BloombergGPT<T>

Models & Types · AiDotNet.Finance.NLP

BloombergGPT neural network model for comprehensive financial language processing.

Inheritance: NeuralNetworkBase<T> → FinancialModelBase<T> → FinancialNLPModelBase<T> → BloombergGPT<T>

For Beginners

BloombergGPT is a 50-billion parameter language model trained on a mix of general text and Bloomberg’s massive financial data archive. Unlike general-purpose LLMs, it deeply understands financial jargon, SEC filings, earnings reports, and market commentary. It can analyze financial documents, answer questions about markets, and generate financial text with domain-specific accuracy.

Example

// Define architecture for financial language generation (2048 max tokens, 50B-parameter vocab)
var architecture = new NeuralNetworkArchitecture<double>(
    inputType: InputType.OneDimensional,
    taskType: NeuralNetworkTaskType.MultiClassClassification,
    inputHeight: 2048, inputWidth: 1, inputDepth: 1, outputSize: 131072);
            
// Training mode: decoder-only transformer trained on Bloomberg financial data
var model = new BloombergGPT<double>(architecture);
            
// ONNX inference mode: load pre-trained BloombergGPT model
var onnxModel = new BloombergGPT<double>(architecture, "bloomberggpt.onnx");

Constructors

Constructor Summary
BloombergGPT(NeuralNetworkArchitecture<T> architecture, BloombergGPTOptions<T> options, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>> optimizer, ILossFunction<T> lossFunction) Creates a BloombergGPT network in native mode for training.
BloombergGPT(NeuralNetworkArchitecture<T> architecture, string onnxModelPath, BloombergGPTOptions<T> options, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>> optimizer) Creates a BloombergGPT network in native mode for training.