Datasets:

massaindustries
/

data-farm-tester-labeling

Error code:   DatasetGenerationError
Exception:    ValueError
Message:      Expected object or value
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1835, in _prepare_split_single
                  for key, table in generator:
                                    ^^^^^^^^^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 621, in wrapped
                  for item in generator(*args, **kwargs):
                              ~~~~~~~~~^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 281, in _generate_tables
                  examples = [ujson_loads(line) for line in batch.splitlines()]
                              ~~~~~~~~~~~^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 20, in ujson_loads
                  return pd.io.json.ujson_loads(*args, **kwargs)
                         ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
              ValueError: Expected object or value
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1379, in compute_config_parquet_and_info_response
                  parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
                                                                        ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      builder, max_dataset_size_bytes=max_dataset_size_bytes
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 956, in stream_convert_to_parquet
                  builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
                  ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1702, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ~~~~~~~~~~~~~~~~~~~~~~~~~~^
                      gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  ):
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1887, in _prepare_split_single
                  raise DatasetGenerationError("An error occurred while generating the dataset") from e
              datasets.exceptions.DatasetGenerationError: An error occurred while generating the dataset

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

family string	subfamily string	system string	user string	input string	gold unknown	metric string	source string	source_id string	task_id string	similarity float64
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: ( e ) ventually , every idea in this film is flushed down the latrine of heroism . Return only the label.	( e ) ventually , every idea in this film is flushed down the latrine of heroism .	"negative"	accuracy	sst2	sst2-487	label-sentiment-0000	0.5592
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: or emptying rat traps . Return only the label.	or emptying rat traps .	"negative"	accuracy	sst2	sst2-635	label-sentiment-0001	0.6418
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: at its worst , it implodes in a series of very bad special effects . Return only the label.	at its worst , it implodes in a series of very bad special effects .	"negative"	accuracy	sst2	sst2-380	label-sentiment-0002	0.6325
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: mr. tsai is a very original artist in his medium , and what time is it there ? Return only the label.	mr. tsai is a very original artist in his medium , and what time is it there ?	"positive"	accuracy	sst2	sst2-52	label-sentiment-0003	0.6046
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: it 's of the quality of a lesser harrison ford movie - six days , seven nights , maybe , or that dreadful sabrina remake . Return only the label.	it 's of the quality of a lesser harrison ford movie - six days , seven nights , maybe , or that dreadful sabrina remake .	"negative"	accuracy	sst2	sst2-338	label-sentiment-0004	0.5803
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: it is amusing , and that 's all it needs to be . Return only the label.	it is amusing , and that 's all it needs to be .	"positive"	accuracy	sst2	sst2-623	label-sentiment-0005	0.6654
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: given how heavy-handed and portent-heavy it is , this could be the worst thing soderbergh has ever done . Return only the label.	given how heavy-handed and portent-heavy it is , this could be the worst thing soderbergh has ever done .	"negative"	accuracy	sst2	sst2-844	label-sentiment-0006	0.6096
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: intriguing documentary which is emotionally diluted by focusing on the story 's least interesting subject . Return only the label.	intriguing documentary which is emotionally diluted by focusing on the story 's least interesting subject .	"positive"	accuracy	sst2	sst2-464	label-sentiment-0007	0.6733
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: there are simply too many ideas floating around -- part farce , part sliding doors , part pop video -- and yet failing to exploit them . Return only the label.	there are simply too many ideas floating around -- part farce , part sliding doors , part pop video -- and yet failing to exploit them .	"negative"	accuracy	sst2	sst2-408	label-sentiment-0008	0.6392
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: a strangely compelling and brilliantly acted psychological drama . Return only the label.	a strangely compelling and brilliantly acted psychological drama .	"positive"	accuracy	sst2	sst2-817	label-sentiment-0009	0.623
