# US 2026/0099766 A1 · Adaptive data loader for bridging legacy data sources and artificial intelligence model training

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0099766 A1
- Document ID: US20260099766A1
- Status: Published application
- Kind: A1
- Application number: 19/064,832
- Filed: 2025-02-27
- Published: 2026-04-09
- Earliest claimed: 2024-10-04
- Assignees: Lenovo Global Technology (UnitedStates) Inc.
- Applicants: Lenovo Global Technology (UnitedStates) Inc.
- Inventors: Jeffrey Daniel Esposito; Aishwarya Dharani Arul; Henry Svendsgaard
- Web page: https://patentgrep.com/patent/US20260099766A1

## CPC

- G06N 3/0985: Hyperparameter optimisation; Meta-learning; Learning-to-learn
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06N 3/045: Combinations of networks
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06N 20/00: Machine learning
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A process includes identifying a plurality of legacy data sources, wherein each legacy data source has a unique legacy data format, and identifying, for each of the legacy data sources, a data loader that is adapted to process data from the legacy data source to form training data having a training data format that is different than the legacy data format. The process further includes causing, for each of the legacy data sources, the identified data loader to process at least a portion of the data from the legacy data source to form a batch of training data having the training data format. Still further, the operations comprise training an artificial intelligence model using the batches of training data formed from each of the legacy data sources.

## Claims

### Claim 1

A computer program product comprising a non-transitory computer readable storage medium and program instructions embodied therein, the program instructions being configured to be executable by a processor to cause the processor to perform operations comprising:
identifying a plurality of legacy data sources, wherein each legacy data source has a unique legacy data format;
identifying, for each of the legacy data sources, a data loader that is adapted to process data from the legacy data source to form training data having a training data format that is different than the legacy data format;
causing, for each of the legacy data sources, the identified data loader to process at least a portion of the data from the legacy data source to form a batch of training data having the training data format; and
training an artificial intelligence model using the batches of training data formed from each of the legacy data sources.

### Claim 2 (depends on 1)

The computer program product of claim 1, wherein identifying, for each of the legacy data sources, a data loader that is adapted to process data from the legacy data source includes:
searching a plurality of data loader records for a data loader record that identifies the legacy data source, wherein each data loader record identifies one of the legacy data sources and a data loader that is uniquely adapted to process the identified legacy data source.

### Claim 3 (depends on 1)

The computer program product of claim 1, wherein the artificial intelligence model is a language model.

### Claim 4 (depends on 1)

The computer program product of claim 1, wherein the training of the artificial intelligence model includes unsupervised training of the artificial intelligence model using the batches of training data formed from each of the legacy data sources.

### Claim 5 (depends on 1)

The computer program product of claim 1, wherein the training data format of the training data formed by each data loader is a standardized format for training of the artificial intelligence model.

### Claim 6 (depends on 1)

The computer program product of claim 1, wherein the training data formed by each data loader differs from the legacy data source by the data format, data type and/or data structure.

### Claim 7 (depends on 1)

The computer program product of claim 1, the operations further comprising:
augmenting the training data formed from the legacy data source to enhance diversity of the training data, wherein the training of the artificial intelligence model uses the processed training data and the augmented training data.

### Claim 8 (depends on 1)

The computer program product of claim 1, wherein the training of the artificial intelligence model enables the artificial intelligence model to perform natural language processing, computer vision, or audio processing.

### Claim 9 (depends on 1)

The computer program product of claim 1, the operations further comprising:
requesting that one or more of the data loaders form an additional batch of training data, wherein the one or more data loaders form the additional batch of training data in response to receiving the request for additional batches of training data;
obtaining the additional batch of training data; and
training of the artificial intelligence model using the additional batch of training data.

### Claim 10 (depends on 1)

The computer program product of claim 1, the operations further comprising:
causing two or more of the data loaders to process data from two or more of the legacy data sources at the same time.

### Claim 11 (depends on 1)

The computer program product of claim 1, the operations further comprising:
causing two or more instances the same data loader to process data from the same legacy data source to form two or more batches of training data at the same time.

### Claim 12 (depends on 1)

The computer program product of claim 1, wherein the training of the artificial intelligence model includes fine-tuning the artificial intelligence model using the batches of training data formed from each of the legacy data sources.

### Claim 13 (depends on 1)

The computer program product of claim 1, the operations further comprising:
determining whether output from the artificial intelligence model includes a hallucination; and
fine-tuning the artificial intelligence model using the batches of training data formed from each of the legacy data sources in response to positively determining that the output from the artificial intelligence model includes a hallucination.

### Claim 14 (depends on 1)

The computer program product of claim 1, wherein the training data formed by the data loaders includes unlabeled data.

### Claim 15 (depends on 1)

The computer program product of claim 1, wherein the legacy data source includes data and metadata, and wherein the data loader uses at least a portion of the data and at least a portion of the metadata to form the training data.

### Claim 16 (depends on 1)

The computer program product of claim 1, wherein the legacy data source is a file, database or other collection of data.

### Claim 17 (depends on 1)

The computer program product of claim 1, the operations further comprising:
measuring a performance metric of the artificial intelligence model during training using a validation dataset; and
adjusting a learning rate used in training of the artificial intelligence model on a subsequent data batch in response to the measured performance metric of the artificial intelligence model.

### Claim 18 (depends on 1)

The computer program product of claim 1, the operations further comprising:
establishing an upper performance threshold value and a lower performance threshold value;
measuring a performance metric of the artificial intelligence model during training using a validation dataset;
decreasing the learning rate used in training of the artificial intelligence model on subsequent data batch in response to the measured performance metric less than the lower performance threshold value; and
increasing the learning rate used in training of the artificial intelligence model on subsequent data batch in response to the measured performance metric being greater than the upper performance threshold value.

### Claim 19 (depends on 18)

The computer program product of claim 18, wherein the performance metric of the artificial intelligence model is measured after the artificial intelligence model has been trained on each batch of training data.

### Claim 20 (depends on 1)

The computer program product of claim 1, wherein the legacy data sources include a plurality of data source versions having the same data source type, wherein each data loader is identified by a version number that reflects unique compatibility with one of the data source versions.
