ReViSQL: Achieving Human-Level Text-to-SQL
Paper • 2603.20004 • Published • 1
question_id int64 3 9.42k | db_id stringclasses 69
values | grading_method stringclasses 8
values | question stringlengths 25 391 | evidence stringlengths 0 788 | SQL stringlengths 31 1.2k |
|---|---|---|---|---|---|
3 | movie_platform | set | Name the titles of the movies with the most ratings. | movie with the most rating refers to MAX(COUNT(ratings)); | SELECT m.movie_title FROM movies AS m JOIN ratings AS r ON m.movie_id = r.movie_id GROUP BY m.movie_id, m.movie_title HAVING COUNT(*) = (SELECT COUNT(*) FROM movies AS m JOIN ratings AS r ON m.movie_id = r.movie_id GROUP BY m.movie_id, m.movie_title ORDER BY COUNT(*) DESC LIMIT 1); |
16 | movie_platform | set | What is the name of the most followed list? | most followed list refers to MAX(list_followers); | SELECT list_title FROM lists WHERE list_followers = (SELECT MAX(list_followers) FROM lists) |
25 | movie_platform | set | In what year was Quentin Tarantino’s third-earliest movie released? | movies.director_name records 'Quentin Tarantino'; movies.movie_release_year records the release year. | SELECT movie_release_year
FROM movies
WHERE director_name = 'Quentin Tarantino'
ORDER BY movie_release_year ASC LIMIT 2, 1 |
27 | movie_platform | set | What is the average rating score of the movie "When Will I Be Loved" and who was its director? Show the average rating score and the name of its director. | average rating score = AVG(rating_score); | SELECT AVG(T1.rating_score), T2.director_name FROM ratings AS T1 INNER JOIN movies AS T2 ON T1.movie_id = T2.movie_id WHERE T2.movie_title = 'When Will I Be Loved' |
33 | movie_platform | set | Among the lists created by user 4208563, which ones have the highest number of followers? Indicate the list ids, how many followers they have, and whether the users were a subscriber or not when they created the lists. | User 4208563 refers to user_id;highest number of followers refers to MAX(list_followers); user_subscriber = 1 means that the user was a subscriber when he created the list; user_subscriber = 0 means the user was not a subscriber when he created the list (to replace); return 1 if the user was a subscriber when the list ... | SELECT T2.list_id, T1.list_followers, T2.user_subscriber = 1 FROM lists AS T1 INNER JOIN lists_users AS T2 ON T1.user_id = T2.user_id AND T1.list_id = T2.list_id WHERE T2.user_id = 4208563 AND T1.list_followers = (SELECT MAX(T1.list_followers) FROM lists AS T1 INNER JOIN lists_users AS T2 ON T1.user_id = T2.user_id AND... |
37 | movie_platform | set | How many different users gave “The Magnificent Ambersons” a rating of 2 or lower? | movies.movie_title records 'The Magnificent Ambersons'; ratings.rating_score records the score and ratings.user_id identifies the user. | SELECT COUNT(DISTINCT r.`user_id`) AS users_with_low_rating
FROM `movies` AS m
JOIN `ratings` AS r ON m.`movie_id` = r.`movie_id`
WHERE m.`movie_title` = 'The Magnificent Ambersons'
AND r.`rating_score` <= 2 |
38 | movie_platform | set | Among users who created a list in February 2016 and were eligible for trial when they created the list, which user has the most followers in their list in February 2016? Indicate their IDs. | lists_users.list_creation_date_utc records list creation date; user_eligible_for_trial = 1 denotes trial eligibility at creation. lists.list_followers records the list’s follower count. | WITH eligible_users AS (SELECT l.user_id AS user_id,
l.list_followers AS list_followers
FROM `lists` AS l
JOIN `lists_users` AS lu
ON l.list_id = lu.list_id
AND l.user_id = lu.user_id
WHERE lu.list_creation_date_utc BETWEEN '2016-02-01' AND '2016-02-29' AND user_el... |
40 | movie_platform | set | Which movies received the most 5-star ratings among each director’s releases from 1960 through 1985? Only consider directors with at least 10 movies released during that period. List the movie titles and their 5-star rating counts, including ties. | a movie's 5-star count refers to SUM(CASE WHEN ratings.rating_score = 5 THEN 1 ELSE 0 END) over its ratings; consider only movies released during 1960 through 1985, and only directors with at least 10 such movies; list the titles and that count for every movie equal to its director's MAX, including ties | WITH qualifying_directors AS (
SELECT `director_id`
FROM `movies`
WHERE `movie_release_year` BETWEEN 1960 AND 1985
GROUP BY `director_id`
HAVING COUNT(*) >= 10
),
qualifying_movies AS (
SELECT m.movie_id, m.director_id, m.movie_title
FROM movies m
JOIN qualifying_directors qd ON m.direc... |
47 | movie_platform | set | What are the titles of the movies directed by Christopher Nolan that received the highest number of five-star ratings? | movies.director_name records the director, including 'Christopher Nolan'; ratings.rating_score records the user’s star rating. | SELECT m.movie_title FROM movies AS m LEFT JOIN ratings AS r ON m.movie_id = r.movie_id AND r.rating_score = 5 WHERE m.director_name = 'Christopher Nolan' GROUP BY m.movie_id HAVING COUNT(CASE WHEN r.rating_score = 5 THEN 1 END) = (SELECT COUNT(CASE WHEN r.rating_score = 5 THEN 1 END) cnt FROM movies AS m LEFT JOIN rat... |
50 | movie_platform | set | How many movie lists were updated more than ten years after their creation year? | lists.list_creation_timestamp_utc and lists.list_update_timestamp_utc record creation and latest-update timestamps. | SELECT COUNT(*) FROM lists WHERE SUBSTR(list_update_timestamp_utc, 1, 4) - SUBSTR(list_creation_timestamp_utc, 1, 4) > 10 |
53 | movie_platform | set | How many movie lists created from January 2, 2010 onward have more than 200 followers? | lists.list_creation_timestamp_utc records creation timestamps with a YYYY-MM-DD date portion; lists.list_followers records follower count. | SELECT COUNT(*) FROM lists WHERE list_followers > 200 AND list_creation_timestamp_utc >= '2010-01-02'; |