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: i 'm just too bored to care . Return only the label.	i 'm just too bored to care .	"negative"	accuracy	sst2	sst2-419	label-sentiment-0010	0.7103
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: a movie with a real anarchic flair . Return only the label.	a movie with a real anarchic flair .	"positive"	accuracy	sst2	sst2-868	label-sentiment-0011	0.6825
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: if you dig on david mamet 's mind tricks ... rent this movie and enjoy ! Return only the label.	if you dig on david mamet 's mind tricks ... rent this movie and enjoy !	"positive"	accuracy	sst2	sst2-174	label-sentiment-0012	0.6463
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: liotta put on 30 pounds for the role , and has completely transformed himself from his smooth , goodfellas image . Return only the label.	liotta put on 30 pounds for the role , and has completely transformed himself from his smooth , goodfellas image .	"positive"	accuracy	sst2	sst2-325	label-sentiment-0013	0.5746
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: another in-your-face wallow in the lower depths made by people who have never sung those blues . Return only the label.	another in-your-face wallow in the lower depths made by people who have never sung those blues .	"negative"	accuracy	sst2	sst2-395	label-sentiment-0014	0.6224
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: the jabs it employs are short , carefully placed and dead-center . Return only the label.	the jabs it employs are short , carefully placed and dead-center .	"positive"	accuracy	sst2	sst2-643	label-sentiment-0015	0.6358
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: it 's a grab bag of genres that do n't add up to a whole lot of sense . Return only the label.	it 's a grab bag of genres that do n't add up to a whole lot of sense .	"negative"	accuracy	sst2	sst2-363	label-sentiment-0016	0.681
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: maud and roland 's search for an unknowable past makes for a haunting literary detective story , but labute pulls off a neater trick in possession : he makes language sexy . Return only the label.	maud and roland 's search for an unknowable past makes for a haunting literary detective story , but labute pulls off a neater trick in possession : he makes language sexy .	"positive"	accuracy	sst2	sst2-406	label-sentiment-0017	0.5889
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: filmmakers who can deftly change moods are treasures and even marvels . Return only the label.	filmmakers who can deftly change moods are treasures and even marvels .	"positive"	accuracy	sst2	sst2-679	label-sentiment-0018	0.7173
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: although huppert 's intensity and focus has a raw exhilaration about it , the piano teacher is anything but fun . Return only the label.	although huppert 's intensity and focus has a raw exhilaration about it , the piano teacher is anything but fun .	"negative"	accuracy	sst2	sst2-544	label-sentiment-0019	0.6303
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: this is not the undisputed worst boxing movie ever , but it 's certainly not a champion - the big loser is the audience . Return only the label.	this is not the undisputed worst boxing movie ever , but it 's certainly not a champion - the big loser is the audience .	"negative"	accuracy	sst2	sst2-209	label-sentiment-0020	0.6024
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: puts a human face on a land most westerners are unfamiliar with . Return only the label.	puts a human face on a land most westerners are unfamiliar with .	"positive"	accuracy	sst2	sst2-605	label-sentiment-0021	0.6827
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: an occasionally funny , but overall limp , fish-out-of-water story . Return only the label.	an occasionally funny , but overall limp , fish-out-of-water story .	"negative"	accuracy	sst2	sst2-238	label-sentiment-0022	0.6529
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: here 's yet another studio horror franchise mucking up its storyline with glitches casual fans could correct in their sleep . Return only the label.	here 's yet another studio horror franchise mucking up its storyline with glitches casual fans could correct in their sleep .	"negative"	accuracy	sst2	sst2-552	label-sentiment-0023	0.6242
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: it 's a demented kitsch mess ( although the smeary digital video does match the muddled narrative ) , but it 's savvy about celebrity and has more guts and energy than much of what will open this year . Return only the label.	it 's a demented kitsch mess ( although the smeary digital video does match the muddled narrative ) , but it 's savvy about celebrity and has more guts and energy than much of what will open this year .	"positive"	accuracy	sst2	sst2-742	label-sentiment-0024	0.5606
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: in its best moments , resembles a bad high school production of grease , without benefit of song . Return only the label.	in its best moments , resembles a bad high school production of grease , without benefit of song .	"negative"	accuracy	sst2	sst2-19	label-sentiment-0025	0.5789
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: the vitality of the actors keeps the intensity of the film high , even as the strafings blend together . Return only the label.	the vitality of the actors keeps the intensity of the film high , even as the strafings blend together .	"positive"	accuracy	sst2	sst2-680	label-sentiment-0026	0.665
