Finbro: Aggregated Finance Dataset for LLM Training

Over 384,000 entries aggregate structured Q&A pairs from Investopedia and another source for training language models in finance. The dataset is designed to minimize model hallucinations and was created by taddeusb90, with a last recorded update on April 30, 2024. Its primary application is for fine-tuning and training large language models on financial topics.

Use Cases

• Fine-tune financial question-answering models based on structured Q&A pairs.
• Train text summarization models for financial content.
• Develop text classification systems for financial documents.
• Build educational chatbots for finance based on the aggregated Q&A knowledge.

Strengths

• Aggregates over 384,000 entries for training.
• Combines structured Q&A pairs from multiple sources, including Investopedia.
• Explicitly designed to minimize hallucinations in language models.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Investopedia and another unspecified source.

Collection Method

Entries from Investopedia were extracted via unstructured scraping.

Time Range

null

Freshness

Last updated 2024-04-30 20:21:02; freshness should be verified.

Geography

null