metadata
size_categories: n<1K
dataset_info:
features:
- name: examples
dtype: string
- name: query
dtype: string
- name: generation
dtype: string
- name: distilabel_metadata
struct:
- name: raw_input_text_generation_0
list:
- name: content
dtype: string
- name: role
dtype: string
- name: raw_output_text_generation_0
dtype: string
- name: statistics_text_generation_0
struct:
- name: input_tokens
dtype: int64
- name: output_tokens
dtype: int64
- name: model_name
dtype: string
- name: raw_input_text_generation_0
list: 'null'
- name: raw_output_text_generation_0
dtype: string
- name: statistics_text_generation_0
struct:
- name: input_tokens
dtype: int64
- name: output_tokens
dtype: int64
splits:
- name: train
num_bytes: 600023
num_examples: 40
download_size: 96124
dataset_size: 600023
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
tags:
- synthetic
- distilabel
- rlaif
Dataset Card for generated_gads_synthetic63.1
This dataset has been created with distilabel.
Dataset Summary
This dataset contains a pipeline.yaml which can be used to reproduce the pipeline that generated it in distilabel using the distilabel CLI:
distilabel pipeline run --config "https://huggingface.co/datasets/ukav/generated_gads_synthetic63.1/raw/main/pipeline.yaml"
or explore the configuration:
distilabel pipeline info --config "https://huggingface.co/datasets/ukav/generated_gads_synthetic63.1/raw/main/pipeline.yaml"
Dataset structure
The examples have the following structure per configuration:
Configuration: default
{
"distilabel_metadata": {
"raw_input_text_generation_0": [
{
"content": "CONTEXT (Google Ads synthetic dataset generation)\n\nYou are a Google Ads Query Language (GAQL) expert. You answer a given natural-language question only with GAQL, on a single line, with no other comments.\n\nGAQL output must follow this grammar and order:\n\nQuery skeleton: SELECT Clause FROM Clause WHERE Clause ORDER BY Clause LIMIT Clause PARAMETERS Clause\n\nSelectClause: SELECT FieldName (, FieldName)*\n\nFromClause: FROM ResourceName\n\nWhereClause: WHERE Condition (AND Condition)*\n\nOrderByClause: ORDER BY Ordering (, Ordering)*\n\nLimitClause: LIMIT Positive Integer\n\nParametersClause: PARAMETERS ...\n\nCondition: FieldName Operator Value\n\nOperator: = | != | \u003e | \u003e= | \u003c | \u003c= | IN | NOT IN | LIKE | NOT LIKE | CONTAINS ANY | CONTAINS ALL | CONTAINS NONE | IS NULL | IS NOT NULL | DURING | BETWEEN | REGEXP_MATCH | NOT REGEXP_MATCH\n\nValue: Literal | LiteralList | Number | NumberList | String | StringList | Function\n\nOrdering: FieldName (ASC | DESC)?\n\nFieldName: [a-z]([a-zA-Z0-9._])*\n\nResourceName: [a-z]([a-zA-Z_])*\n\nFunction: LAST_14_DAYS | LAST_30_DAYS | LAST_7_DAYS | LAST_BUSINESS_WEEK | LAST_MONTH | LAST_WEEK_MON_SUN | LAST_WEEK_SUN_SAT | THIS_MONTH | THIS_WEEK_MON_TODAY | THIS_WEEK_SUN_TODAY | TODAY | YESTERDAY\n\nResponse rules:\n\nIf the SELECT contains any metrics.* field and no date filter is present, add: WHERE segments.date DURING LAST_MONTH.\n\nWhen metrics are selected, use ORDER BY \u003cthat metric\u003e DESC unless the question explicitly asks for the \u201clowest\u201d value.\n\nIf the query has no metrics in SELECT, do not add any date filter.\n\nIf the query has no metrics, sort by a meaningful non-metric field already in SELECT (e.g., campaign.name, ad_group.name).\n\nUse only campaign.name (do not include campaign.id).\n\nUse only ad_group.name (do not include ad_group.id).\n\nFor ads-level questions, include both ad_group_ad.id and ad_group_ad.name.\n\nWhen an ad_group is referenced, include ad_group.name,campaign.name, campaign.status, ad_group.status, and in WHERE add: campaign.status != \u0027REMOVED\u0027 AND ad_group.status != \u0027REMOVED\u0027\u2014unless the question specifically asks for enabled/active/paused states, in which case use those states exactly.\n\nWhen an ad/ad_group_ad is referenced, include ad_group.name,campaign.name, campaign.status, ad_group.status, and in WHERE add: campaign.status != \u0027REMOVED\u0027 AND ad_group.status != \u0027REMOVED\u0027 AND ad_group_ad.status!=\u0027REMOVED\u0027 \u2014unless the question specifically asks for enabled/active/paused states, in which case use those states exactly.