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: `` mostly martha '' is a bright , light modern day family parable that wears its heart on its sleeve for all to see . Return only the label.	`` mostly martha '' is a bright , light modern day family parable that wears its heart on its sleeve for all to see .	"positive"	accuracy	sst2	sst2-533	label-sentiment-0027	0.6021
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: by getting myself wrapped up in the visuals and eccentricities of many of the characters , i found myself confused when it came time to get to the heart of the movie . Return only the label.	by getting myself wrapped up in the visuals and eccentricities of many of the characters , i found myself confused when it came time to get to the heart of the movie .	"negative"	accuracy	sst2	sst2-158	label-sentiment-0028	0.7029
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: delivers the same old same old , tarted up with latin flava and turned out by hollywood playas . Return only the label.	delivers the same old same old , tarted up with latin flava and turned out by hollywood playas .	"negative"	accuracy	sst2	sst2-176	label-sentiment-0029	0.6253
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: hardly a masterpiece , but it introduces viewers to a good charitable enterprise and some interesting real people . Return only the label.	hardly a masterpiece , but it introduces viewers to a good charitable enterprise and some interesting real people .	"positive"	accuracy	sst2	sst2-91	label-sentiment-0030	0.6854
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: what is 100 % missing here is a script of even the most elemental literacy , an inkling of genuine wit , and anything resembling acting . Return only the label.	what is 100 % missing here is a script of even the most elemental literacy , an inkling of genuine wit , and anything resembling acting .	"negative"	accuracy	sst2	sst2-840	label-sentiment-0031	0.6285
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: ... a boring parade of talking heads and technical gibberish that will do little to advance the linux cause . Return only the label.	... a boring parade of talking heads and technical gibberish that will do little to advance the linux cause .	"negative"	accuracy	sst2	sst2-337	label-sentiment-0032	0.6583
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: further proof that the epicenter of cool , beautiful , thought-provoking foreign cinema is smack-dab in the middle of dubya 's axis of evil . Return only the label.	further proof that the epicenter of cool , beautiful , thought-provoking foreign cinema is smack-dab in the middle of dubya 's axis of evil .	"positive"	accuracy	sst2	sst2-648	label-sentiment-0033	0.625
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: a woman 's pic directed with resonance by ilya chaiken . Return only the label.	a woman 's pic directed with resonance by ilya chaiken .	"positive"	accuracy	sst2	sst2-164	label-sentiment-0034	0.6337
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: while the ideas about techno-saturation are far from novel , they 're presented with a wry dark humor . Return only the label.	while the ideas about techno-saturation are far from novel , they 're presented with a wry dark humor .	"positive"	accuracy	sst2	sst2-425	label-sentiment-0035	0.6026
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: immersing us in the endlessly inventive , fiercely competitive world of hip-hop djs , the project is sensational and revelatory , even if scratching makes you itch . Return only the label.	immersing us in the endlessly inventive , fiercely competitive world of hip-hop djs , the project is sensational and revelatory , even if scratching makes you itch .	"positive"	accuracy	sst2	sst2-437	label-sentiment-0036	0.5877
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: the film is based on truth and yet there is something about it that feels incomplete , as if the real story starts just around the corner . Return only the label.	the film is based on truth and yet there is something about it that feels incomplete , as if the real story starts just around the corner .	"negative"	accuracy	sst2	sst2-547	label-sentiment-0037	0.6921
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: ( a ) shapeless blob of desperate entertainment . Return only the label.	( a ) shapeless blob of desperate entertainment .	"negative"	accuracy	sst2	sst2-681	label-sentiment-0038	0.6458
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: it 's a charming and often affecting journey . Return only the label.	it 's a charming and often affecting journey .	"positive"	accuracy	sst2	sst2-0	label-sentiment-0039	0.6836
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: a coda in every sense , the pinochet case splits time between a minute-by-minute account of the british court 's extradition chess game and the regime 's talking-head survivors . Return only the label.	a coda in every sense , the pinochet case splits time between a minute-by-minute account of the british court 's extradition chess game and the regime 's talking-head survivors .	"positive"	accuracy	sst2	sst2-266	label-sentiment-0040	0.5921