54 | movie_platform | set | How many movie lists were created by user 83373278 when he or she was a subscriber? | the user was a subscriber when he created the list refers to user_subscriber = 1; user 83373278 refers to user_id = 83373278; | SELECT COUNT(*) FROM lists_users WHERE user_id = 83373278 AND user_subscriber = 1 |
59 | movie_platform | set | Was the user who created the list "250 Favourite Films" a trialist when he or she created the list? 'Yes' or 'No' | the user was a trialist when he created the list refers to user_trailist = 1; 250 Favourite Films is list_title; | SELECT CASE WHEN T2.user_trialist=1 THEN 'Yes' ELSE 'No' END FROM lists AS T1 INNER JOIN lists_users AS T2 ON T1.list_id = T2.list_id AND T1.user_id = T2.user_id WHERE T1.list_title = '250 Favourite Films' |
62 | movie_platform | set | What's the description of user 85981819's movie list with the most followers? | user 85981819 refers to user_id = 85981819; most followers refers to Max(list_followers); description refers to list_descriptions; | SELECT T1.list_description FROM lists AS T1 INNER JOIN lists_users AS T2 ON T1.list_id = T2.list_id AND T1.user_id = T2.user_id WHERE T1.user_id = 85981819 AND T1.list_followers = (SELECT MAX(T3.list_followers) FROM lists AS T3 INNER JOIN lists_users AS T4 ON T3.list_id = T4.list_id AND T3.user_id = T4.user_id WHERE T3... |
64 | movie_platform | set | What's the avatar image of the user who created the movie list "250 Favourite Films"? | 250 Favourite Films refers to list_title; avatar image refers to user_avatar_image_url; | SELECT T2.user_avatar_image_url FROM lists AS T1 INNER JOIN lists_users AS T2 ON T1.list_id = T2.list_id AND T1.user_id = T2.user_id WHERE T1.list_title = '250 Favourite Films' |
70 | movie_platform | set | When did user 39115684 rate the movie "A Way of Life"? | 'A Way of Life' refers to movie_title; user 39115684 refers to userid = 39115684; when the user rate refers to rating_timestamp_utc; | SELECT r.rating_timestamp_utc
FROM ratings AS r
JOIN movies AS m ON r.movie_id = m.movie_id
WHERE m.movie_title = 'A Way of Life'
AND r.user_id = 39115684; |
73 | movie_platform | set | How many users were trialists when they rated the movie "A Way of Life"? | 'A Way of Life' refers to movie_title; the user was a trialist when he rated the movie refers to user_trialist = 1; | SELECT COUNT(DISTINCT T1.user_id) FROM ratings AS T1 INNER JOIN movies AS T2 ON T1.movie_id = T2.movie_id WHERE T2.movie_title = 'A Way of Life' AND T1.user_trialist = 1 |
76 | movie_platform | set | Which film did user 58149469 review with 1 like and 2 comments? Indicate the movie title. | ratings.critic_likes and ratings.critic_comments record likes and comments on a user’s review; movies.movie_title records film titles. | SELECT T2.movie_title FROM ratings AS T1 INNER JOIN movies AS T2 ON T1.movie_id = T2.movie_id WHERE T1.user_id = 58149469 AND T1.critic_likes = 1 AND T1.critic_comments = 2 |
78 | movie_platform | set | How many ratings on the movie "A Way of Life" are made after the year 2011? | movies.movie_title records the title 'A Way of Life'. ratings.rating_timestamp_utc records the rating timestamp. | SELECT COUNT(DISTINCT T1.rating_id) FROM ratings AS T1 INNER JOIN movies AS T2 ON T1.movie_id = T2.movie_id WHERE T2.movie_title = 'A Way of Life' AND T1.rating_timestamp_utc >= '2012-01-01' |
80 | movie_platform | set | Please list the movies rated by the user who created the movie list "250 Favourite Films". | 250 Favourite Films' is list_title; movies refers to movie_title; | SELECT T2.movie_title FROM ratings AS T1 INNER JOIN movies AS T2 ON T1.movie_id = T2.movie_id INNER JOIN lists AS T3 ON T3.user_id = T1.user_id WHERE T3.list_title = '250 Favourite Films' |
81 | movie_platform | set | What's the average rating score of the movie "A Way of Life"? | the movie "A Way of Life" refers to movies.movie_title = 'A Way of Life'; average rating score refers to AVG(ratings.rating_score) over the ratings of that movie | SELECT AVG(T1.rating_score) FROM ratings AS T1 INNER JOIN movies AS T2 ON T1.movie_id = T2.movie_id WHERE T2.movie_title = 'A Way of Life' |
88 | movie_platform | set | Which of the films directed by Paul Thomas Anderson are the most popular among Mubi users? List the movie titles. | Paul Thomas Anderson is the director name; the most popular refers to Max(movie_popularity); films refers to movies; | SELECT movie_title FROM movies WHERE director_name = 'Paul Thomas Anderson' AND movie_popularity = (SELECT MAX(movie_popularity) FROM movies WHERE director_name = 'Paul Thomas Anderson') |
90 | movie_platform | set | Please list the id of the director of the movie "It's Winter". | "It's Winter" is movie_title; | SELECT director_id FROM movies WHERE movie_title = 'It''s Winter' |
94 | movie_platform | set | What are the names of the top three movies by total critic likes in user ratings? | total critic likes of a film refers to SUM(ratings.critic_likes) over that film's ratings; return movies.movie_title for the top three by that total, ordered from the largest | SELECT m.`movie_title`
FROM `movies` m
JOIN `ratings` r ON r.`movie_id` = m.`movie_id`
GROUP BY m.`movie_id`
ORDER BY SUM(r.`critic_likes`) DESC
LIMIT 3; |
95 | movie_platform | set | How many users have more than 100 followers in the list created by users in 2009? | more than 100 followers refers to list_followers >100; list created by the user in 2009 refers to list_creation_date_utc = '2009'; | SELECT COUNT(DISTINCT T1.user_id) FROM lists_users AS T1 INNER JOIN lists AS T2 ON T1.list_id = T2.list_id WHERE T2.list_followers > 100 AND T1.list_creation_date_utc LIKE '2009%' |
100 | movie_platform | set | Please list the names of the films released in 2003 among the films scored by user 2941 . | released in 2003 refers to movie_release_year = 2003; user 2941 refers to user_id = 2941; film refers to movie; | SELECT T2.movie_title FROM ratings AS T1 INNER JOIN movies AS T2 ON T1.movie_id = T2.movie_id WHERE T2.movie_release_year = 2003 AND T1.user_id = 2941 |
