dataset: common_gen templates: 45778bd5-fddc-4c60-a77b-4eff5bed1c0b: !Template answer_choices: null id: 45778bd5-fddc-4c60-a77b-4eff5bed1c0b jinja: "Ignoring the order of the concepts: {{ concepts | join(\", \") }}; \n\ Generate a sentence with all the concepts :\n|||\n{{target}}" metadata: !TemplateMetadata choices_in_prompt: false metrics: - BLEU - ROUGE original_task: true name: Given concepts - type 2 reference: Ignoring the order of the concepts:... 684f1859-0b8d-4efe-82e1-7218838813cd: !Template answer_choices: null id: 684f1859-0b8d-4efe-82e1-7218838813cd jinja: 'Put the concepts together to form a sentence: {{ concepts | join(", ") }}. ||| {{target}}' metadata: !TemplateMetadata choices_in_prompt: false metrics: - BLEU - ROUGE original_task: true name: Put together reference: This is similar to a task description a4991cc7-cc91-4f37-af80-1983a02eb950: !Template answer_choices: null id: a4991cc7-cc91-4f37-af80-1983a02eb950 jinja: "Construct a sentence with the word {{ concepts | choice }}. \n\nHint:\ \ Use {{concepts | join(\", \")}} to restrict the output sentence.\n|||\n{{target}}" metadata: !TemplateMetadata choices_in_prompt: false metrics: - BLEU - ROUGE original_task: true name: choice in concept centric sentence generation reference: '' b2033df4-7922-45b8-a113-e7784135cea9: !Template answer_choices: null id: b2033df4-7922-45b8-a113-e7784135cea9 jinja: "{% set seq = [ \n'From the concepts mentioned below, generate a sentence:',\ \ \n'Convert the concepts to a sentence:', \n'Given the list of concepts, write\ \ a sentence:'\n] %} \n{{ seq | choice }}\n{{ concepts | join(\", \") }}\n|||\n\ {{target}}" metadata: !TemplateMetadata choices_in_prompt: false metrics: - BLEU - ROUGE original_task: true name: random task template prompt reference: With a randomly chosen task template, generate a sentence from the list of concepts b7012213-04c4-424d-85fb-39d63d8a0ca2: !Template answer_choices: null id: b7012213-04c4-424d-85fb-39d63d8a0ca2 jinja: 'What are the topics in the sentence: {{target}} ||| {{ concepts | join(", ") }}' metadata: !TemplateMetadata choices_in_prompt: false metrics: - BLEU - ROUGE original_task: false name: topics from the sentence reference: The template generates a random topic from the sentence ed215962-8e51-45e7-b025-6e822f877098: !Template answer_choices: null id: ed215962-8e51-45e7-b025-6e822f877098 jinja: "We have the sentence: {{target}}; \nExtract all the key concepts: \n|||\n\ {{ concepts | join(\", \") }}" metadata: !TemplateMetadata choices_in_prompt: false metrics: - BLEU - ROUGE original_task: false name: sentence to concepts reference: Template identifies the concepts from the sentence f3fce748-89e0-4b88-94bb-72ddb9a04d10: !Template answer_choices: null id: f3fce748-89e0-4b88-94bb-72ddb9a04d10 jinja: 'Can you write a sentence about the topic {{concepts | choice}}? ||| {{target}}' metadata: !TemplateMetadata choices_in_prompt: false metrics: - BLEU - ROUGE original_task: false name: topic to sentence reference: 'Choose a random topic and ask the model to write a sentence about it ' fa787974-86dd-4f66-b2d7-6d3523ce00e1: !Template answer_choices: null id: fa787974-86dd-4f66-b2d7-6d3523ce00e1 jinja: "Humans can easily string together abstract concepts to form a coherent\ \ sentence. \nFor example, with the concepts {{ concepts | join(\", \") }},\ \ a simple sentence can be \n|||\n{{target}}" metadata: !TemplateMetadata choices_in_prompt: false metrics: - BLEU - ROUGE original_task: true name: Example prompt reference: The prompt is in the form of an example fbeebb4f-cba1-4bc4-80ec-758a3c134033: !Template answer_choices: null id: fbeebb4f-cba1-4bc4-80ec-758a3c134033 jinja: "Given the list of concepts: {{ concepts | join(\", \") }}; \nGenerate\ \ a sentence with all the concepts :\n|||\n{{target}}" metadata: !TemplateMetadata choices_in_prompt: false metrics: - BLEU - ROUGE original_task: true name: Given concepts type 1 reference: 'The prompt has the prefix "Given the .." '