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: uncommonly stylish but equally silly ... the picture fails to generate much suspense , nor does it ask searching enough questions to justify its pretensions . Return only the label.	uncommonly stylish but equally silly ... the picture fails to generate much suspense , nor does it ask searching enough questions to justify its pretensions .	"negative"	accuracy	sst2	sst2-160	label-sentiment-0041	0.6865
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: green might want to hang onto that ski mask , as robbery may be the only way to pay for his next project . Return only the label.	green might want to hang onto that ski mask , as robbery may be the only way to pay for his next project .	"negative"	accuracy	sst2	sst2-81	label-sentiment-0042	0.5256
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: the film may appear naked in its narrative form ... but it goes deeper than that , to fundamental choices that include the complexity of the catholic doctrine Return only the label.	the film may appear naked in its narrative form ... but it goes deeper than that , to fundamental choices that include the complexity of the catholic doctrine	"positive"	accuracy	sst2	sst2-401	label-sentiment-0043	0.6472
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: funny but perilously slight . Return only the label.	funny but perilously slight .	"positive"	accuracy	sst2	sst2-481	label-sentiment-0044	0.7217
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: this illuminating documentary transcends our preconceived vision of the holy land and its inhabitants , revealing the human complexities beneath . Return only the label.	this illuminating documentary transcends our preconceived vision of the holy land and its inhabitants , revealing the human complexities beneath .	"positive"	accuracy	sst2	sst2-41	label-sentiment-0045	0.617
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: turns potentially forgettable formula into something strangely diverting . Return only the label.	turns potentially forgettable formula into something strangely diverting .	"positive"	accuracy	sst2	sst2-123	label-sentiment-0046	0.7325
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: the fly-on-the-wall method used to document rural french school life is a refreshing departure from the now more prevalent technique of the docu-makers being a visible part of their work . Return only the label.	the fly-on-the-wall method used to document rural french school life is a refreshing departure from the now more prevalent technique of the docu-makers being a visible part of their work .	"positive"	accuracy	sst2	sst2-524	label-sentiment-0047	0.6459
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: if you believe any of this , i can make you a real deal on leftover enron stock that will double in value a week from friday . Return only the label.	if you believe any of this , i can make you a real deal on leftover enron stock that will double in value a week from friday .	"negative"	accuracy	sst2	sst2-253	label-sentiment-0048	0.6581
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: it confirms fincher 's status as a film maker who artfully bends technical know-how to the service of psychological insight . Return only the label.	it confirms fincher 's status as a film maker who artfully bends technical know-how to the service of psychological insight .	"positive"	accuracy	sst2	sst2-794	label-sentiment-0049	0.6282
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: pumpkin means to be an outrageous dark satire on fraternity life , but its ambitions far exceed the abilities of writer adam larson broder and his co-director , tony r. abrams , in their feature debut . Return only the label.	pumpkin means to be an outrageous dark satire on fraternity life , but its ambitions far exceed the abilities of writer adam larson broder and his co-director , tony r. abrams , in their feature debut .	"negative"	accuracy	sst2	sst2-135	label-sentiment-0050	0.5914
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: displaying about equal amounts of naiveté , passion and talent , beneath clouds establishes sen as a filmmaker of considerable potential . Return only the label.	displaying about equal amounts of naiveté , passion and talent , beneath clouds establishes sen as a filmmaker of considerable potential .	"positive"	accuracy	sst2	sst2-359	label-sentiment-0051	0.58
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: dazzling in its complexity , disturbing for its extraordinary themes , the piano teacher is a film that defies categorisation . Return only the label.	dazzling in its complexity , disturbing for its extraordinary themes , the piano teacher is a film that defies categorisation .	"positive"	accuracy	sst2	sst2-484	label-sentiment-0052	0.6487
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: the film flat lines when it should peak and is more missed opportunity and trifle than dark , decadent truffle . Return only the label.	the film flat lines when it should peak and is more missed opportunity and trifle than dark , decadent truffle .	"negative"	accuracy	sst2	sst2-86	label-sentiment-0053	0.5954
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: works hard to establish rounded characters , but then has nothing fresh or particularly interesting to say about them . Return only the label.	works hard to establish rounded characters , but then has nothing fresh or particularly interesting to say about them .	"negative"	accuracy	sst2	sst2-347	label-sentiment-0054	0.7042