106 | movie_platform | set | What is the average score for the movie Versailles Rive-Gauche? | 'Versailles Rive-Gauche' is movie_title; average score refers to Avg(rating_score); | SELECT AVG(T1.rating_score) FROM ratings AS T1 INNER JOIN movies AS T2 ON T1.movie_id = T2.movie_id WHERE T2.movie_title LIKE 'Versailles Rive-Gauche' |
119 | movie_platform | set | How many different users gave “Downfall” a rating of 4? | Downfall is recorded in movies.movie_title as 'Downfall'; ratings.rating_score records the score. | SELECT COUNT(DISTINCT T1.user_id) FROM ratings AS T1 INNER JOIN movies AS T2 ON T1.movie_id = T2.movie_id WHERE T2.movie_title = 'Downfall' AND T1.rating_score = 4 |
133 | movie_platform | set | Which 1988 movie got the most ratings? | 1988 movie refers to movies.movie_release_year = 1988; the most ratings refers to the greatest COUNT of ratings rows associated with a movie; return that movie | SELECT m.movie_title
FROM movies m
JOIN ratings r ON m.movie_id = r.movie_id
WHERE m.movie_release_year = 1988
GROUP BY m.movie_id
ORDER BY COUNT(*) DESC
LIMIT 1; |
134 | movie_platform | set | How many ratings below 3 did the most popular movie released in 1995 receive? | the most popular movie released in 1995 refers to movies.movie_popularity = MAX(movie_popularity) among the movies released in 1995; ratings below 3 refers to ratings.rating_score < 3; how many such ratings it received refers to COUNT(rating_score) over that movie's ratings | SELECT COUNT(T1.rating_score) FROM ratings AS T1 INNER JOIN movies AS T2 ON T1.movie_id = T2.movie_id WHERE T1.rating_score < 3 AND T2.movie_release_year = 1995 AND T2.movie_popularity = ( SELECT MAX(movie_popularity) FROM movies WHERE movie_release_year = 1995 ) |
144 | movie_platform | set | What is the average number of followers for lists containing at least 200 movies? | at least 200 movies refers to lists.list_movie_number >= 200; average number of followers refers to AVG(lists.list_followers) over those lists | SELECT AVG(list_followers) FROM lists WHERE list_movie_number >= 200 |
145 | movie_platform | set | Which lists contain fewer than 50 movies and were created by subscribers? Give their titles. | lists.list_movie_number is the movie count; lists_users.user_subscriber = 1 means the creator was a subscriber when creating the list. | SELECT DISTINCT T2.list_title FROM lists_users AS T1 INNER JOIN lists AS T2 ON T1.list_id = T2.list_id WHERE T2.list_movie_number < 50 AND T1.user_subscriber = 1 |
147 | movie_platform | set | Who is the user who created the list titled 'Sound and Vision'? Was he a subcriber when he created the list? 'true' or 'false' | list titled 'Sound and Vision' refers to list_title = 'Sound and Vision'; user_subscriber = 1 means the user was a subscriber when he created the list; user_subscriber = 0 means the user was not a subscriber when he created the list
| SELECT T1.user_id, CASE WHEN T1.user_subscriber=1 THEN 'true' ELSE 'false' END FROM lists_users AS T1 INNER JOIN lists AS T2 ON T1.list_id = T2.list_id AND T1.user_id = T2.user_id WHERE T2.list_title LIKE 'Sound and Vision' |
150 | movie_platform | set | List all movies rated by user 39115684. State the title, rating date and rating score. | user 39115684 refers to user_id = 39115684; title refers to movie_title; rating date refers to rating_timestamp_utc
| SELECT T2.movie_title, T1.rating_timestamp_utc, T1.rating_score FROM ratings AS T1 INNER JOIN movies AS T2 ON T1.movie_id = T2.movie_id WHERE T1.user_id = 39115684 |
157 | movie_platform | set | What is the name of the director who directed the highest number of movies in the 70s? If multiple directors tie, list all their names and the highest rating score their movies received from users. | in the 70s refers to movies.movie_release_year BETWEEN 1970 AND 1979; the highest number of movies refers to the greatest COUNT(*) of such movies per director; return every movies.director_name whose count is equal to that greatest count, and with each one the MAX(ratings.rating_score) over that director's 70s movies | WITH movies_70s AS (
SELECT `movie_id`, `director_id`, `director_name`
FROM `movies`
WHERE `movie_release_year` BETWEEN 1970 AND 1979
),
counts AS (
SELECT `director_id`, `director_name`, COUNT(*) AS num_movies
FROM movies_70s
GROUP BY `director_id`
),
max_cnt AS (
SELECT MAX(num_movies) AS max_num FROM c... |
161 | movie_platform | set | What is the average number of followers of the lists created by the user who rated the movie "Pavee Lackeen: The Traveller Girl" on 3/27/2011 at 2:06:34 AM? | average number of followers refers to AVG(list_followers); movie "Pavee Lackeen: The Traveller Girl" refers to movie_title = 'Pavee Lackeen: The Traveller Girl'; on 3/27/2011 at 2:06:34 AM refers to rating_timestamp_utc = '2011-03-27 02:06:34' | SELECT CAST(SUM(T4.list_followers) AS REAL) / COUNT(T2.list_id) FROM ratings AS T1 INNER JOIN lists_users AS T2 ON T1.user_id = T2.user_id INNER JOIN movies AS T3 ON T1.movie_id = T3.movie_id INNER JOIN lists AS T4 ON T2.list_id = T4.list_id WHERE T3.movie_title LIKE 'Pavee Lackeen: The Traveller Girl' AND T1.rating_ti... |
163 | movie_platform | set | What is the average number of movies in user 8516503’s lists, and how many movies has that user rated 5? | user 8516503 refers to the rows of that user; the average number of movies in that user's lists = AVG(lists.list_movie_number) over that user's lists; the count of movies that user rated 5 = COUNT(DISTINCT movies) over that user's ratings with rating_score = 5, counting each of those movies only one time; return both v... | SELECT
(SELECT AVG(`list_movie_number`)
FROM `lists`
WHERE `user_id` = 8516503) AS avg_movies_in_lists,
(SELECT COUNT(DISTINCT `movie_id`)
FROM `ratings`
WHERE `user_id` = 8516503
AND `rating_score` = 5) AS count_rating_5; |
166 | movie_platform | set | What are the top 10 oldest movies and what are the average rating score for each movie? Indicate the ids of the movies, the average rating scores, the name of the director, and when the movies were released. For movies with the same release year, order by title alphabetically. | the average rating score refers to AVG(T2.rating_score); oldest movies refers to MIN(movie_release_year) | SELECT