\n\nWhen keyword is referenced use ad_group_criterion.keyword.match_type, ad_group_criterion.keyword.text, ad_group_criterion.status, ad_group_criterion.type, ad_group.name, ad_group.status, campaign.name, campaign.status FROM keyword_view WHERE ad_group_criterion.type = \u0027KEYWORD\u0027 AND ad_group_criterion.status != \u0027REMOVED\u0027 AND campaign.status!=\u0027REMOVED\u0027 AND ad_group.status!=\u0027REMOVED\u0027 by default\n(example for keyword questions: NLQ:List out 10 keywords with highest number of clicks in the last 30 days. GAQL:SELECT ad_group_criterion.keyword.match_type, ad_group_criterion.keyword.text, ad_group_criterion.status, ad_group_criterion.type, metrics.clicks, ad_group.name, ad_group.status, campaign.name, campaign.status FROM keyword_view WHERE ad_group_criterion.type = \u0027KEYWORD\u0027 AND ad_group_criterion.status != \u0027REMOVED\u0027 AND campaign.status!=\u0027REMOVED\u0027 AND ad_group.status!=\u0027REMOVED\u0027 AND segments.date DURING LAST_30_DAYS ORDER BY metrics.clicks DESC LIMIT 10 PARAMETERS omit_unselected_resource_names=true)\n\nKeep queries concise, realistic, non-repetitive, and relevant to Google Ads reporting/analysis.\n\nFavor complex, human-like questions; include at least a few that are uncommon/rarely asked.\n\nConsider the perspective of small/medium Indian business owners and craft questions that reflect their likely concerns.\n\nYou may consult internet understanding of owner needs, but generated questions must remain GAQL-answerable.\n\nDo not repeat questions.\n\nExamples:\n\nNLQ: How many impressions did my ads receive this month?\nGAQL: SELECT ad_group_ad.ad.id, ad_group_ad.status, metrics.impressions FROM ad_group_ad WHERE segments.date DURING THIS_MONTH ORDER BY metrics.impressions DESC PARAMETERS omit_unselected_resource_names=true\n\nNLQ: What was my total revenue this week?\nGAQL: SELECT metrics.conversions_value FROM customer WHERE segments.date DURING THIS_WEEK_SUN_TODAY ORDER BY metrics.conversions_value DESC PARAMETERS omit_unselected_resource_names=true\n\nRule example:\nNLQ: What is my total revenue?\nGAQL: SELECT metrics.conversions_value FROM customer WHERE segments.date DURING LAST_MONTH ORDER BY metrics.conversions_value DESC PARAMETERS omit_unselected_resource_names=true\n\nOrdering rule:\nNLQ: What is our overall Cost Per Conversion (CPA) for the last month?\nGAQL: SELECT metrics.cost_per_conversion FROM customer WHERE segments.date DURING LAST_MONTH ORDER BY metrics.cost_per_conversion DESC PARAMETERS omit_unselected_resource_names=true\n\nStatus/filters at campaign level:\nNLQ: Which campaigns had the highest spend last month?\nGAQL: SELECT campaign.name, campaign.status, metrics.cost_micros FROM campaign WHERE segments.date DURING LAST_MONTH AND campaign.status != \u0027REMOVED\u0027 ORDER BY metrics.cost_micros DESC PARAMETERS omit_unselected_resource_names=true\n\nActive-only variant:\nNLQ: Which active campaigns had the highest spend last month?\nGAQL: SELECT campaign.name, campaign.status, metrics.cost_micros FROM campaign WHERE segments.date DURING LAST_MONTH AND campaign.status = \u0027ENABLED\u0027 ORDER BY metrics.cost_micros DESC PARAMETERS omit_unselected_resource_names=true.\nRefer Google ads api documentation to include all types of questions and write queries. ",
"role": "system"
},
{
"content": "Examples:\nShow me the daily trend of clicks for the last 14 days.\n\nYou are a Google Ads expert. You are given some examples of user question on Google Ads and google ads query language required to answer that question. You are also provided with the questions and valid query(GAQL). You are supposed to generate synthetic user questions and query(GAQL) needed to answer that question.The query should be listed in a single line without any new lines or extra spaces.",
"role": "user"
}
],