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: stephen rea , aidan quinn , and alan bates play desmond 's legal eagles , and when joined by brosnan , the sight of this grandiloquent quartet lolling in pretty irish settings is a pleasant enough thing , ` tis . Return only the label.	stephen rea , aidan quinn , and alan bates play desmond 's legal eagles , and when joined by brosnan , the sight of this grandiloquent quartet lolling in pretty irish settings is a pleasant enough thing , ` tis .	"positive"	accuracy	sst2	sst2-860	label-sentiment-0055	0.6057
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: but what are adults doing in the theater at all ? Return only the label.	but what are adults doing in the theater at all ?	"negative"	accuracy	sst2	sst2-731	label-sentiment-0056	0.5465
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: having had the good sense to cast actors who are , generally speaking , adored by the movie-going public , khouri then gets terrific performances from them all . Return only the label.	having had the good sense to cast actors who are , generally speaking , adored by the movie-going public , khouri then gets terrific performances from them all .	"positive"	accuracy	sst2	sst2-336	label-sentiment-0057	0.6075
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: it 's difficult to imagine the process that produced such a script , but here 's guessing that spray cheese and underarm noises played a crucial role . Return only the label.	it 's difficult to imagine the process that produced such a script , but here 's guessing that spray cheese and underarm noises played a crucial role .	"negative"	accuracy	sst2	sst2-96	label-sentiment-0058	0.6486
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: an overemphatic , would-be wacky , ultimately tedious sex farce . Return only the label.	an overemphatic , would-be wacky , ultimately tedious sex farce .	"negative"	accuracy	sst2	sst2-392	label-sentiment-0059	0.6695
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: nicks , seemingly uncertain what 's going to make people laugh , runs the gamut from stale parody to raunchy sex gags to formula romantic comedy . Return only the label.	nicks , seemingly uncertain what 's going to make people laugh , runs the gamut from stale parody to raunchy sex gags to formula romantic comedy .	"negative"	accuracy	sst2	sst2-25	label-sentiment-0060	0.5989
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: no screen fantasy-adventure in recent memory has the showmanship of clones ' last 45 minutes . Return only the label.	no screen fantasy-adventure in recent memory has the showmanship of clones ' last 45 minutes .	"positive"	accuracy	sst2	sst2-685	label-sentiment-0061	0.5579
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: if the first men in black was money , the second is small change . Return only the label.	if the first men in black was money , the second is small change .	"negative"	accuracy	sst2	sst2-559	label-sentiment-0062	0.6826
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: majidi is an unconventional storyteller , capable of finding beauty in the most depressing places . Return only the label.	majidi is an unconventional storyteller , capable of finding beauty in the most depressing places .	"positive"	accuracy	sst2	sst2-677	label-sentiment-0063	0.6003
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: it 's like watching a nightmare made flesh . Return only the label.	it 's like watching a nightmare made flesh .	"negative"	accuracy	sst2	sst2-272	label-sentiment-0064	0.6355
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: add yet another hat to a talented head , clooney 's a good director . Return only the label.	add yet another hat to a talented head , clooney 's a good director .	"positive"	accuracy	sst2	sst2-859	label-sentiment-0065	0.6115
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: there 's too much falseness to the second half , and what began as an intriguing look at youth fizzles into a dull , ridiculous attempt at heart-tugging . Return only the label.	there 's too much falseness to the second half , and what began as an intriguing look at youth fizzles into a dull , ridiculous attempt at heart-tugging .	"negative"	accuracy	sst2	sst2-715	label-sentiment-0066	0.6407
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: reign of fire looks as if it was made without much thought -- and is best watched that way . Return only the label.	reign of fire looks as if it was made without much thought -- and is best watched that way .	"positive"	accuracy	sst2	sst2-230	label-sentiment-0067	0.6063
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: but taken as a stylish and energetic one-shot , the queen of the damned can not be said to suck . Return only the label.	but taken as a stylish and energetic one-shot , the queen of the damned can not be said to suck .	"positive"	accuracy	sst2	sst2-728	label-sentiment-0068	0.5786
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: writer/director joe carnahan 's grimy crime drama is a manual of precinct cliches , but it moves fast enough to cover its clunky dialogue and lapses in logic . Return only the label.	writer/director joe carnahan 's grimy crime drama is a manual of precinct cliches , but it moves fast enough to cover its clunky dialogue and lapses in logic .	"positive"	accuracy	sst2	sst2-760	label-sentiment-0069	0.5999