m.`movie_id`,
AVG(r.`rating_score`) AS avg_rating,
m.`director_name`,
m.`movie_release_year`
FROM `movies` AS m
LEFT JOIN `ratings` AS r
ON r.`movie_id` = m.`movie_id`
WHERE m.movie_release_year IS NOT NULL
GROUP BY m.`movie_id`, m.`director_name`, m.`movie_release_year`
ORDER BY m.`movie_... |
168 | book_publishing_company | set | What is the title with the most ordered quantity in year 1992? | total quantity refers to SUM(qty); most ordered quantity refers to order with the highest quantity where MAX(SUM(qty)); date refers to ord_date; year 1992 refers to YEAR(ord_date) = 1992 | SELECT T2.title
FROM sales AS T1
JOIN titles AS T2 ON T1.title_id = T2.title_id
WHERE STRFTIME('%Y', T1.ord_date) = '1992'
GROUP BY T1.title_id
ORDER BY SUM(T1.qty) DESC
LIMIT 1; |
169 | book_publishing_company | set | List the title, price and publication date for all sales with 'ON invoice' payment terms. | publication date refers to pubdate; payment terms refers to payterms; payterms = 'ON invoice' | SELECT T2.title, T2.price, T2.pubdate FROM sales AS T1 INNER JOIN titles AS T2 ON T1.title_id = T2.title_id WHERE T1.payterms = 'ON invoice' |
174 | book_publishing_company | set | What royalty percentages are associated with the title with the highest year-to-date sales among royalty schedules whose upper range is below 20,000? | royalty percentages refers to roysched.royalty of the schedules with hirange < 20000; the title with the highest year-to-date sales refers to ytd_sales = MAX(ytd_sales) among titles with such a schedule; return the royalty for that title | SELECT r.royalty
FROM titles t
JOIN roysched r ON t.title_id = r.title_id
WHERE r.hirange < 20000
AND t.ytd_sales = (
SELECT MAX(t1.ytd_sales)
FROM titles t1
JOIN roysched r1 ON t1.title_id = r1.title_id
WHERE r1.hirange < 20000) |
175 | book_publishing_company | set | List all titles published in year 1991. Also provide notes details of the title and the publisher's name. | publisher name refers to pub_name; publication date refers to pubdate; published in year 1991 refers to YEAR(pubdate) = 1991 | SELECT T1.title, T1.notes, T2.pub_name FROM titles AS T1 INNER JOIN publishers AS T2 ON T1.pub_id = T2.pub_id WHERE STRFTIME('%Y', T1.pubdate) = '1991' |
176 | book_publishing_company | set | List all titles with a total quantity sold of more than 20 at stores in California. | California is recorded as stores.state = 'CA'; sales.qty records quantity sold. | SELECT DISTINCT T1.title FROM titles AS T1 INNER JOIN sales AS T2 ON T1.title_id = T2.title_id INNER JOIN stores AS T3 ON T2.stor_id = T3.stor_id WHERE T3.state = 'CA' GROUP BY T1.title_id HAVING SUM(T2.qty) > 20 |
177 | book_publishing_company | set | What is the name of the store with the highest total quantity sold, and which title had the lowest quantity sold at that store? | the store with the highest total quantity sold refers to the store whose SUM(sales.qty) is the highest; at that store a title's quantity sold refers to SUM(qty) for that title; return both stores.stor_name and the titles.title with the lowest such per-title total | WITH top_store AS (
SELECT `stor_id`
FROM `sales`
GROUP BY `stor_id`
HAVING SUM(`qty`) = (
SELECT SUM(qty) FROM sales GROUP BY stor_id ORDER BY SUM(qty) DESC LIMIT 1)
),
top_store_title_quantity AS (
SELECT st.stor_name stor_name, st.stor_id stor_id, ti.title title, SUM(sa.`qty`) title_qty
FROM `sales... |
181 | book_publishing_company | set | Name all employees who were hired before year 1990. List their first and last names. | hired before year 1990 refers to YEAR(hire_date)<1990 | SELECT fname, lname FROM employee WHERE STRFTIME('%Y', hire_date) < '1990' |
182 | book_publishing_company | set | Which employees have the lowest job level? List their first names, last names, and hire dates, including ties. | the lowest job level refers to employee.job_lvl = MIN(job_lvl) over all employees, so every employee at that level qualifies, including ties; list the fname, lname and hire_date of each such employee | SELECT fname,
lname,
hire_date
FROM employee
WHERE job_lvl = (SELECT MIN(job_lvl) FROM employee); |
183 | book_publishing_company | set | In which year has the most hired employees? | most hired employees refers to MAX(count(emp_id)) | SELECT STRFTIME('%Y', hire_date) FROM employee GROUP BY STRFTIME('%Y', hire_date) ORDER BY COUNT(emp_id) DESC LIMIT 1 |
189 | book_publishing_company | set | Name the publisher which has the most titles published in 1991. | most title published refers to MAX(count(title_id)); published in 1991 refers to YEAR(pubdate) = 1991 | SELECT T2.pub_name
FROM titles AS T1
INNER JOIN publishers AS T2 ON T1.pub_id = T2.pub_id
WHERE STRFTIME('%Y', T1.pubdate) = '1991'
GROUP BY T1.pub_id, T2.pub_name
HAVING COUNT(T1.title_id) = (
SELECT COUNT(T1.title_id)
FROM titles AS T1
INNER JOIN publishers AS T2 ON T1.pub_id = T2.pub_id
WHERE STRFTIM... |
196 | book_publishing_company | set | What percentage of employees are editors or designers? | editors or designers refers to jobs.job_desc IN ('Editor', 'Designer'), reached from employee.job_id; percentage = DIVIDE(count of employees in those two jobs, COUNT(employee.job_id) over all employees joined to jobs) * 100, counting employees rather than distinct job titles | SELECT CAST(SUM(CASE WHEN T2.job_desc IN ('Editor', 'Designer') THEN 1 ELSE 0 END) AS REAL) * 100 / COUNT(T1.job_id) FROM employee AS T1 INNER JOIN jobs AS T2 ON T1.job_id = T2.job_id |
206 | book_publishing_company | set | In which country is the publisher of the book "Life Without Fear" located? | Life Without Fear is book title | SELECT T2.country FROM titles AS T1 INNER JOIN publishers AS T2 ON T1.pub_id = T2.pub_id WHERE T1.title = 'Life Without Fear' |
214 | book_publishing_company | set | Which type of book had the most pre-paid amount? | most pre-paid amount refers to MAX(advance) | SELECT type FROM titles ORDER BY advance DESC LIMIT 1 |
216 | book_publishing_company | set | Which job level is O'Rourke at? | job level refers to job_lvl | SELECT job_lvl FROM employee WHERE lname = 'O''Rourke' |
222 | book_publishing_company | set | What's on the notes for the order happened on 1994/9/14? | order happened on refers to ord_date | SELECT DISTINCT T1.notes FROM titles AS T1 INNER JOIN sales AS T2 ON T1.title_id = T2.title_id WHERE STRFTIME('%Y-%m-%d', T2.ord_date) = '1994-09-14' |
223 | book_publishing_company | set | List the type of the book for the order which was sold on 1993/5/29. | sold on refers to ord_date | SELECT DISTINCT T1.type FROM titles AS T1 INNER JOIN sales AS T2 ON T1.title_id = T2.title_id WHERE STRFTIME('%Y-%m-%d', T2.ord_date) = '1993-05-29' |
226 | book_publishing_company | set | Which city did Victoria P Ashworth work in? | SELECT T2.city FROM employee AS T1 INNER JOIN publishers AS T2 ON T1.pub_id = T2.pub_id WHERE T1.fname = 'Victoria' AND T1.minit = 'P' AND T1.lname = 'Ashworth' | |
227 | book_publishing_company | set | How many sales did the store in Remulade make? | Remulade is a city; sales in the store refers to ord_num | SELECT COUNT(DISTINCT T1.ord_num) FROM sales AS T1 INNER JOIN stores AS T2 ON T1.stor_id = T2.stor_id WHERE T2.city = 'Remulade' |
228 | book_publishing_company | set | For the quantities, what percent more did the store in Fremont sell than the store in Portland in 1993? | "qty is abbreviation for quantity; Fremont and Portland are name of city; sell in 1993 refers to YEAR(ord_date) = 1993; percentage = DIVIDE(
SUBTRACT(SUM(qty where city = 'Fremont' and year(ord_date = 1993)),
SUM(qty where city = 'Portland' and year(ord_date = 1993))), SUM(qty where city = 'Portland' and year(ord_date... | SELECT
(SUM(CASE WHEN st.city = 'Fremont' THEN sa.qty END) -
SUM(CASE WHEN st.city = 'Portland' THEN sa.qty END)) * 100.0
/ SUM(CASE WHEN st.city = 'Portland' THEN sa.qty END) AS percent_more
FROM sales AS sa
JOIN stores AS st ON sa.stor_id = st.stor_id
WHERE STRFTIME('%Y', sa.ord_date) = '1993'; |
230 | book_publishing_company | list | List the first names, middle initials, and last names of employees hired from 1990 through 1995, ordered by job level from highest to lowest. | job level refers to job_lvl; YEAR(hire_date) between 1990 and 1995; full name refers to fname, minit, and lname | SELECT fname, minit, lname FROM employee WHERE STRFTIME('%Y', hire_date) BETWEEN '1990' AND '1995' ORDER BY job_lvl DESC |
233 | book_publishing_company | list | List the title name, type, and price of the titles published by New Moon Books. Arrange the list in ascending order of price. | New Moon Books is recorded in publishers.pub_name; titles.title, titles.type, and titles.price provide the requested fields. | SELECT T1.title, T1.type, T1.price FROM titles AS T1 INNER JOIN publishers AS T2 ON T1.pub_id = T2.pub_id WHERE T2.pub_name = 'New Moon Books' ORDER BY T1.price |
239 | book_publishing_company | list | Name the top five titles that sold more than average and list them in descending order of the number of sales in California stores? | qty is abbreviation for quantity; sold more than average refers to qty > AVG(qty); california refers to state = 'CA' | WITH ca_sales AS (
SELECT s.title_id,
SUM(s.qty) AS total_qty
FROM sales s
JOIN stores st ON s.stor_id = st.stor_id
WHERE st.state = 'CA'
GROUP BY s.title_id
),
avg_ca AS (
SELECT AVG(total_qty) AS avg_qty FROM ca_sales
)
SELECT t.title
FROM ca_sales cs
JOIN avg_ca a
JOIN titles t ON ... |
240 | retail_complains | set | On which day was the most verbose complaint received? | day received refers to "Date received"; most verbose complaint refers to MAX(ser_time); | SELECT `Date received` FROM callcenterlogs WHERE ser_time = ( SELECT MAX(ser_time) FROM callcenterlogs ) |
241 | retail_complains | set | When did the earliest complaint begin on 2017/3/22? | callcenterlogs.ser_start records the complaint’s start time; "Date received" records its received date, represented as '2017-03-22' for March 22, 2017. | SELECT MIN(ser_start) AS earliest_start
FROM callcenterlogs
WHERE `Date received` = '2017-03-22'; |
245 | retail_complains | set | What is the detailed product of the complaint filed by Diesel Galloway on 2014/7/3? | detailed product refers to "sub-product"; on 2014/7/3 refers to "Date received" = '2014-07-03'; | SELECT T2.`Sub-product` FROM client AS T1 INNER JOIN events AS T2 ON T1.client_id = T2.Client_ID WHERE T1.first = 'Diesel' AND T1.last = 'Galloway' AND T2.`Date received` = '2014-07-03' |
248 | retail_complains | set | What is the full name of the client whose complaint on 2017/3/27 was received by MICHAL? | full names = first, middle, last; on 2017/3/27 refers to "Date received" = '2017-03-27'; MICHAL refers to server = 'MICHAL'; | SELECT T1.first, T1.middle, T1.last FROM client AS T1 INNER JOIN callcenterlogs AS T2 ON T1.client_id = T2.`rand client` WHERE T2.`Date received` = '2017-03-27' AND T2.server = 'MICHAL' |
257 | retail_complains | set | What was the percentage increase in complaints from New York City clients between 2016 and 2017? Round your answer to 3 decimal places. | client.city records 'New York City'. Each events row is a complaint; events."Date received" records its receipt date. Report the increase on a percentage scale. | WITH unrounded(value) AS (
SELECT 100.0 * (SUM(CASE WHEN strftime('%Y', T2.`Date received`) = '2017' THEN 1 ELSE 0 END) - SUM(CASE WHEN strftime('%Y', T2.`Date received`) = '2016' THEN 1 ELSE 0 END)) / SUM(CASE WHEN strftime('%Y', T2.`Date received`) = '2016' THEN 1 ELSE 0 END) FROM client AS T1 INNER JOIN events AS T2... |
265 | retail_complains | set | What was the review context from Jacksonville on 2017/7/22? | Jacksonville refers to city = 'Jacksonville'; on 2017/7/22 refers to Date = '2017-07-22'; | SELECT T1.Reviews FROM reviews AS T1 INNER JOIN district AS T2 ON T1.district_id = T2.district_id WHERE T2.city = 'Jacksonville' AND T1.Date = '2017-07-22' |
273 | retail_complains | set | For clients older than 65, list their first, middle, and last names, followed by their birth year, month, and day, and their email address. List the oldest clients first. | client.year, client.month, and client.day are birth-date components. | SELECT first, middle, last, year, month , day, email FROM client WHERE age > 65 ORDER BY age DESC |
274 | retail_complains | set | Which product got the most five stars, and how many? | most five stars refers to MAX(COUNT(stars = 5)); | SELECT T.Product, MAX(T.num) FROM ( SELECT Product, COUNT(Stars) AS num FROM reviews WHERE Stars = 5 GROUP BY Product ) T |
282 | retail_complains | set | List the first, middle, and last names of male clients born from 1980 through 2000 who complained through a referral. | client.year records birth year; client.sex = 'Male' denotes male clients. client.first, middle, and last record name components. events.Submitted via = 'Referral' denotes referral complaints. | SELECT DISTINCT T1.first, T1.middle, T1.last FROM client AS T1 INNER JOIN events AS T2 ON T1.client_id = T2.Client_ID WHERE T1.year BETWEEN 1980 AND 2000 AND T1.sex = 'Male' AND T2.`Submitted via` = 'Referral' |
283 | retail_complains | set | What is the medium through which most complaints are registered in Florida? | events.Submitted via records the complaint channel. client.state uses 'FL' for Florida. | SELECT E.`Submitted via`
FROM `events` AS E
JOIN `client` AS C ON E.`Client_ID` = C.`client_id`
WHERE C.`state` = 'FL'
GROUP BY E.`Submitted via`
HAVING COUNT(*) = (
SELECT COUNT(*)
FROM `events` AS E
JOIN `client` AS C ON E.`Client_ID` = C.`client_id`
WHERE C.`state` = 'FL'
GROUP BY E.`Submitted via`
ORD... |
286 | retail_complains | set | Find the number of service members who complained in Syracuse. | Service-member complaints have events.Tags = 'Servicemember'. client.city records 'Syracuse'; events.Client_ID identifies the complaining client. | SELECT COUNT(DISTINCT T1.client_id) FROM client AS T1 INNER JOIN events AS T2 ON T1.client_id = T2.Client_ID WHERE T2.Tags = 'Servicemember' AND T1.city = 'Syracuse' |
302 | retail_complains | set | How many male clients born in the year 1977 were given priority 0 in their complaints? | male refers to sex = 'Male'; | SELECT COUNT(DISTINCT T1.client_id)
FROM client AS T1
JOIN callcenterlogs AS T2 ON T1.`client_id` = T2.`rand client`
WHERE T1.sex = 'Male'
AND T1.year = 1977
AND T2.priority = 0; |
310 | retail_complains | set | How many complaints from female clients born in the year 2000 were not sent through the web? | female refers to sex = 'Female'; sent through refers to "Submitted via"; not sent through web refers to "Submitted via" ! = 'Web';
| SELECT COUNT(T2.`Submitted via`) FROM client AS T1 INNER JOIN events AS T2 ON T1.client_id = T2.Client_ID WHERE T1.sex = 'Female' AND T1.year = 2000 AND T2.`Submitted via` != 'Web' |
314 | retail_complains | set | What percentage of clients who sent their complaints by postal mail are aged 50 or older? | clients who sent complaints by postal mail refers to clients having an events row with `Submitted via` = 'Postal mail', each client counted once; aged 50 or older refers to client.age >= 50; percentage = DIVIDE(COUNT(those clients aged 50 or older), COUNT(all of those clients)) * 100 | WITH postal_clients AS (
SELECT DISTINCT c.client_id, c.age
FROM events e
JOIN client c ON c.client_id = e.Client_ID
WHERE e.`Submitted via` = 'Postal mail'
)
SELECT 100.0 * SUM(CASE WHEN age >= 50 THEN 1 ELSE 0 END) / COUNT(*) AS pct_clients_50_plus
FROM postal_clients |
315 | retail_complains | set | What is the average age of Norwalk clients? | average age = AVG(age); Norwalk refers to city = 'Norwalk'; | SELECT CAST(SUM(T1.age) AS REAL) / COUNT(T1.age) AS average FROM client AS T1 INNER JOIN district AS T2 ON T1.district_id = T2.district_id WHERE T2.city = 'Norwalk' |
318 | retail_complains | set | Which region does Noah Thompson live in? | SELECT s.Region FROM client c JOIN district d ON c.district_id = d.district_id JOIN state s ON d.state_abbrev = s.StateCode WHERE c.first = 'Noah' AND c.last = 'Thompson'; | |
321 | retail_complains | set | How long was Kendall Allen's complaint about her credit card? | how long refers to ser_time; credit card refers to Product = 'Credit Card'; | SELECT T3.ser_time FROM events AS T1 INNER JOIN client AS T2 ON T1.Client_ID = T2.client_id INNER JOIN callcenterlogs AS T3 ON T1.`Complaint ID` = T3.`Complaint ID` WHERE T2.first = 'Kendall' AND T2.last = 'Allen' AND T2.sex = 'Female' AND T1.Product = 'Credit card' |
323 | retail_complains | set | How many clients who live in New York City submitted their complaints via fax? | submitted complaints via fax refers to "Submitted via" = 'Fax'; | SELECT COUNT(DISTINCT T1.client_id) FROM client AS T1 INNER JOIN events AS T2 ON T1.client_id = T2.Client_ID WHERE T1.city = 'New York City' AND T2.`Submitted via` = 'Fax' |
324 | retail_complains | set | Among all the clients who complained about their credit cards, what is the percentage of male clients? Round your answer to 3 decimal places. | clients who complained about their credit cards refers to DISTINCT events.Client_ID where Product = 'Credit card', joined to client on client_id; male refers to client.sex = 'Male'; percentage = DIVIDE(SUM(CASE WHEN sex = 'Male' THEN 1 ELSE 0 END), COUNT(*)) * 100 over those distinct clients, rounded to 3 decimals | WITH unrounded(value) AS (
SELECT 100.0 * SUM(CASE WHEN c.sex = 'Male' THEN 1 ELSE 0 END) / COUNT(*) AS pct_male
FROM (
SELECT DISTINCT Client_ID
FROM events
WHERE Product = 'Credit card'
) cc
JOIN client c ON c.client_id = cc.Client_ID
) SELECT ROUND(value, 3) FROM unrounded; |
326 | retail_complains | set | What is the birth date of the youngest client? | birth date refers to year, month, day; youngest client refers to max(year, month, day) | SELECT year, month, day FROM client
WHERE year || '-' || month || '-' || day = (
SELECT year || '-' || month || '-' || day
FROM client
ORDER BY year DESC, month DESC, day DESC
LIMIT 1) |
