Name: Gemma-rewrite-nbroad
Creator: Kaggle
License: https://www.apache.org/licenses/

About this Dataset

Gemma-rewrite-nbroad

Essays rewritten by gemma-7b-it on A100 in bfloat16 using TGI

The original essays were taken from https://huggingface.co/datasets/euclaise/writingprompts, which is from Reddit.

V1

2.1k essays. Prompts created using chatGPT. Essays generated by gemma-7b-it on A100 in bfloat16 There are about 100 different prompts, so each prompt was used multiple times.

V2

2.4k essays, each with a unique prompt (some prompts are very similar, though not exactly the same). Prompts created by mixtral 8x7b. Essays generated by gemma-7b-it on A100 in bfloat16.

Image created using DALLE in chatGPT.

Tables

Nbroad V1

@kaggle.nbroad_gemma_rewrite_nbroad.nbroad_v1

6.38 MB
2,166 rows
4 columns

CREATE TABLE nbroad_v1 (
  "id" VARCHAR,
  "original_text" VARCHAR,
  "rewrite_prompt" VARCHAR,
  "rewritten_text" VARCHAR
);

Nbroad V2

@kaggle.nbroad_gemma_rewrite_nbroad.nbroad_v2

7.28 MB
2,400 rows
4 columns

CREATE TABLE nbroad_v2 (
  "id" VARCHAR,
  "original_text" VARCHAR,
  "rewrite_prompt" VARCHAR,
  "rewritten_text" VARCHAR
);