualization or a visualization axis, and provide a user with visualizations of the unified time stream based on selections.

Abstract: An agent deployment system may prompt a user to define a feature space for an agent to operate as a character in an environment, where the feature space includes an observation space available to the agent in the environment and an action space available to the agent in the environment. The system may further construct the agent based on the feature space, where the agent is suitable for operating as the character in the environment by receiving observations from the observation space and performing actions from the action space. The system may further iteratively refine the agent by operating the agent in the environment with human interaction. The system may further deploy the agent in the environment.