Maplebear Inc.

Saurav Manchanda

Min Xie

Ramasubramanian Balasubramanian

Haixun Wang

Shishir Kumar Prasad

Prithvishankar Srinivasan

Shih-Ting Lin

Krishnakumar Subramanian

Riddhima Sejpal

Jonathan Newman

Luis Manrique

Gordon McCreight

Embodiments relate to utilizing a language model to automatically generate a novel recipe with refined content, which can be offered to a user of an online system. The online system generates a first prompt for input into a large language model (LLM), the first prompt including a plurality of task requests for generating initial content of a recipe. The online system requests the LLM to generate, based on the first prompt input into the LLM, the initial content of the recipe. The online system generates a second prompt for input into the LLM, the second prompt including the initial content of the recipe and contextual information about the recipe. The online system requests the LLM to generate, based on the second prompt input into the LLM, refined content of the recipe. The online system stores the recipe with the refined content in a database of the online system.

Using language model to generate recipe with refined content

An online concierge system uses a machine learning click through rate model to select promoted items based on user embeddings, item embeddings, and search query embeddings. Embeddings obtained by an embedding model may be used as inputs to the click through rate model. The embedding model may be trained using different actions to score the strength of a customer interaction with an item. For example, a customer purchasing an item may be a stronger signal than a customer placing an item in a shopping cart, which in turn may be a stronger signal than a customer clicking on an item. The online concierge system generates a ranking of candidate promoted items based on the search query and using the click through rate model. Based on the ranking, the online concierge system displays promoted items along with the organic search results to the customer.

Machine learning model for click through rate predication using three vector representations

A computer system uses clustering and a large language model (LLM) to normalize attribute tuples for items stored in a database of an online system. The online system collects attribute tuples, each attribute tuple comprising an attribute type and an attribute value for an item. The online system initially clusters the attribute tuples into a first plurality of clusters. The online system generates prompts for input into the LLM, each prompt including a subset of attribute tuples grouped into a respective cluster of the first plurality. Based on the prompts, the LLM generates a second plurality of clusters, each cluster including one or more attribute tuples that have a common attribute type and a common attribute value. The online system maps each attribute tuple to a respective normalized attribute tuple associated with each cluster. The online system rewrites each attribute tuple in the database to a corresponding normalized attribute tuple.

Using unsupervised clustering and language model to normalize attribute tuples of items in a database

An online concierge system selects content for presentation to a user by using a product scoring engine. The product scoring engine generates a user embedding for user data and a query embedding for query data. The product scoring engine generates an anchor embedding based on the user embedding and the query embedding, where the anchor embedding is an embedding in a product embedding space. The product scoring engine compares the anchor embedding to a set of product embeddings to score a set of products for presentation to a user.

Non-transitory computer-readable medium, and method for context-based content-scoring for an online concierge system

An online concierge system trains a classification model as a domain adversarial neural network from training data labeled with source classes from a source domain that do not overlap with target classes from a target domain output by the classification model. The online concierge system maps one or more source classes to a target class. The classification model extracts features from an image, classifies whether an image is from the source domain or the target domain, and predicts a target class for an image from the extracted features. The classification model includes a gradient reversal layer between feature extraction layers and the domain classifier that is used during training, so the feature extraction layers extract domain invariant features from an image.

Training a classification model using labeled training data that does not overlap with target classifications for the classification model

A server receives a plurality of product data entries from a plurality of retailer computing systems. Each product data entry includes a product identifier uniquely identifying a corresponding physical product and descriptive data of the corresponding physical product. A subset of the plurality of product data entries having a same product identifier is determined. An embedding vector representative of a product data entry in the subset is pairwise compared with each of respective embedding vectors representative of other product data entries in the subset other than the product data entry to compute respective vector similarity metrics. A pooled semantic similarity metric for the product data entry based on the computed respective vector similarity metrics. It is determined whether the product data entry is an outlier in the subset based on the pooled semantic similarity metric. A notification is transmitted to a client device of a user based on the determination.

Verifying association between physical products and identifiers of physical products

Machine learning model for click through rate prediction using three vector representations

An online concierge system uses a domain-adaptive suggestion module to score products that may be presented to a user as suggestions in response to a user's search query. The domain-adaptive suggestion module receives data that is relevant to scoring products as suggestions in response to a search query. The domain-adaptive suggestion module uses one or more domain-neutral representation models to generate a domain-neutral representation of the received data. The domain-neutral representation is a featurized representation of the received data that can be used by machine-learning models in the search domain or the suggestion domain. The domain-adaptive suggestion module then scores products by applying one or more machine-learning models to domain-neutral representations generated based on those products. By using domain-neutral representations, the domain-adaptive suggestion module can be trained based on training examples from a similar prediction task in a different domain.

Growing community of inventors

Minneapolis, MN, United States of America

Saurav Manchanda