Datasets:
metadata
license: afl-3.0
task_categories:
- text-classification
language:
- en
pretty_name: 'Lifelike Quotient: Identifying Antropomorphism towards Chatbots'
size_categories:
- n<1K
Lifelike Quotient: Identifying Antropomorphism towards Chatbots
This dataset is used for training both a regression model and a reward model for identifying anthropomorphism towards chatbots.
The dataset contains 215 test-set samples and 2500 train-set samples. Each sample contains a prompt and a conversational agent's response. Each prompt does have 5 responses, with each response representing a gradual increase in anthropomorphism. Samples also include anthropomorphistic properties, which are used for identifying anthropomorphism towards chatbots.
The test-set include human-annotated scores. The train-set include LLM-as-a-judge scores.
This dataset was constructed for the exam of a master thesis, which developed the Lifelike Quotient, a model trained for identifying anthropomorphism towards chatbots.