Asharox/

summarization-sft-100k

Sentiment classification dataset in Alpaca instruction format for supervised fine-tuning (SFT). 100,000 examples sourced and normalised from:

mteb/mteb-human-tweet-sentiment-classification stanfordnlp/sst2 SetFit/sst5 (5-class collapsed to 3) jbeno/sentiment_merged

All labels normalised to 3 classes: negative, neutral, positive

	Format




	

Column Description

instruction Task prompt with label options… See the full description on the dataset page: https://huggingface.co/datasets/Asharox/summarization-sft-100k.