bradfordlevy commited on
Commit
55df575
·
verified ·
1 Parent(s): 00e6abb

Upload ./README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +2 -4
README.md CHANGED
@@ -33,16 +33,14 @@ configs:
33
  ---
34
 
35
  # 🫘🧮 BeanCounter
36
-
37
  ## Datset Summary
38
-
39
  BeanCounter is a low-toxicity, large-scale, and open dataset of business-oriented text. See [Wang and Levy (2024)](https://arxiv.org/abs/2409.17827) for details of the data collection, analysis, and some explorations of using the data for continued pre-training.
40
 
41
- The data is sourced from the Electronic Data Gathering and Retrieval (EDGAR) system operated by the United States Securities and Exchange Commission (SEC). Specifically all filings submitted to EDGAR from 1996 through 2023 (validation splits are based on a random sample of data from January and February of 2024). We include three configurations of the dataset: `clean`, `default`, and `sample`. These consist of:
42
 
43
  - `clean`: 159B tokens of cleaned text
44
  - `default`: 111B tokens of cleaned and deduplicated text (referred to as "final" in the paper)
45
- - `fraud`: 0.3B tokens of text filed during periods of fraud based on SEC [Accounting and Auditing Enforcement Releases](https://www.sec.gov/enforcement-litigation/accounting-auditing-enforcement-releases) and [Litigation Releases](https://www.sec.gov/enforcement-litigation/litigation-releases) (NOT deduplicated)
46
  - `sample`: 1.1B tokens randomly sampled from `default` stratified by year
47
 
48
  ## How can I use this?
 
33
  ---
34
 
35
  # 🫘🧮 BeanCounter
 
36
  ## Datset Summary
 
37
  BeanCounter is a low-toxicity, large-scale, and open dataset of business-oriented text. See [Wang and Levy (2024)](https://arxiv.org/abs/2409.17827) for details of the data collection, analysis, and some explorations of using the data for continued pre-training.
38
 
39
+ The data is sourced from the Electronic Data Gathering and Retrieval (EDGAR) system operated by the United States Securities and Exchange Commission (SEC). Specifically all filings submitted to EDGAR from 1996 through 2023 (validation splits are based on a random sample of data from January and February of 2024). We include four configurations of the dataset: `clean`, `default`, `fraud`, and `sample`. These consist of:
40
 
41
  - `clean`: 159B tokens of cleaned text
42
  - `default`: 111B tokens of cleaned and deduplicated text (referred to as "final" in the paper)
43
+ - `fraud`: 0.3B tokens of text filed during periods of fraud according to SEC [Accounting and Auditing Enforcement Releases](https://www.sec.gov/enforcement-litigation/accounting-auditing-enforcement-releases) and [Litigation Releases](https://www.sec.gov/enforcement-litigation/litigation-releases) (Note that this content is not deduplicated)
44
  - `sample`: 1.1B tokens randomly sampled from `default` stratified by year
45
 
46
  ## How can I use this?