Abstract: Methods, systems, apparatuses, devices, and computer program products are described. A system may support retrieval-augmented generation (RAG) for a large language model (LLM). The system may use adaptive learning to improve the RAG process. For example, the system may implement a context-based embedding function to contextualize the RAG for the specific LLM or a specific tenant or user using the LLM. The context-based embedding function may project document vectors from a generic vector space into a context-based vector space for document retrieval. The system may retrieve a document using the context-based vector space to provide additional contextual information to the LLM to improve the LLM's output. The system may adaptively train the context-based embedding function based on the LLM, user feedback, or both. For example, the system may train the context-based embedding function to improve alignment of document retrieval likelihoods with confidence metrics for the outputs of the LLM.

Abstract: A service may receive, from users of an application that uses a prompt for accessing an LLM, a set of feedback indications associated with a set of responses from the LLM based on the prompt including a set of parameters. The service may transmit, to a first LLM, the set of feedback indications and the set of responses to obtain a set of feedback evaluations. The service may transmit, to a second LLM, the set of feedback evaluations to obtain a summary of the set of feedback evaluations. The service may transmit, to a third LLM, the summary of the set of feedback evaluations and the prompt associated with the set of responses to obtain a set of updated parameters for the prompt. The service may then configure the application to use the prompt with the set of updated parameters for accessing the LLM.

Abstract: Disclosed herein are systems and methods for generating custom datasets. For example, a method may include using one or more computer systems to gather a first dataset comprising example data relevant to a use case. The method may also include using a first artificial intelligence (AI) model implemented by the one or more computer systems to generate a second dataset. Input to the first AI model includes at least a portion of the first dataset. The method may also include configuring a second AI model using the second dataset. The gathering, generating, and configuring may occur within an integrated platform.

Abstract: A method may include obtaining a generative artificial intelligence (AI) model that includes a set of weights and that is associated with an explicit reward model and an implicit reward model. The method may include zeroing a partition function of the implicit reward model. The method may include obtaining feedback data associated with the explicit reward model that includes preference feedback data, binary feedback data, score feedback data, or any combination thereof. The method may include generating the explicit reward model based on the feedback data. The method may include fine-tuning the set of weights of the generative AI model based on a comparison of the explicit reward model and the implicit reward model and further based on the feedback data. The method may include receiving a query and generating, based on the query and the fine-tuned set of weights, a response that is responsive to the query.

Abstract: Embodiments also provide an LLM adapter training and merging framework that builds a new neural network model by merging a first LLM (stronger) with an adapter that has been trained in conjunction with a second LLM (weaker). Specifically, the adapter may be trained in conjunction with a smaller LLM to perform a specific task or adapt to a particular domain. The trained adapter is then merged with a different (larger) LLM to produce a new model. In this way, developers may select compatible LLMs as base models to merge with trained adapters to produce new models without additional training and/or finetuning the adapter with different LLMs. The one-time domain specific adapter training may be applied to any subsequent developments in merging compatible models with the trained specific adapter, thus enhancing computational efficiency of neural network model adaptation.

Abstract: Embodiments provide a merging framework that selectively merges pretrained model parameters of an LLM and retrained adapter weights. Specifically, the merging framework measures a similarity metric between a pretrained base LLM and an adapter that is retrained for a specific task or domain, and then prunes one or more components (weights or layers) of the adapter that have a high similarity with the base LLM and thus are likely to be redundant. The pruned adapter with only sparse features that are most dissimilar to the base LLM is then merged with the base LLM to produce a new neural network model that is adapted for the specific task or domain. In this way, redundant features may be pruned from adapter modules before merging.

Abstract: Embodiments described herein provide a parallel adapter-based training paradigm that trains multiple adapters in parallel for specific tasks or domains. The trained adapters are then selectively merged with a base neural network to produce a new finetuned neural network that is finetuned to perform the specific tasks. In this way, the parallel training largely improves computational efficiency to train or adapt a neural network for different tasks without repeated retraining of the entire neural network.

Abstract: In some systems, a set of sentences of a relatively large document may be vectorized into a set of vectors via an embedding model for summarization. Further, a subset of vectors of the set of vectors may be selected via a farthest point sampling (FPS) procedure based on a vector-space distance between respective vectors of the subset of vectors. Moreover, the subset of vectors that are associated with a subset of sentences may be ordered based on the order of the subset of sentences within the set of sentences of the document. Further, to generate a summary of the document, a query may be transmitted to a large language model (LLM) that includes a summarization prompt and the subset of sentences that correspond with the selected subset of vectors. A summary of the document may then be received from the LLM based on transmitting the query.

Abstract: A computer determines social media influencers in a specific topic by receiving a dataset of information associated with a website, the information including a first list of users of the website and a list of content that each user posts on the website, wherein each user is associated with other users from the first list of users. The computer determines initial values representing variables of the dataset of information on the website, wherein the variables include one or more topics for the list of content that each user from the first list of users posts on the website. The computer performs an iteration of Gibbs Sampling utilizing the initial values. The computer determines the one or more new values representing variables of the dataset represent a distribution of the one or more topics for the list of content that each user from the first list of users posts.

Abstract: A computer determines social media influencers in a specific topic by receiving a dataset of information associated with a website, the information including a first list of users of the website and a list of content that each user posts on the website, wherein each user is associated with other users from the first list of users. The computer determines initial values representing variables of the dataset of information on the website, wherein the variables include one or more topics for the list of content that each user from the first list of users posts on the website. The computer performs an iteration of Gibbs Sampling utilizing the initial values. The computer determines the one or more new values representing variables of the dataset represent a distribution of the one or more topics for the list of content that each user from the first list of users posts.

Abstract: A computer determines social media influencers in a specific topic. The computer receives a dataset of information on a website, the information including a list of users of the website and a list of content that each user posts, wherein each user is associated with one or more other users. The computer identifies a plurality of variables associated with the dataset, wherein the plurality of variables represent the information of the dataset on the website. The computer executes a topic specific search based on the plurality of variables, the topic search providing at least another list of users representing influencers in a specific topic.

Abstract: A computer determines social media influencers in a specific topic. The computer receives a dataset of information on a website, the information including a list of users of the website and a list of content that each user posts, wherein each user is associated with one or more other users. The computer identifies a plurality of variables associated with the dataset, wherein the plurality of variables represent the information of the dataset on the website. The computer executes a topic specific search based on the plurality of variables, the topic search providing at least another list of users representing influencers in a specific topic.