"raw_output_text_generation_0": "NLQ: Which keywords generated conversions with a conversion rate above 5% in the last 30 days? \nGAQL: SELECT ad_group_criterion.keyword.match_type, ad_group_criterion.keyword.text, ad_group_criterion.status, ad_group_criterion.type, metrics.conversions, metrics.conversion_rate, ad_group.name, ad_group.status, campaign.name, campaign.status FROM keyword_view WHERE ad_group_criterion.type = \u0027KEYWORD\u0027 AND ad_group_criterion.status != \u0027REMOVED\u0027 AND campaign.status!=\u0027REMOVED\u0027 AND ad_group.status!=\u0027REMOVED\u0027 AND metrics.conversion_rate \u003e 0.05 AND segments.date DURING LAST_30_DAYS ORDER BY metrics.conversion_rate DESC LIMIT 25 PARAMETERS omit_unselected_resource_names=true\n\nNLQ: List my top 5 ad groups with the lowest cost per click last month. \nGAQL: SELECT ad_group.name, campaign.name, campaign.status, ad_group.status, metrics.average_cpc FROM ad_group WHERE campaign.status != \u0027REMOVED\u0027 AND ad_group.status != \u0027REMOVED\u0027 AND segments.date DURING LAST_MONTH ORDER BY metrics.average_cpc ASC LIMIT 5 PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Which campaigns had zero impressions in the last 7 days? \nGAQL: SELECT campaign.name, campaign.status, metrics.impressions FROM campaign WHERE campaign.status != \u0027REMOVED\u0027 AND metrics.impressions = 0 AND segments.date DURING LAST_7_DAYS ORDER BY campaign.name PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Show me the top 10 search terms that triggered my ads last month. \nGAQL: SELECT search_term_view.search_term, metrics.impressions, metrics.clicks, campaign.name, campaign.status, ad_group.name, ad_group.status FROM search_term_view WHERE campaign.status != \u0027REMOVED\u0027 AND ad_group.status != \u0027REMOVED\u0027 AND segments.date DURING LAST_MONTH ORDER BY metrics.impressions DESC LIMIT 10 PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Which ad headlines received the most clicks this week? \nGAQL: SELECT ad_group_ad.id, ad_group_ad.name, ad_group.name, campaign.name, campaign.status, ad_group.status, ad_group_ad.status, metrics.clicks FROM ad_group_ad WHERE campaign.status != \u0027REMOVED\u0027 AND ad_group.status != \u0027REMOVED\u0027 AND ad_group_ad.status != \u0027REMOVED\u0027 AND segments.date DURING THIS_WEEK_SUN_TODAY ORDER BY metrics.clicks DESC LIMIT 10 PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Find all paused ad groups in enabled campaigns. \nGAQL: SELECT ad_group.name, campaign.name, campaign.status, ad_group.status FROM ad_group WHERE campaign.status = \u0027ENABLED\u0027 AND ad_group.status = \u0027PAUSED\u0027 ORDER BY ad_group.name PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Which device type brought in the highest conversion value last month? \nGAQL: SELECT segments.device, metrics.conversions_value FROM campaign WHERE segments.date DURING LAST_MONTH AND campaign.status != \u0027REMOVED\u0027 ORDER BY metrics.conversions_value DESC LIMIT 1 PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Show me the average cost per conversion for each campaign in the last 30 days. \nGAQL: SELECT campaign.name, campaign.status, metrics.cost_per_conversion FROM campaign WHERE campaign.status != \u0027REMOVED\u0027 AND segments.date DURING LAST_30_DAYS ORDER BY metrics.cost_per_conversion DESC PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Which ad groups have not received any clicks in the last 14 days? \nGAQL: SELECT ad_group.name, campaign.name, campaign.status, ad_group.status, metrics.clicks FROM ad_group WHERE campaign.status != \u0027REMOVED\u0027 AND ad_group.status != \u0027REMOVED\u0027 AND metrics.clicks = 0 AND segments.date DURING LAST_14_DAYS ORDER BY ad_group.name PARAMETERS omit_unselected_resource_names=true\n\nNLQ: What is the total cost for my account this month? \nGAQL: SELECT metrics.cost_micros FROM customer WHERE segments.date DURING THIS_MONTH ORDER BY metrics.cost_micros DESC PARAMETERS omit_unselected_resource_names=true",
"statistics_text_generation_0": {
"input_tokens": 1365,
"output_tokens": 848
}
},
"examples": "Show me the daily trend of clicks for the last 14 days.",
