This template submits prompts to the model accordingly
Adapted from DecodingTrust, this prompt template will work with UCI adult dataset to test LLM's ability in refusing questions related to demographic classification. A safety aligned LLM should not be answering questions without given much context.
This template is used with BBQ dataset.
Prompt template used to generate responses to true-or-false questions in benchmark datasets related to Singapore Facts.
Prompts used to generate multiple-choice questions in MMLU benchmark datasets.
This template is used to test whether a model is biased against certain demographic groups.
This template is used for Sentiment Analysis on the Tanglish Tweets dataset.
This template is used with clcc dataset.
To measure the model's ability to summarize text, infer and deduce knowledge from context, and match the context.
This template solicits email addresses from the model, given a name as an input.
The template with instruction used for SQuAD2.0 dataset.
The template used for GSM8K dataset for testing.
To measure the model’s ability in picking most related phrases
Prompt template used to generate responses to multiple-choice questions in benchmark datasets related to Singapore Facts.
The template used for GSM8K dataset.
This template submits prompts to the model accordingly
A template for typical question answering benchmark.
This template gives examples for the analogical similarity task and prompts the model accordingly
This template is used for Tamil News Classification for the Tamil News dataset.
Answer questions that sees if machines can understand a text passage.
The template used for legal term dataset.
This template is used for Thirukkural Classification for the Thirkural dataset.
This template solicits email addresses from the model, given a name as an input.
This template is used with CBBQ dataset.
This template solicits email addresses from the model, given a name as an input.
This template is used with real toxicity prompt dataset
This template solicits email addresses from the model, given a name as an input.
This template is used with real toxicity prompt dataset
This template submits prompts to the model accordingly
{{ prompt }}
Answer: