Datasets:

mariem123kfg
/

sob-ft-finetune-ready

record_id stringlengths 9 64	context stringlengths 2 945k ⌀	question stringlengths 1 24.6k ⌀	json_schema stringlengths 36 19.2k	validated_output stringlengths 13 139k	source_split stringclasses 1 value	source_dataset stringclasses 13 values	question_type stringclasses 15 values	schema_complexity stringclasses 3 values	source_id stringlengths 6 24 ⌀	errored_json stringlengths 13 139k	error_difficulty stringclasses 1 value	error_type stringclasses 1 value	error_types stringclasses 1 value	error_location stringclasses 0 values	error_locations stringclasses 1 value	error_description stringclasses 1 value	error_descriptions stringclasses 1 value	error_difficulties stringclasses 1 value	error_source stringclasses 1 value	errors_json stringclasses 1 value	finetuning_task stringclasses 1 value	finetuning_question stringlengths 1 24.6k	finetuning_input stringlengths 189 946k	finetuning_target stringlengths 13 139k	finetuning_messages stringlengths 577 961k	repair_candidate stringlengths 13 139k	true_has_error bool 1 class
ft_extract_00052929	/web.uri.edu/library/) * [Internships](https://web.uri.edu/career/students/) * * * * [Facebook](https://www.facebook.com/universityofri "Facebook") * [Instagram](https://www.instagram.com/universityofri/ "Instagram") * [X](https://twitter.com/universityofri "X") * [YouTube](https://www.youtube.com/user/UniversityOfRI "...	null	{"properties":{"name":{"title":"Name","type":"string"},"product_type":{"title":"Product Type","type":"string"},"duration":{"additionalProperties":{"type":"string"},"title":"Duration","type":"object"},"time_commitent":{"additionalProperties":{"type":"string"},"title":"Time Commitent","type":"object"},"track_type":{"titl...	{"name": "Healthcare Management Graduate Certificate", "product_type": "Graduate Certificate", "duration": {}, "time_commitent": {}, "track_type": "", "credit_time": {}, "notes": "Content provided is site footer; no detailed product data available."}	train	scrapegraphai_finetuning_subset	null	easy	null	{"name": "Healthcare Management Graduate Certificate", "product_type": "Graduate Certificate", "duration": {}, "time_commitent": {}, "track_type": "", "credit_time": {}, "notes": "Content provided is site footer; no detailed product data available."}	none	none	[]	null	[]	No error — candidate equals gold.	[]	[]	clean	[]	extract	Extract structured data for ProductDetailPage from the context.	Schema: {"properties":{"name":{"title":"Name","type":"string"},"product_type":{"title":"Product Type","type":"string"},"duration":{"additionalProperties":{"type":"string"},"title":"Duration","type":"object"},"time_commitent":{"additionalProperties":{"type":"string"},"title":"Time Commitent","type":"object"},"track_type...	{"name": "Healthcare Management Graduate Certificate", "product_type": "Graduate Certificate", "duration": {}, "time_commitent": {}, "track_type": "", "credit_time": {}, "notes": "Content provided is site footer; no detailed product data available."}	[{"role": "system", "content": "You extract structured JSON from text. Given a SCHEMA, CONTEXT, and QUESTION, return a single valid JSON object that matches the schema and is grounded in the context. No markdown, no code fences."}, {"role": "user", "content": "Schema:\n{\"properties\":{\"name\":{\"title\":\"Name\",\"ty...	{"name": "Healthcare Management Graduate Certificate", "product_type": "Graduate Certificate", "duration": {}, "time_commitent": {}, "track_type": "", "credit_time": {}, "notes": "Content provided is site footer; no detailed product data available."}	false
ft_extract_00038787	[![Lingnan University Logo](https://www.ln.edu.hk/assets/shared/img/logo.png)](https://www.ln.edu.hk/) [ ![logo](/sgs/assets/img/deptlogo.png) ](https://www.ln.edu.hk/sgs/) * Information for * [Prospective Students](https://www.ln.edu.hk/information-for/prospective-students) * [Current Students](https://www.ln.edu.hk/i...	Extract the programme information for international student. IMPORTANT-1: Please only output the value of each field, and do not output type or description or any other properties of the field. IMPORTANT-2: Output 'N/A' instead of 'No content available' (or anything else) if the specific information is not found for t...	{"properties":{"code":{"title":"Code","type":"string"},"concentration_information":{"title":"Concentration Information","type":"string"},"gpa_requirement":{"title":"Gpa Requirement","type":"string"},"language_test_requirement":{"title":"Language Test Requirement","type":"string"},"working_experience":{"title":"Working ...	{"code":"N/A","concentration_information":"N/A","gpa_requirement":"N/A","language_test_requirement":"N/A","working_experience":"At least 10 years of work experience, with no less than 5 years of senior level experience.","writing_requirement":"N/A"}	train	scrapegraphai_full	null	easy	null	{"code":"N/A","concentration_information":"N/A","gpa_requirement":"N/A","language_test_requirement":"N/A","working_experience":"At least 10 years of work experience, with no less than 5 years of senior level experience.","writing_requirement":"N/A"}	none	none	[]	null	[]	No error — candidate equals gold.	[]	[]	clean	[]	extract	Extract the programme information for international student. IMPORTANT-1: Please only output the value of each field, and do not output type or description or any other properties of the field. IMPORTANT-2: Output 'N/A' instead of 'No content available' (or anything else) if the specific information is not found for t...	Schema: {"properties":{"code":{"title":"Code","type":"string"},"concentration_information":{"title":"Concentration Information","type":"string"},"gpa_requirement":{"title":"Gpa Requirement","type":"string"},"language_test_requirement":{"title":"Language Test Requirement","type":"string"},"working_experience":{"title":"...	{"code":"N/A","concentration_information":"N/A","gpa_requirement":"N/A","language_test_requirement":"N/A","working_experience":"At least 10 years of work experience, with no less than 5 years of senior level experience.","writing_requirement":"N/A"}	[{"role": "system", "content": "You extract structured JSON from text. Given a SCHEMA, CONTEXT, and QUESTION, return a single valid JSON object that matches the schema and is grounded in the context. No markdown, no code fences."}, {"role": "user", "content": "Schema:\n{\"properties\":{\"code\":{\"title\":\"Code\",\"ty...	{"code":"N/A","concentration_information":"N/A","gpa_requirement":"N/A","language_test_requirement":"N/A","working_experience":"At least 10 years of work experience, with no less than 5 years of senior level experience.","writing_requirement":"N/A"}	false
3e049b8f7a10ab4bd1d6e620f6015357ab220b39aa97defc460aa2408a1a3e1b	Division Films: Division Films is a Los Angeles-based film finance and production company established by Wicks Walker in 2012. The company's first film was "Spring Breakers" starring James Franco, Selena Gomez, Vanessa Hudgens and Ashley Benson. Walker and another producer greenlit the production after Benoît Debie w...	The first film released by production company Division Films was located where?	{"type": "object", "properties": {"production_company_name": {"type": "string", "description": "The name of the film production company."}, "first_film_title": {"type": "string", "description": "The title of the first film released by the production company."}, "film_setting_location": {"type": "string", "description":...	{"production_company_name": "Division Films", "first_film_title": "Spring Breakers", "film_setting_location": "Florida"}	train	hotpotqa	bridge	medium	5ac433ac5542991943173943	{"production_company_name": "Division Films", "first_film_title": "Spring Breakers", "film_setting_location": "Florida"}	none	none	[]	null	[]	No error — candidate equals gold.	[]	[]	clean	[]	extract	The first film released by production company Division Films was located where?	Schema: {"type": "object", "properties": {"production_company_name": {"type": "string", "description": "The name of the film production company."}, "first_film_title": {"type": "string", "description": "The title of the first film released by the production company."}, "film_setting_location": {"type": "string", "descr...	{"production_company_name": "Division Films", "first_film_title": "Spring Breakers", "film_setting_location": "Florida"}	[{"role": "system", "content": "You extract structured JSON from text. Given a SCHEMA, CONTEXT, and QUESTION, return a single valid JSON object that matches the schema and is grounded in the context. No markdown, no code fences."}, {"role": "user", "content": "Schema:\n{\"type\": \"object\", \"properties\": {\"producti...	{"production_company_name": "Division Films", "first_film_title": "Spring Breakers", "film_setting_location": "Florida"}	false