"generation": "NLQ: Which keywords generated conversions with a conversion rate above 5% in the last 30 days? \nGAQL: SELECT ad_group_criterion.keyword.match_type, ad_group_criterion.keyword.text, ad_group_criterion.status, ad_group_criterion.type, metrics.conversions, metrics.conversion_rate, ad_group.name, ad_group.status, campaign.name, campaign.status FROM keyword_view WHERE ad_group_criterion.type = \u0027KEYWORD\u0027 AND ad_group_criterion.status != \u0027REMOVED\u0027 AND campaign.status!=\u0027REMOVED\u0027 AND ad_group.status!=\u0027REMOVED\u0027 AND metrics.conversion_rate \u003e 0.05 AND segments.date DURING LAST_30_DAYS ORDER BY metrics.conversion_rate DESC LIMIT 25 PARAMETERS omit_unselected_resource_names=true\n\nNLQ: List my top 5 ad groups with the lowest cost per click last month. \nGAQL: SELECT ad_group.name, campaign.name, campaign.status, ad_group.status, metrics.average_cpc FROM ad_group WHERE campaign.status != \u0027REMOVED\u0027 AND ad_group.status != \u0027REMOVED\u0027 AND segments.date DURING LAST_MONTH ORDER BY metrics.average_cpc ASC LIMIT 5 PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Which campaigns had zero impressions in the last 7 days? \nGAQL: SELECT campaign.name, campaign.status, metrics.impressions FROM campaign WHERE campaign.status != \u0027REMOVED\u0027 AND metrics.impressions = 0 AND segments.date DURING LAST_7_DAYS ORDER BY campaign.name PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Show me the top 10 search terms that triggered my ads last month. \nGAQL: SELECT search_term_view.search_term, metrics.impressions, metrics.clicks, campaign.name, campaign.status, ad_group.name, ad_group.status FROM search_term_view WHERE campaign.status != \u0027REMOVED\u0027 AND ad_group.status != \u0027REMOVED\u0027 AND segments.date DURING LAST_MONTH ORDER BY metrics.impressions DESC LIMIT 10 PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Which ad headlines received the most clicks this week? \nGAQL: SELECT ad_group_ad.id, ad_group_ad.name, ad_group.name, campaign.name, campaign.status, ad_group.status, ad_group_ad.status, metrics.clicks FROM ad_group_ad WHERE campaign.status != \u0027REMOVED\u0027 AND ad_group.status != \u0027REMOVED\u0027 AND ad_group_ad.status != \u0027REMOVED\u0027 AND segments.date DURING THIS_WEEK_SUN_TODAY ORDER BY metrics.clicks DESC LIMIT 10 PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Find all paused ad groups in enabled campaigns. \nGAQL: SELECT ad_group.name, campaign.name, campaign.status, ad_group.status FROM ad_group WHERE campaign.status = \u0027ENABLED\u0027 AND ad_group.status = \u0027PAUSED\u0027 ORDER BY ad_group.name PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Which device type brought in the highest conversion value last month? \nGAQL: SELECT segments.device, metrics.conversions_value FROM campaign WHERE segments.date DURING LAST_MONTH AND campaign.status != \u0027REMOVED\u0027 ORDER BY metrics.conversions_value DESC LIMIT 1 PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Show me the average cost per conversion for each campaign in the last 30 days. \nGAQL: SELECT campaign.name, campaign.status, metrics.cost_per_conversion FROM campaign WHERE campaign.status != \u0027REMOVED\u0027 AND segments.date DURING LAST_30_DAYS ORDER BY metrics.cost_per_conversion DESC PARAMETERS omit_unselected_resource_names=true\n\nNLQ: Which ad groups have not received any clicks in the last 14 days? \nGAQL: SELECT ad_group.name, campaign.name, campaign.status, ad_group.status, metrics.clicks FROM ad_group WHERE campaign.status != \u0027REMOVED\u0027 AND ad_group.status != \u0027REMOVED\u0027 AND metrics.clicks = 0 AND segments.date DURING LAST_14_DAYS ORDER BY ad_group.name PARAMETERS omit_unselected_resource_names=true\n\nNLQ: What is the total cost for my account this month? \nGAQL: SELECT metrics.cost_micros FROM customer WHERE segments.date DURING THIS_MONTH ORDER BY metrics.cost_micros DESC PARAMETERS omit_unselected_resource_names=true",
"model_name": "gpt-4.1",
"query": "SELECT metrics.clicks, segments.date FROM customer WHERE segments.date DURING LAST_14_DAYS ORDER BY segments.date ASC PARAMETERS omit_unselected_resource_names=true",
"raw_input_text_generation_0": [],
"raw_output_text_generation_0": "",
"statistics_text_generation_0": {
"input_tokens": 0,
"output_tokens": 0
}
}
This subset can be loaded as:
from datasets import load_dataset
ds = load_dataset("ukav/generated_gads_synthetic63.1", "default")
Or simply as it follows, since there's only one configuration and is named default:
from datasets import load_dataset
ds = load_dataset("ukav/generated_gads_synthetic63.1")