Dataset
earnings_calls_transcripts
The awinml/earnings_calls_transcripts dataset on the Hugging Face Hub packages a small collection of earnings-call transcript segments formatted as chat-style messages for fine-tuning language models. It is distributed as a parquet file and is tagged for text modality with libraries including datasets, pandas, mlcroissant, and polars.
Data preview
Dataset card
What this dataset contains
The release consists of a single text table built around a messages column, where each row stores a list of chat turns rather than raw transcript prose. The provided sample rows show a system-style prompt instructing a model to act as a financial earnings analyst, followed by analyst and possibly transcript content, indicating the data has been pre-formatted for instruction tuning rather than left as unstructured narrative. The release is small in scale, with a train/test split totaling roughly 1,250 rows and a size category between 1K and 10K examples.
Structure and fields
The schema is centered on a messages field whose entries are lists of objects, each carrying a content string and a role indicator, mirroring the OpenAI/HF chat-template convention used for conversational fine-tuning. The content element is typed as a string Value, while role supplies the speaker tag such as system, user, or assistant. No additional columns such as ticker, fiscal period, speaker name, or sentiment label are exposed in the schema, so structural metadata about the underlying earnings calls is not directly available without further processing.
How analysts use it
For alternative-data practitioners, the dataset is best viewed as a training resource for building small chat-tuned models that can summarize or answer questions about earnings-call language, rather than as a primary source of company-level signals. Because ticker, date, and speaker fields are absent from the published schema, it does not directly support fund-specific event studies or factor backtests, and any downstream mapping back to public-company transcripts would require external joins. Its main practical value is as a lightweight, ready-to-load parquet for prototyping earnings-call NLP workflows before scaling to larger proprietary corpora.
Documentation & references 1
Related datasets
Browse moreParsed eligibility criteria from 13,229 ClinicalTrials.gov studies representing the candidate pool gathered by two first-stage retrievers during TREC Clinical Trials 2021–2023, formatted as typed entity-relation graphs for reranking models.
The clinical-trials-patient-graphs dataset is a small tabular collection of patient-level clinical records distributed in Parquet format by the 2001jdev publisher. It organizes entities and relations extracted from trial data into structured rows spanning 2021 through 2023.
This dataset is a parsed snapshot of ClinicalTrials.gov records, distributed in Parquet format on the Hugging Face Hub under an unspecified license. It contains roughly 52,000 trial entries drawn from public registry filings.
clinical-trials-trec-qrels is a tabular relevance-judgment file distributed by the 2001jdev user on the Hugging Face Hub. It maps clinical-trial topic identifiers to NCT registry IDs with graded relevance scores.
The clinical-trials-trec-topics dataset is a small tabular collection of clinical-trial topic records distributed in Parquet format on the Hugging Face Hub. It appears to be derived from TREC clinical-trial retrieval topics, with rows partitioned by year.
SEC EDGAR 10-K & 10-Q financial statements for 2,600+ NASDAQ companies.
Discussion
Questions or observations about this dataset? Join the conversation on X.
No comments yet. Start the conversation on X.