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: from the opening scenes , it 's clear that all about the benjamins is a totally formulaic movie . Return only the label.	from the opening scenes , it 's clear that all about the benjamins is a totally formulaic movie .	"negative"	accuracy	sst2	sst2-545	label-sentiment-0070	0.5774
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: pacino is brilliant as the sleep-deprived dormer , his increasing weariness as much existential as it is physical . Return only the label.	pacino is brilliant as the sleep-deprived dormer , his increasing weariness as much existential as it is physical .	"positive"	accuracy	sst2	sst2-186	label-sentiment-0071	0.5833
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: ... a magnificent drama well worth tracking down . Return only the label.	... a magnificent drama well worth tracking down .	"positive"	accuracy	sst2	sst2-738	label-sentiment-0072	0.6267
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: i 've always dreamed of attending cannes , but after seeing this film , it 's not that big a deal . Return only the label.	i 've always dreamed of attending cannes , but after seeing this film , it 's not that big a deal .	"negative"	accuracy	sst2	sst2-548	label-sentiment-0073	0.6112
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: feature debuter d.j. caruso directs a crack ensemble cast , bringing screenwriter tony gayton 's narcotics noir to life . Return only the label.	feature debuter d.j. caruso directs a crack ensemble cast , bringing screenwriter tony gayton 's narcotics noir to life .	"positive"	accuracy	sst2	sst2-567	label-sentiment-0074	0.5833
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: it 's great escapist fun that recreates a place and time that will never happen again . Return only the label.	it 's great escapist fun that recreates a place and time that will never happen again .	"positive"	accuracy	sst2	sst2-109	label-sentiment-0075	0.6499
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: a misogynistic piece of filth that attempts to pass itself off as hip , young adult entertainment . Return only the label.	a misogynistic piece of filth that attempts to pass itself off as hip , young adult entertainment .	"negative"	accuracy	sst2	sst2-188	label-sentiment-0076	0.649
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: two hours fly by -- opera 's a pleasure when you do n't have to endure intermissions -- and even a novice to the form comes away exhilarated . Return only the label.	two hours fly by -- opera 's a pleasure when you do n't have to endure intermissions -- and even a novice to the form comes away exhilarated .	"positive"	accuracy	sst2	sst2-671	label-sentiment-0077	0.5992
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: i had to look away - this was god awful . Return only the label.	i had to look away - this was god awful .	"negative"	accuracy	sst2	sst2-29	label-sentiment-0078	0.6929
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: offers much to enjoy ... and a lot to mull over in terms of love , loyalty and the nature of staying friends . Return only the label.	offers much to enjoy ... and a lot to mull over in terms of love , loyalty and the nature of staying friends .	"positive"	accuracy	sst2	sst2-636	label-sentiment-0079	0.679
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: not only unfunny , but downright repellent . Return only the label.	not only unfunny , but downright repellent .	"negative"	accuracy	sst2	sst2-346	label-sentiment-0080	0.6502
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: the story and the friendship proceeds in such a way that you 're watching a soap opera rather than a chronicle of the ups and downs that accompany lifelong friendships . Return only the label.	the story and the friendship proceeds in such a way that you 're watching a soap opera rather than a chronicle of the ups and downs that accompany lifelong friendships .	"negative"	accuracy	sst2	sst2-292	label-sentiment-0081	0.6325
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: corny , schmaltzy and predictable , but still manages to be kind of heartwarming , nonetheless . Return only the label.	corny , schmaltzy and predictable , but still manages to be kind of heartwarming , nonetheless .	"positive"	accuracy	sst2	sst2-223	label-sentiment-0082	0.6097
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: the words , ` frankly , my dear , i do n't give a damn , ' have never been more appropriate . Return only the label.	the words , ` frankly , my dear , i do n't give a damn , ' have never been more appropriate .	"negative"	accuracy	sst2	sst2-645	label-sentiment-0083	0.6204
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: this is n't even madonna 's swept away . Return only the label.	this is n't even madonna 's swept away .	"negative"	accuracy	sst2	sst2-734	label-sentiment-0084	0.618
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: a marvel like none you 've seen . Return only the label.	a marvel like none you 've seen .	"positive"	accuracy	sst2	sst2-435	label-sentiment-0085	0.6494
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: it 's too bad that the helping hand he uses to stir his ingredients is also a heavy one . Return only the label.	it 's too bad that the helping hand he uses to stir his ingredients is also a heavy one .	"negative"	accuracy	sst2	sst2-461	label-sentiment-0086	0.6646
