Abstract: Methods, systems, apparatuses, and computer program products are described herein that enable execution of a large AI model on a memory-constrained target device that is communicatively connected to a parameter server, which stores a master copy of the AI model. The AI model may be dissected into smaller portions (e.g., layers or sub-layers), and each portion may be executed as efficiently as possible on the target device. After execution of one portion of the AI model is finished, another portion of the AI model may be downloaded and executed at the target device. To improve efficiency, the input samples may be divided into microbatches, and a plurality of microbatches executing in sequential order may form a minibatch. The size of the group of microbatches or minibatch can be manually or automatically adjusted to reduce the communication overhead.

Abstract: The present techniques generally concern techniques for managing horticultural load shedding. A system includes a plurality of horticultural light sources powered by an electrical utility, the plurality of horticultural light sources being disposed in a horticultural structure to irradiate at least one plant or crop; a control module in data communication with the electrical utility, the control module being adapted to receive information associated with at least one load shedding event and, in response thereto, produce a control signal; and a rendering module adapted to receive and process the control signal; and send illumination instructions to the plurality of horticultural light sources to adjust an emission profile of the plurality of horticultural light, thereby mitigating potentially negative effects on said at least one plant or crop growth. A method is also provided.

Abstract: Embodiments of the present disclosure include systems and methods for compressing tokens based on positions for training data that is used to train transformer models. In some embodiments, a set of input data for training a transformer model is received. The set of input data comprises a set of tokens and a set of position values. A first token in the set of tokens that is the same as a second token in the set of tokens is identified. The position value representing the first token with the position value representing the second token are combined. The set of tokens is modified by removing the first token from the set of tokens. A set of training data is generated to comprise the modified set of tokens and the set of position values. The transformer model is trained using the set of training data.

Abstract: There are provided techniques, including systems and methods, for adjusting illumination conditions provided by a pool of light sources. The adjustments made to the illumination conditions are based on one or more objectives, hence enabling the possibility of generating objectives-based light recipes or scenarios. The system is connectable to the pool of light sources through appropriate components and protocols. The system includes a light scenario generation module, a rendering module, and a user interface, and the method is implemented using these components.

Abstract: Embodiments of the present disclosure include systems and methods for packing tokens to train sequence models. In some embodiments, a plurality of datasets for training a sequence model is received. Each dataset in the plurality of datasets includes a sequence of correlated tokens. A set of training data is generated that includes a subset of a sequence of tokens from a first dataset in the plurality of datasets and a subset of a sequence of tokens from a second, different dataset in the plurality of datasets. The sequence model is trained using the set of training data.

Abstract: Embodiments of the present disclosure include systems and methods for forcing weights of transformer model layers when training a transformer model. In some embodiments, input data is received at a first layer included in a transformer model. The input data is processed through the first layer of the transformer model to produce a first output data. The first output data is processed through the first layer of the transformer model to produce a second output data. The first output data is processed through a second layer included in the transformer model to produce a third output data. A difference is calculated between the second output data and the third output data. Weights included in the first layer of the transformer model are adjusted based on the calculated difference.

Abstract: Systems and methods related to hardware-assisted gradient optimization using streamed gradients are described. An example method in a system comprising a memory configured to store weights associated with a neural network model comprising L layers, where L is an integer greater than one, a gradient optimizer, and a plurality of workers is described. The method includes during a single burst cycle moving a first set of gradients, received from each of the plurality of workers, from at least one gradient buffer to the gradient optimizer and moving weights from at least one buffer, coupled to the memory, to the gradient optimizer. The method further includes during the single burst cycle writing back the new weights, calculated by the gradient optimizer, to the memory. The method further includes during the single burst cycle transmitting the new weights, from the gradient optimizer, to each of the plurality of workers.

Abstract: There is provided a method for determining pollination illumination patterns within a horticultural structure, including defining a control zone associated with a first light source, the first light source being configured to illuminate a control plant or crop according to a control lighting scenario; defining a test zone associated with a second light source, the second light source being configured to illuminate a test plant or crop according to a test lighting scenario; illuminating the control plant or crop with the first light source according to the control lighting scenario; illuminating the test plant or crop with the second light source according to the test lighting scenario in the test zone; and following a release of pollinators, determining whether the test lighting scenario impacts a pollination-related property of the test plant or crop with respect to a pollination-related property of the control plant or crop.

Abstract: There is provided a method for determining pest management illumination patterns including defining a control zone, the control zone being associated with a first light source, the first light source being configured to illuminate a control plant or crop according to a control lighting scenario; in the control zone, illuminating the control plant or crop with the first light source according to the control lighting scenario; defining test zones, each test zone being associated with a horticultural light source; for each test zone of the test zones: identifying an undesirable organism; providing a natural predator to the undesirable organism in the test zone; and illuminating a test plant or crop according to a corresponding test lighting scenario with the light source, the test lighting scenario having test optical property affecting a behavior of the natural predator.