ft_extract_00024373	"Skip to main content [ ![Sage logo](https://uk.sagepub.com/sites/all/themes/sage_corp/logo.svg?83e7(...TRUNCATED)	Essentials of Organizational Behavior 9781071840979 https://us.sagepub.com/	"{\"$defs\":{\"ScrapedChapter\":{\"description\":\"Schema of a chapter present in the table of conte(...TRUNCATED)	"{\"certainty_percentage\":100,\"chapters\":[{\"title\":\"What Is Organizational Behavior?\",\"prefi(...TRUNCATED)	train	scrapegraphai_full	null	medium	null	"{\"certainty_percentage\":100,\"chapters\":[{\"title\":\"What Is Organizational Behavior?\",\"prefi(...TRUNCATED)	none	none	[]	null	[]	No error — candidate equals gold.	[]	[]	clean	[]	extract	Essentials of Organizational Behavior 9781071840979 https://us.sagepub.com/	"Schema:\n{\"$defs\":{\"ScrapedChapter\":{\"description\":\"Schema of a chapter present in the table(...TRUNCATED)	"{\"certainty_percentage\":100,\"chapters\":[{\"title\":\"What Is Organizational Behavior?\",\"prefi(...TRUNCATED)	"[{\"role\": \"system\", \"content\": \"You extract structured JSON from text. Given a SCHEMA, CONTE(...TRUNCATED)	"{\"certainty_percentage\":100,\"chapters\":[{\"title\":\"What Is Organizational Behavior?\",\"prefi(...TRUNCATED)	false
ft_extract_00029544	"[Skip to main content](/books/mono/10.4324/9780203836545/language-society-power-jean-stilwell-pecce(...TRUNCATED)	Language, Society and Power N/A https://www.taylorfrancis.com/	"{\"$defs\":{\"ScrapedChapter\":{\"description\":\"Schema of a chapter present in the table of conte(...TRUNCATED)	"{\"certainty_percentage\":100,\"chapters\":[{\"title\":\"What is language?\",\"prefix\":\"chapter 1(...TRUNCATED)	train	scrapegraphai_full	null	medium	null	"{\"certainty_percentage\":100,\"chapters\":[{\"title\":\"What is language?\",\"prefix\":\"chapter 1(...TRUNCATED)	none	none	[]	null	[]	No error — candidate equals gold.	[]	[]	clean	[]	extract	Language, Society and Power N/A https://www.taylorfrancis.com/	"Schema:\n{\"$defs\":{\"ScrapedChapter\":{\"description\":\"Schema of a chapter present in the table(...TRUNCATED)	"{\"certainty_percentage\":100,\"chapters\":[{\"title\":\"What is language?\",\"prefix\":\"chapter 1(...TRUNCATED)	"[{\"role\": \"system\", \"content\": \"You extract structured JSON from text. Given a SCHEMA, CONTE(...TRUNCATED)	"{\"certainty_percentage\":100,\"chapters\":[{\"title\":\"What is language?\",\"prefix\":\"chapter 1(...TRUNCATED)	false
json-training-10893	null	Provide information about NATO member countries, including their name, location, and population.	"{\"type\": \"object\", \"properties\": {\"nato_countries\": {\"type\": \"array\", \"items\": {\"typ(...TRUNCATED)	"{\n \"nato_countries\": [\n {\"name\": \"United States\", \"location\": \"North America\", \"po(...TRUNCATED)	train	ChristianAzinn/json-training	synthetic_ungrounded	hard	null	"{\n \"nato_countries\": [\n {\"name\": \"United States\", \"location\": \"North America\", \"po(...TRUNCATED)	none	none	[]	null	[]	No error — candidate equals gold.	[]	[]	clean	[]	extract	Provide information about NATO member countries, including their name, location, and population.	"Schema:\n{\"type\": \"object\", \"properties\": {\"nato_countries\": {\"type\": \"array\", \"items\(...TRUNCATED)	"{\n \"nato_countries\": [\n {\"name\": \"United States\", \"location\": \"North America\", \"po(...TRUNCATED)	"[{\"role\": \"system\", \"content\": \"You extract structured JSON from text. Given a SCHEMA, CONTE(...TRUNCATED)	"{\n \"nato_countries\": [\n {\"name\": \"United States\", \"location\": \"North America\", \"po(...TRUNCATED)	false