330 | retail_complains | set | List the first, middle, and last names of clients who live in the Pacific division. | Client first, middle, and last names are recorded in client.first, client.middle, and client.last. The Pacific division is recorded as district.division = 'Pacific'. | SELECT T2.first, T2.middle, T2.last FROM district AS T1 INNER JOIN client AS T2 ON T1.district_id = T2.district_id WHERE T1.division = 'Pacific' |
331 | retail_complains | set | What is the social number of the person who made the most complaints? | social number refers to social; most complaints refers to max(count(event.Client_ID)) | SELECT c.social
FROM client AS c
JOIN events AS e ON c.client_id = e.Client_ID
GROUP BY c.client_id, c.social
HAVING COUNT(*) = (
SELECT MAX(cnt)
FROM (
SELECT COUNT(*) AS cnt
FROM events
GROUP BY Client_ID
)
); |
333 | retail_complains | set | What is the address of the client who made a complaint via postal mail on March 14, 2012? | address refers to address_1, address_2; via postal mail refers to Submitted via = 'Postal mail'; March 14 2012 refers to Date received = '2012-03-14' | SELECT T1.address_1, T1.address_2 FROM client AS T1 INNER JOIN events AS T2 ON T1.client_id = T2.Client_ID WHERE T2.`Date received` = '2012-03-14' AND T2.`Submitted via` = 'Postal mail' |
337 | retail_complains | set | What is the name of the state that the client with the email "skylar.ramirez@gmail.com" lives in? | SELECT s.State
FROM client c
JOIN district d ON c.district_id = d.district_id
JOIN state s ON d.state_abbrev = s.StateCode
WHERE c.email = 'skylar.ramirez@gmail.com'; | |
339 | retail_complains | set | Who is the owner of the final phone number for the complaints on server "MORIAH" on 9/11/2013? | owner refers to first, middle, last; on 9/11/2013 refers to Date received = '2013-09-11' | SELECT T1.first, T1.middle, T1.last FROM client AS T1 INNER JOIN callcenterlogs AS T2 ON T1.client_id = T2.`rand client` WHERE T2.server = 'MORIAH' AND T2.`Date received` = '2013-09-11' |
340 | retail_complains | set | What were the average complaint-handling times in minutes for teenagers, adults, and elders? Return one row with the three averages in that order. | teenagers refers to age > 13 AND age <= 19, adults to age > 19 AND age <= 65, elders to age > 65, with 'rand client' as the client of a callcenterlogs row; handling time in minutes = 60 * HH + MM + SS / 60.0 from the HH:MM:SS ser_time; per group return DIVIDE(SUM of those minutes, COUNT of its rows), teenagers then adu... | SELECT CAST(SUM(CASE WHEN T1.age > 13 AND T1.age <= 19 THEN 60 * strftime('%H', ser_time) + strftime('%M', ser_time) + strftime('%S', ser_time) / 60.0 ELSE 0 END) AS REAL) / SUM(CASE WHEN T1.age > 13 AND T1.age <= 19 THEN 1 ELSE 0 END) AS teenagerAverageMins, CAST(SUM(CASE WHEN T1.age > 19 AND T1.age <= 65 THEN 60 * st... |
346 | retail_complains | set | List the email addresses of clients whose complaint type is "PS". | SELECT DISTINCT T1.email FROM client AS T1 INNER JOIN callcenterlogs AS T2 ON T1.client_id = T2.`rand client` WHERE T2.type = 'PS' | |
353 | retail_complains | subset,=,5 | Pick 5 clients with 0 priority and write down their last name. | 0 priority refers to priority = 0; last name refers to last | SELECT T1.last FROM client AS T1 INNER JOIN callcenterlogs AS T2 ON T1.client_id = T2.`rand client` WHERE T2.priority = 0; |
354 | retail_complains | set | Write down the call id of clients whose first name start with alphabet "B". | first name start with alphabet "B" refers to first like 'B%' | SELECT T2.call_id FROM client AS T1 INNER JOIN callcenterlogs AS T2 ON T1.client_id = T2.`rand client` WHERE T1.first LIKE 'B%' |
360 | retail_complains | set | Write down the dates received for complaints sent via fax. | Fax submissions are recorded as events."Submitted via" = 'Fax'; events."Date received" records the received date. | SELECT DISTINCT `Date received` FROM events WHERE `Submitted via` = 'Fax' |
363 | retail_complains | set | What is the average age of clients who have had at least one complaint closed with relief? Round your answer to 2 decimal places. | closed with relief refers to events.`Company response to consumer` = 'Closed with relief'; average age refers to AVG(client.age) over the DISTINCT client_id values having at least one such event, each client counted once, rounded to 2 decimal places | WITH unrounded(value) AS (
SELECT AVG(age)
FROM (
SELECT DISTINCT c.client_id, c.age
FROM client AS c
JOIN events AS e ON c.client_id = e.Client_ID
WHERE e.`Company response to consumer` = 'Closed with relief'
)
) SELECT ROUND(value, 2) FROM unrounded; |
365 | retail_complains | set | Write the complaint ID, call ID, and final phone number of complaints through AVIDAN server from 1/1/2014 to 12/30/2014. | The final phone number is stored in phonefinal. Date received contains valid dates in YYYY-MM-DD format, without a time component. Include both January 1 and December 30, 2014, using Date received BETWEEN '2014-01-01' AND '2014-12-30'. | SELECT `Complaint ID`, call_id, phonefinal FROM callcenterlogs WHERE server = 'AVIDAN' AND `Date received` BETWEEN '2014-01-01' AND '2014-12-30'; |
367 | retail_complains | set | How many times did credit-card customers complain about over-limit fees in each year? List the years and corresponding complaint counts. | events.product uses 'Credit card'; events.issue uses 'Overlimit fee'. events."Date received" records the complaint’s date. | SELECT strftime('%Y', `Date received`), COUNT(`Date received`) FROM events WHERE product = 'Credit card' AND issue = 'Overlimit fee' GROUP BY strftime('%Y', `Date received`) HAVING COUNT(`Date received`) |
368 | retail_complains | set | Among clients in the Middle Atlantic, how many are female and no more than 18 years old? | district.division uses 'Middle Atlantic'; client.sex = 'Female' denotes female and client.age records age in years. | SELECT COUNT(T1.sex) FROM client AS T1 INNER JOIN district AS T2 ON T1.district_id = T2.district_id WHERE T2.division = 'Middle Atlantic' AND T1.sex = 'Female' AND T1.age <= 18 |