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: ( director ) o'fallon manages to put some lovely pictures up on the big screen , but his skill at telling a story -- he also contributed to the screenplay -- falls short . Return only the label.	( director ) o'fallon manages to put some lovely pictures up on the big screen , but his skill at telling a story -- he also contributed to the screenplay -- falls short .	"negative"	accuracy	sst2	sst2-575	label-sentiment-0087	0.6284
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: this is a train wreck of an action film -- a stupefying attempt by the filmmakers to force-feed james bond into the mindless xxx mold and throw 40 years of cinematic history down the toilet in favor of bright flashes and loud bangs . Return only the label...	this is a train wreck of an action film -- a stupefying attempt by the filmmakers to force-feed james bond into the mindless xxx mold and throw 40 years of cinematic history down the toilet in favor of bright flashes and loud bangs .	"negative"	accuracy	sst2	sst2-79	label-sentiment-0088	0.5895
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: the film contains no good jokes , no good scenes , barely a moment when carvey 's saturday night live-honed mimicry rises above the level of embarrassment . Return only the label.	the film contains no good jokes , no good scenes , barely a moment when carvey 's saturday night live-honed mimicry rises above the level of embarrassment .	"negative"	accuracy	sst2	sst2-795	label-sentiment-0089	0.5431
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: ... turns so unforgivably trite in its last 10 minutes that anyone without a fortified sweet tooth will likely go into sugar shock . Return only the label.	... turns so unforgivably trite in its last 10 minutes that anyone without a fortified sweet tooth will likely go into sugar shock .	"negative"	accuracy	sst2	sst2-349	label-sentiment-0090	0.6436
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: leigh 's film is full of memorable performances from top to bottom . Return only the label.	leigh 's film is full of memorable performances from top to bottom .	"positive"	accuracy	sst2	sst2-169	label-sentiment-0091	0.632
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: deadeningly dull , mired in convoluted melodrama , nonsensical jargon and stiff-upper-lip laboriousness . Return only the label.	deadeningly dull , mired in convoluted melodrama , nonsensical jargon and stiff-upper-lip laboriousness .	"negative"	accuracy	sst2	sst2-502	label-sentiment-0092	0.6288
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: it gets onto the screen just about as much of the novella as one could reasonably expect , and is engrossing and moving in its own right . Return only the label.	it gets onto the screen just about as much of the novella as one could reasonably expect , and is engrossing and moving in its own right .	"positive"	accuracy	sst2	sst2-726	label-sentiment-0093	0.6321
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: manages to be both repulsively sadistic and mundane . Return only the label.	manages to be both repulsively sadistic and mundane .	"negative"	accuracy	sst2	sst2-58	label-sentiment-0094	0.6735
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: for all its technical virtuosity , the film is so mired in juvenile and near-xenophobic pedagogy that it 's enough to make one pine for the day when godard can no longer handle the rigors of filmmaking . Return only the label.	for all its technical virtuosity , the film is so mired in juvenile and near-xenophobic pedagogy that it 's enough to make one pine for the day when godard can no longer handle the rigors of filmmaking .	"negative"	accuracy	sst2	sst2-800	label-sentiment-0095	0.5477
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: looks and feels like a project better suited for the small screen . Return only the label.	looks and feels like a project better suited for the small screen .	"negative"	accuracy	sst2	sst2-136	label-sentiment-0096	0.6836
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: i just loved every minute of this film . Return only the label.	i just loved every minute of this film .	"positive"	accuracy	sst2	sst2-534	label-sentiment-0097	0.6316
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: visually rather stunning , but ultimately a handsome-looking bore , the true creativity would have been to hide treasure planet entirely and completely reimagine it . Return only the label.	visually rather stunning , but ultimately a handsome-looking bore , the true creativity would have been to hide treasure planet entirely and completely reimagine it .	"negative"	accuracy	sst2	sst2-597	label-sentiment-0098	0.6319
labeling	sentiment	You are a data labeling engine. Follow the output contract exactly.	Label the sentiment of the text as positive or negative. Text: this time mr. burns is trying something in the martin scorsese street-realist mode , but his self-regarding sentimentality trips him up again . Return only the label.	this time mr. burns is trying something in the martin scorsese street-realist mode , but his self-regarding sentimentality trips him up again .	"negative"	accuracy	sst2	sst2-213	label-sentiment-0099	0.6345