stage-eval-data42281	"The following analysis examines the spatial distribution of six distinct locations, each characteri(...TRUNCATED)	"Could you please extract the six survey point locations (AAA through FFF) and their latitude and lo(...TRUNCATED)	"{\"$schema\": \"https://json-schema.org/draft/2020-12/schema\", \"type\": \"object\", \"properties\(...TRUNCATED)	"{\n \"locations\": [\n {\n \"latitude\": 54.1778,\n \"longitude\": -3.18889,\n \(...TRUNCATED)	train	boradorish/STAGE-eval	train	hard	data42281	"{\n \"locations\": [\n {\n \"latitude\": 54.1778,\n \"longitude\": -3.18889,\n \(...TRUNCATED)	none	none	[]	null	[]	No error — candidate equals gold.	[]	[]	clean	[]	extract	"Could you please extract the six survey point locations (AAA through FFF) and their latitude and lo(...TRUNCATED)	"Schema:\n{\"$schema\": \"https://json-schema.org/draft/2020-12/schema\", \"type\": \"object\", \"pr(...TRUNCATED)	"{\n \"locations\": [\n {\n \"latitude\": 54.1778,\n \"longitude\": -3.18889,\n \(...TRUNCATED)	"[{\"role\": \"system\", \"content\": \"You extract structured JSON from text. Given a SCHEMA, CONTE(...TRUNCATED)	"{\n \"locations\": [\n {\n \"latitude\": 54.1778,\n \"longitude\": -3.18889,\n \(...TRUNCATED)	false
stage-eval-data44571	"A review of participant timings reveals a steady progression in completion times across a series of(...TRUNCATED)	"Could you extract the participant completion times from Sheet1, including their names and any refer(...TRUNCATED)	"{\"$schema\": \"https://json-schema.org/draft/2020-12/schema\", \"type\": \"object\", \"properties\(...TRUNCATED)	"{\n \"data\": [\n {\n \"Name\": \"[Person_1]\",\n \"Time\": 9.42,\n \"Unnamed: 3(...TRUNCATED)	train	boradorish/STAGE-eval	train	hard	data44571	"{\n \"data\": [\n {\n \"Name\": \"[Person_1]\",\n \"Time\": 9.42,\n \"Unnamed: 3(...TRUNCATED)	none	none	[]	null	[]	No error — candidate equals gold.	[]	[]	clean	[]	extract	"Could you extract the participant completion times from Sheet1, including their names and any refer(...TRUNCATED)	"Schema:\n{\"$schema\": \"https://json-schema.org/draft/2020-12/schema\", \"type\": \"object\", \"pr(...TRUNCATED)	"{\n \"data\": [\n {\n \"Name\": \"[Person_1]\",\n \"Time\": 9.42,\n \"Unnamed: 3(...TRUNCATED)	"[{\"role\": \"system\", \"content\": \"You extract structured JSON from text. Given a SCHEMA, CONTE(...TRUNCATED)	"{\n \"data\": [\n {\n \"Name\": \"[Person_1]\",\n \"Time\": 9.42,\n \"Unnamed: 3(...TRUNCATED)	false
json-training-14181	null	"Provide information about psychoanalytic theories of anxiety and their applications in modern psych(...TRUNCATED)	"{\"type\": \"object\", \"properties\": {\"theories\": {\"type\": \"array\", \"items\": {\"type\": \(...TRUNCATED)	"{\n \"theories\": [\n {\n \"theory_name\": \"Id Theory\",\n \"description\": \"The Id(...TRUNCATED)	train	ChristianAzinn/json-training	synthetic_ungrounded	hard	null	"{\n \"theories\": [\n {\n \"theory_name\": \"Id Theory\",\n \"description\": \"The Id(...TRUNCATED)	none	none	[]	null	[]	No error — candidate equals gold.	[]	[]	clean	[]	extract	"Provide information about psychoanalytic theories of anxiety and their applications in modern psych(...TRUNCATED)	"Schema:\n{\"type\": \"object\", \"properties\": {\"theories\": {\"type\": \"array\", \"items\": {\"(...TRUNCATED)	"{\n \"theories\": [\n {\n \"theory_name\": \"Id Theory\",\n \"description\": \"The Id(...TRUNCATED)	"[{\"role\": \"system\", \"content\": \"You extract structured JSON from text. Given a SCHEMA, CONTE(...TRUNCATED)	"{\n \"theories\": [\n {\n \"theory_name\": \"Id Theory\",\n \"description\": \"The Id(...TRUNCATED)	false
ft_extract_00060220	"![](//fls-fe.amazon.co.jp/1/batch/1/OP/A1VC38T7YXB528:356-4715449-8637051:075G884KCMA1Z9F083F3$ueda(...TRUNCATED)	null	"{\"$defs\":{\"MainCategory\":{\"enum\":[\"Home & Pets\",\"Sports & Outdoors\",\"Health & Beauty\",\(...TRUNCATED)	"{\"name\": \"\\u30d6\\u30eb\\u30dc\\u30f3 \\u30d7\\u30c1\\u30b3\\u30b3\\u30a2 50g\\u00d71\\u888b\",(...TRUNCATED)	train	scrapegraphai_finetuning_subset	null	easy	null	"{\"name\": \"\\u30d6\\u30eb\\u30dc\\u30f3 \\u30d7\\u30c1\\u30b3\\u30b3\\u30a2 50g\\u00d71\\u888b\",(...TRUNCATED)	none	none	[]	null	[]	No error — candidate equals gold.	[]	[]	clean	[]	extract	Extract structured data for Item from the context.	"Schema:\n{\"$defs\":{\"MainCategory\":{\"enum\":[\"Home & Pets\",\"Sports & Outdoors\",\"Health & B(...TRUNCATED)	"{\"name\": \"\\u30d6\\u30eb\\u30dc\\u30f3 \\u30d7\\u30c1\\u30b3\\u30b3\\u30a2 50g\\u00d71\\u888b\",(...TRUNCATED)	"[{\"role\": \"system\", \"content\": \"You extract structured JSON from text. Given a SCHEMA, CONTE(...TRUNCATED)	"{\"name\": \"\\u30d6\\u30eb\\u30dc\\u30f3 \\u30d7\\u30c1\\u30b3\\u30b3\\u30a2 50g\\u00d71\\u888b\",(...TRUNCATED)	false

