Name: DAIGT Proper Train Dataset
Creator: Kaggle
License: https://opensource.org/licenses/MIT

Version 2 updated on 11/2/2023:

Since there is no proper train dataset for LLM - Detect AI Generated Text competition, I decided to create one.

Ingredients (please upvote the included datasets!):

New version includes:

Version 3 updated on 11/3/2023:

Additional 2400+ AI examples generated with Mistral 7B instruct and a new prompt (let's see how it works!)

Version 4 updated on 11/5/2023:

@kaggle.thedrcat_daigt_proper_train_dataset.train_drcat_01

CREATE TABLE train_drcat_01 (
  "text" VARCHAR,
  "label" BIGINT,
  "source" VARCHAR,
  "fold" BIGINT
);

@kaggle.thedrcat_daigt_proper_train_dataset.train_drcat_02

CREATE TABLE train_drcat_02 (
  "essay_id" VARCHAR,
  "text" VARCHAR,
  "label" BIGINT,
  "source" VARCHAR,
  "prompt" VARCHAR,
  "fold" BIGINT
);

@kaggle.thedrcat_daigt_proper_train_dataset.train_drcat_03

CREATE TABLE train_drcat_03 (
  "essay_id" VARCHAR,
  "text" VARCHAR,
  "label" BIGINT,
  "source" VARCHAR,
  "prompt" VARCHAR,
  "fold" BIGINT
);

@kaggle.thedrcat_daigt_proper_train_dataset.train_drcat_04

CREATE TABLE train_drcat_04 (
  "essay_id" VARCHAR,
  "text" VARCHAR,
  "label" BIGINT,
  "source" VARCHAR,
  "prompt" VARCHAR,
  "fold" BIGINT
);