Abstract: Embodiments of the present disclosure include systems and methods for compressing and decompressing data generated by sub-blocks in a neural network. In some embodiment, an input matrix is received at a compression block in the neural network. The compression block compresses the input matrix into a compressed matrix and outputs the compressed matrix. The compressed matrix has a reduced dimensionality relative to a dimensionality of the input matrix. A decompression block retrieves the compressed matrix. The decompression block decompresses compressed matrix into a decompressed matrix and outputs the decompressed matrix. The decompressed matrix has a same dimensionality as the dimensionality of the input matrix. The compression and decompression blocks are optimized based on feedback received from the neural network.

Abstract: The present techniques generally concern methods and systems for managing undesired effects in an electrical grid, which may include rapid voltage change(s) and/or flicker(s). The system includes an event detection module operatively connected to a plurality of horticultural light sources. The event detection module is configured to determine a power usage of the horticultural light sources, based on illumination conditions, detect an event affecting the illumination conditions, determine whether the event causes the undesired effects in the electrical grid, based on an evolution of the power usage of the horticultural light sources in response to the event, and send illumination instructions to the horticultural light sources to adjust the power usage of the horticultural light sources, if the event causes the undesired effects.

Abstract: The present techniques generally concern methods and systems for monitoring and managing reactive power from horticultural lighting sources in an electrical grid. The techniques provided herein include determining or predicting distortive effects produced by the horticultural lighting sources, evaluating a power factor of the horticultural light sources, and based on a target power factor, adjusting the power factor of the horticultural light sources. The techniques described herein allow for an optimization of the power factor of the horticultural lighting sources in order to reduce, mitigate or eliminate the negative effects generally associated with the operation of horticultural light sources on the electrical grid.

Abstract: A system for simulating and implementing lighting scenarios for driving a horticultural light source in accordance with illumination instructions.

Abstract: Embodiments of the present disclosure include systems and methods for packing tokens to train sequence models. In some embodiments, a plurality of datasets for training a sequence model is received. Each dataset in the plurality of datasets includes a sequence of correlated tokens. A set of training data is generated that includes a subset of a sequence of tokens from a first dataset in the plurality of datasets and a subset of a sequence of tokens from a second, different dataset in the plurality of datasets. The sequence model is trained using the set of training data.

Abstract: Embodiments of the present disclosure include systems and methods for determining position values for training data that is used to train transformer models. In some embodiments, a set of input data for training a transformer model is received. The set of input data comprises a set of tokens. Based on an offset value, a set of successive position values for the set of tokens is determined. Each position value in the set of successive position values represents a position of a token in the set of tokens relative to other tokens in the set of tokens. A set of training data is generated to comprise the set of tokens and the set of successive position values. The transformer model is trained using the set of training data.

Abstract: Embodiments of the present disclosure include systems and methods for packing tokens to train sequence models. In some embodiments, a plurality of datasets for training a sequence model is received. Each dataset in the plurality of datasets includes a sequence of correlated tokens. A set of training data is generated that includes a subset of a sequence of tokens from a first dataset in the plurality of datasets and a subset of a sequence of tokens from a second, different dataset in the plurality of datasets. The sequence model is trained using the set of training data.

Abstract: Embodiments of the present disclosure include systems and methods for training transformer models using position masking. In some embodiments, a set of data for training a transformer model is received. The set of data includes a sequence of tokens and a set of position values. Each position value in the set of position values represents a position of a token in the sequence of tokens relative to other tokens in the sequence of tokens. A subset of the set of position values in the set of data is selected. Each position value in the subset of the set of position values is replaced with a second defined value to form a second set of defined values. The transformer model is trained using the set of data.

Abstract: Embodiments of the present disclosure include systems and methods for reducing hardware resource utilization by residual neural networks. In some embodiments, a first matrix is received at a layer included in a neural network. The first matrix is compressed to produce a second matrix. The second matrix has a reduced dimensionality relative to a dimensionality of the first matrix. The second matrix is processed through a network block in the layer included in the neural network. The processed second matrix is expanded to produce a third matrix. The third matrix has a dimensionality that is equal to a dimensionality of the first matrix. The third matrix is added to the first matrix to produce a fourth matrix.

Abstract: The present techniques generally concern techniques for managing horticultural load shedding. A system includes a plurality of horticultural light sources powered by an electrical utility, the plurality of horticultural light sources being disposed in a horticultural structure to irradiate at least one plant or crop; a control module in data communication with the electrical utility, the control module being adapted to receive information associated with at least one load shedding event and, in response thereto, produce a control signal; and a rendering module adapted to receive and process the control signal; and send illumination instructions to the plurality of horticultural light sources to adjust an emission profile of the plurality of horticultural light, thereby mitigating potentially negative effects on said at least one plant or crop growth. A method is also provided.

Abstract: Systems and methods related to hardware-assisted gradient optimization using streamed gradients are described. An example method in a system comprising a memory configured to store weights associated with a neural network model comprising L layers, where L is an integer greater than one, a gradient optimizer, and a plurality of workers is described. The method includes during a single burst cycle moving a first set of gradients, received from each of the plurality of workers, from at least one gradient buffer to the gradient optimizer and moving weights from at least one buffer, coupled to the memory, to the gradient optimizer. The method further includes during the single burst cycle writing back the new weights, calculated by the gradient optimizer, to the memory. The method further includes during the single burst cycle transmitting the new weights, from the gradient optimizer, to each of the plurality of workers.