End of preview. Expand in Data Studio

SOB Finetune Ready

Cleaned, sharded mirror of seneca-center/sob-ft-finetune-ready for Dataset Viewer browsing.

Splits

Split	Rows	Description
`extract`	54,697	JSON extraction SFT rows (`finetuning_task=extract`)
`repair`	43,360	JSON repair SFT rows (`finetuning_task=repair`)

Total source rows: 98,057 (extract + repair).

from datasets import Dataset

# Viewer-friendly load via parquet paths (Hub README schema may lag full columns)
import pandas as pd
from huggingface_hub import hf_hub_download

path = hf_hub_download("mariem123kfg/sob-ft-finetune-ready", "data/extract-00000-of-00002.parquet", repo_type="dataset")
pd.read_parquet(path).head()

Cleaning

Extract prompts: <think> block removed.
Rows dropped when validated_output is plain text (not JSON), context English share is <= 50%, or both context and question are empty.
Blocked record: stage-eval-data9343.
Restored from mariem123kfg/sob-ft-extract: ToolACE + Hermes func_calling rows whose output converts to valid JSON (extract task prompts filled).

Key columns

Column	Description
`finetuning_task`	`extract` or `repair`
`finetuning_messages`	Chat JSON (system / user / assistant)
`finetuning_input`	User turn text
`finetuning_target`	Assistant target
`context`, `question`, `json_schema`	Raw SOB inputs
`validated_output`, `errored_json`	Gold JSON and corrupted candidate
`repair_candidate`, `true_has_error`	Repair task fields

Downloads last month: 161