End of preview.

Data Farm Tester — Labeling

A sealed 1,000-task labeling benchmark for data-cleaning/labeling models. It was built to measure the capability the product actually needs, after the existing benchmark turned out to be dominated by SQL and a deterministic cleaning contract.

Subsets

subset	n	source	output	metric
sentiment	200	SST-2 validation	positive/negative	accuracy
structured	200	SST-2 validation	JSON `{label, confidence}`	label accuracy + JSON validity
intent	200	Banking77 test	one of 77 intents	accuracy
nli	200	MNLI matched	entailment/neutral/contradiction	accuracy
ner	200	CoNLL-2003 test	JSON entity array	entity F1

Leakage validation

Against 1,029,903 texts / 293,131 unique from the full SFT lineage:

gate	result
exact input overlap	0
payload-line overlap (≥20 chars)	0
13-gram overlap	0
BGE max cosine	0.799 (threshold 0.80)
tasks dropped for similarity	14
source train-split overlap in the benchmark	0 / 1,000
internal duplicates	0

Results

Protocol: vLLM, greedy, thinking off, 4096 context.

subset	base	v9	v15	α=0.5
sentiment	89.0%	83.5%	83.5%	88.5%
structured	90.0%	86.0%	65.0%	90.5%
JSON validity	100%	97.0%	75.0%	100%
intent	48.5%	45.5%	43.0%	50.5%
nli	66.5%	55.5%	53.0%	70.5%
ner F1	38.3%	37.0%	27.8%	44.2%
macro	66.5%	61.5%	54.5%	68.8%

The full SFT (v15) loses 11 macro points, including the JSON contract (validity 100% → 79%) and NER (−10.7 F1). α=0.5 is the best checkpoint on every subset and significantly better than base on NER (+6.4 F1).

Files

datafarm_tester.jsonl — 1,000 tasks
manifest.json — counts, thresholds, SHA-256
preds/{base,v9,v15,alpha50}.jsonl
scores/{base,v9,v15,alpha50}.jsonl and *.summary.json
ATTRIBUTION.md

Caveats

200 tasks per subset → ±6 point CIs.
English only.
The JSON subset uses sentiment labels; schema-heavy extraction is not yet covered.
Public corpora may be in pretraining; the base-vs-fine-tune delta is still valid.

Attribution

SST-2 and MNLI (GLUE, CC-BY-SA-4.0), Banking77 (CC-BY-4.0), CoNLL-2003 (CC-BY-SA-4.0).

Downloads last month: 107

Total file size:

2.24 MB