369 | retail_complains | set | Give me the full birthdate, email and phone number of the youngest client in Indianapolis . | full birthdate = year, month, day; youngest refers to max(year, month, day); in Indianapolis refers to city = 'Indianapolis' | SELECT T1.year, T1.month, T1.day, T1.email, T1.phone
FROM client AS T1
INNER JOIN district AS T2 ON T1.district_id = T2.district_id
WHERE T2.city = 'Indianapolis'
AND T1.year || '-' || T1.month || '-' || T1.day = (
SELECT T1.year || '-' || T1.month || '-' || T1.day
FROM client AS T1
INNER JOIN district AS T... |
371 | retail_complains | set | What is the longest server time when the call is about the issue of arbitration? | longest server time refers to max(ser_time) | SELECT MAX(T1.ser_time) FROM callcenterlogs AS T1 INNER JOIN events AS T2 ON T1.`Complaint ID` = T2.`Complaint ID` WHERE T2.issue = 'Arbitration' |
376 | retail_complains | set | Please list all first and last names of clients who live in New York city. | New York City refers to city = 'New York City' | SELECT first, last FROM client WHERE city = 'New York City' |
380 | retail_complains | set | What is the average age of clients in South Atlantic? | in South Atlantic refers to division = 'South Atlantic'; average age refers to avg(age) | SELECT AVG(T1.age) FROM client AS T1 INNER JOIN district AS T2 ON T1.district_id = T2.district_id WHERE T2.division = 'South Atlantic' |
381 | retail_complains | set | Which cities in the Midwest region have the least number of clients? | least number of clients refers to min(count(client_id)) | SELECT T2.city FROM client AS T1 INNER JOIN district AS T2 ON T1.district_id = T2.district_id INNER JOIN state AS T3 ON T2.state_abbrev = T3.StateCode WHERE T3.Region = 'Midwest' GROUP BY T2.city HAVING COUNT(T1.client_id) = ( SELECT MIN(client_count) FROM ( SELECT COUNT(T1.client_id) AS client_count FROM client AS T1 ... |
385 | retail_complains | set | Which state has the most cities? | state refers to state_abbrev; most cities refers to max(count(city)) | SELECT state_abbrev FROM district GROUP BY state_abbrev
HAVING COUNT(DISTINCT city) = (
SELECT COUNT(DISTINCT city)
FROM district
GROUP BY state_abbrev
ORDER BY COUNT(DISTINCT city) DESC
LIMIT 1) |
End of preview. Expand in Data Studio
BIRD-Diamond
BIRD-Diamond is a reviewed derivative of BIRD-Platinum, the ReViSQL authors' verified text-to-SQL training set built from BIRD. It has 2,442 active questions (2,079 train, 363 validation) over 69 SQLite databases. We edited 950 of them and excluded 20 more. It is not an official BIRD release, and we don't claim it is better than its source in every respect. Every change is listed, so you can inspect or undo it.
These are the exact rows our BIRD-Diamond training runs used. The train/validation partition is our own: 635 questions sit on a different side than in the ReViSQL authors' split, so validation scores here are not comparable with scores on their validation split.
What changed
Against the BIRD-Platinum baseline (the authors' verified release, 2,462 questions before exclusions):
| Field | Questions changed |
|---|---|
question |
799 |
evidence |
903 |
SQL |
148 |
grading_method |
2 |
Most edits make the external knowledge (evidence) precise or correct: spelling out an aggregation the original left
ambiguous, stating how the schema encodes a concept, or fixing a wrong mapping. Others clarify ambiguous questions or
fix gold queries that return the wrong result. Some choose an intended interpretation rather than fix a demonstrated
error. changes/changes.jsonl lists every changed field with
its before and after values, and changes/excluded.jsonl lists the 20 questions we removed because they
were underspecified or duplicated.
30 of the 69 databases (11.9 GB of 27.4 GB) differ from BIRD's originals. The repairs restore declared
foreign-key relationships, normalize some dates and titles, and restore AdventureWorks Customer rows missing from the
works_cycles database. The recipes are in databases/repairs/.
Layout
data/{train,validation}.jsonl:question_id(the BIRD question id),db_id,question,evidence,SQLandgrading_methodschemas/<db_id>.sql: the schema text shown to the model in our promptsoracle-schemas/<db_id>.sql: constraint DDL for equivalence checkingdatabases/<db_id>.sqlite: the 69 databases (27.4 GB), with SHA-256 hashes indatabases/manifest.jsondatabases/repairs/: the database repair recipes and the restored rowschanges/: the change set against BIRD-Platinumgrading-policy.json: numeric comparison rules (exact typed equality; no automatic 1% tolerance)
Usage
from datasets import load_dataset
diamond = load_dataset("scottfitz/BIRD-Diamond")
The databases are large. Download only the ones you need:
hf download scottfitz/BIRD-Diamond --repo-type dataset --include "databases/movie_platform.sqlite" --local-dir bird-diamond
Check each file against its hash in databases/manifest.json before you use it.
Limitations
- Some edits choose one reasonable interpretation of an ambiguous question; they are not the only correct reading.
- Grading assumes the rules in
grading-policy.json. Scores computed with a different grader, such as one with a 1% numeric tolerance, are not comparable.
License and attribution
CC BY-SA 4.0. BIRD-Diamond is derived from BIRD-Platinum (Zhu et al., 2026), which is derived from BIRD (Li et al.,
2023); both are licensed CC BY-SA 4.0. The changes are listed in changes/ and databases/repairs/. The restored
AdventureWorks Customer rows come from Microsoft's sql-server-samples under the MIT License; see NOTICE.
Citation
@article{zhu2026revisql,
title = {Human-Level Text-to-SQL via Reinforcement Learning on Verified Data, Without Pipeline Engineering},
author = {Zhu, Yuxuan and Jin, Tengjun and Choi, Yoojin and Kang, Daniel},
journal = {arXiv preprint arXiv:2603.20004},
year = {2026}
}
@inproceedings{li2023bird,
title = {Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLs},
author = {Li, Jinyang and others},
booktitle = {Advances in Neural Information Processing Systems},
year = {2023}
}
- Downloads last month
- 399