SECBERTOptions<T>

Options & Configuration · AiDotNet.Models.Options

Configuration options for SEC-BERT model specialized for SEC filings analysis.

Inheritance: ModelOptions → SECBERTOptions<T>

For Beginners

SEC-BERT specializes in regulatory filing language:

The Key Insight: SEC filings use formal legal and accounting language that differs from general financial news. Terms like “material adverse effect”, “going concern”, and “contingent liability” have specific meanings that SEC-BERT understands.

What Problems Does SEC-BERT Solve?

  • Analyzing 10-K/10-Q annual and quarterly reports
  • Processing 8-K current reports for material events
  • Extracting risk factors from Item 1A disclosures
  • Understanding MD&A (Management Discussion and Analysis) sections
  • Detecting changes in disclosure language over time

Key Benefits:

  • Trained on millions of SEC filing documents
  • Understands regulatory terminology and structure
  • Captures subtle changes in disclosure tone
  • Effective for compliance and risk assessment tasks

How It Works

SEC-BERT is a BERT model fine-tuned specifically on SEC filings including 10-K, 10-Q, 8-K, and other regulatory documents to understand financial disclosure language.

Reference: Loukas et al., “SEC-BERT: A Domain-Specific Language Model for SEC Filings”, 2022.

Constructors

Constructor Summary
SECBERTOptions() Initializes a new instance with default SEC-BERT configuration.
SECBERTOptions(SECBERTOptions<T> other) Initializes a new instance by copying from another instance.

Properties

Property Summary
FinancialNLPTaskType TaskType { get; set; } Task type for the SEC-BERT model (default: Classification).
double DropoutRate { get; set; } Dropout rate (default: 0.1).
double LearningRate { get; set; } Gets or sets the Adam learning rate used to fine-tune SEC-BERT.
int HiddenDimension { get; set; } Hidden dimension (default: 768 for BERT-base).
int IntermediateDimension { get; set; } Intermediate feed-forward dimension (default: 3072).
int MaxSequenceLength { get; set; } Maximum sequence length in tokens (default: 512).
int NumAttentionHeads { get; set; } Number of attention heads (default: 12).
int NumClasses { get; set; } Number of output classes for classification tasks (default: 2 for binary).
int NumLayers { get; set; } Number of transformer layers (default: 12).
int VocabularySize { get; set; } Vocabulary size (default: 30522 for BERT-base).

Methods

Method Summary
void Validate() Validates the SEC-BERT options.