Abstract: Methods, devices, and systems for training a deep learning recommendation model (DLRM), including: offloading one or more storage tables to a memory device; training a DLRM based on training data, wherein during the training, a storage table prefetched from the memory device is loaded according to a training phase; and storing feature data obtained during the training into the loaded storage table.

Abstract: A method prepares data for analysis. A user interface includes a flow pane, which includes a flow diagram with nodes, each node having an intermediate set and specifying an operation. A first user input selects a first node, which specifies a first operation. In response, the method displays node internals for the first node, including a description of the first operation. A second user input combines a second node with the first node. The second node specifies a second operation. In response to the second user input, the method (i) removes the second node from the flow diagram, (ii) determines ordering for the first and second operations, (iii) updates the node internals for the first node to specify the first and second operations, ordered according to the ordering, and (iv) displays the updated node internals, including descriptions of the first and second operations, sequenced according to the determined ordering.