Comprehensive and Efficient Distillation for Lightweight Sentiment Analysis Models
Paper • 2510.24425 • Published
Dataset Preview
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed because of a cast error
Error code: DatasetGenerationCastError
Exception: DatasetGenerationCastError
Message: An error occurred while generating the dataset
All the data files must have the same columns, but at some point there are 1 new columns ({'Qwen/Qwen3-32B_output'}) and 1 missing columns ({'meta-llama/Meta-Llama-3.1-70B-Instruct_output'}).
This happened while the json dataset builder was generating data using
hf://datasets/zhang-yice/sentiment-distillation-v2/Qwen3-32B_50k.jsonl (at revision 0ee48fc022bb45ad9f2c66c43fc1c73db6c3f4fb)
Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1831, in _prepare_split_single
writer.write_table(table)
File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 714, in write_table
pa_table = table_cast(pa_table, self._schema)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2272, in table_cast
return cast_table_to_schema(table, schema)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2218, in cast_table_to_schema
raise CastError(
datasets.table.CastError: Couldn't cast
text: string
cluster_id: string
selected_cluster_id: string
task_name: string
instruction: string
input: string
ppl: double
ifd: double
avg_rank_score: double
text_length: int64
Qwen/Qwen3-32B_output: string
to
{'text': Value('string'), 'cluster_id': Value('string'), 'selected_cluster_id': Value('string'), 'task_name': Value('string'), 'instruction': Value('string'), 'input': Value('string'), 'ppl': Value('float64'), 'ifd': Value('float64'), 'avg_rank_score': Value('float64'), 'text_length': Value('int64'), 'meta-llama/Meta-Llama-3.1-70B-Instruct_output': Value('string')}
because column names don't match
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1339, in compute_config_parquet_and_info_response
parquet_operations = convert_to_parquet(builder)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 972, in convert_to_parquet
builder.download_and_prepare(
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 894, in download_and_prepare
self._download_and_prepare(
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 970, in _download_and_prepare
self._prepare_split(split_generator, **prepare_split_kwargs)
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1702, in _prepare_split
for job_id, done, content in self._prepare_split_single(
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1833, in _prepare_split_single
raise DatasetGenerationCastError.from_cast_error(
datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset
All the data files must have the same columns, but at some point there are 1 new columns ({'Qwen/Qwen3-32B_output'}) and 1 missing columns ({'meta-llama/Meta-Llama-3.1-70B-Instruct_output'}).
This happened while the json dataset builder was generating data using
hf://datasets/zhang-yice/sentiment-distillation-v2/Qwen3-32B_50k.jsonl (at revision 0ee48fc022bb45ad9f2c66c43fc1c73db6c3f4fb)
Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)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.
text string | cluster_id string | selected_cluster_id string | task_name string | instruction string | input string | ppl float64 | ifd float64 | avg_rank_score float64 | text_length int64 | meta-llama/Meta-Llama-3.1-70B-Instruct_output string |
|---|---|---|---|---|---|---|---|---|---|---|
#PeoplesDrugMart Terrace is #nowplaying Glory of Love by #PeterCetera #cubevenue | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.049498 | 0.000009 | 0 | 80 | [] | ||
@USER EVEN THO THEY SAYING U WON .. ITS INEVITABLE OF STOPPING WHAT GOD GOT TRUMP DOING .. #FREETHECHILDREN #SEXSLAVERY #PIZZAGATE #KILLUMINATI #FUCKTHEDEEPSTATE | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.031331 | 0.000009 | 0 | 161 | [] | ||
Its a brass hammer. Not much to say really. Have not put it through much use yet since it is mainly for more specialty jobs. Seems well made. MADE IN USA which is a major plus. Bought it for firearms work. Maybe the 16oz was a little heavy for that, I imagine I will be choked up on the handle close to the head with thi... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.075839 | 0.085112 | 0.388889 | 608 | [
{
"phrase": "Seems well made.",
"sentiment_intensity": "slightly friendly"
},
{
"phrase": "MADE IN USA which is a major plus.",
"sentiment_intensity": "friendly"
}
] | ||
You get what you pay for and this hat applies. I would point out that for the price, the value of this hat is good. While I would say it is a bit flimsy and cheaply put together, it certainly will work for a simple cheap boonie hat. I have the Under Armour version of a boonie and there is no denying that there is no co... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.057639 | 0.131628 | 0.416667 | 474 | [
{
"phrase": "I would point out that for the price, the value of this hat is good",
"sentiment_intensity": "friendly"
},
{
"phrase": "All in all I am happy with the purchase",
"sentiment_intensity": "friendly"
}
] | ||
I've been going for a few months now and I get the shellac over my nails with the gel nail polish. I'm a life long nail biter so this has definitely upped my professional appearance related to my nails. Kimberly does my nails and they come out great looking every time. The gel polish lasts for the entire two weeks unti... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.037805 | 0.102517 | 0.416667 | 507 | [
{
"phrase": "I'm a life long nail biter so this has definitely upped my professional appearance related to my nails",
"sentiment_intensity": "friendly"
},
{
"phrase": "Kimberly does my nails and they come out great looking every time",
"sentiment_intensity": "friendly"
},
{
"phrase": "I ... | ||
Fiskars basically says this hatchet is indestructible -- bold words, but once you see this item you will question them no longer. Unless you swing it for a half-hour into a brick wall or throw it in a roaring fire, this is a forever hatchet. Besides making it free, there is nothing Fiskars can do to improve it. | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.06932 | 0.055862 | 0.416667 | 312 | [
{
"phrase": "Fiskars basically says this hatchet is indestructible -- bold words, but once you see this item you will question them no longer.",
"sentiment_intensity": "friendly"
},
{
"phrase": "Besides making it free, there is nothing Fiskars can do to improve it.",
"sentiment_intensity": "very... | ||
A classic film ! . This is one of the best films that i have ever watched . it has a very strong storyline , with scenes of action and plenty of drama throughout . A film that had such a wide variety of genre , action , drama , thriller , romance . although there is romantic reference , there is not too much of it , an... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.076427 | 0.209964 | 0.416667 | 396 | [
{
"phrase": "A classic film!",
"sentiment_intensity": "friendly"
},
{
"phrase": "This is one of the best films that i have ever watched",
"sentiment_intensity": "very friendly"
},
{
"phrase": "A must-see film!",
"sentiment_intensity": "extremely friendly"
}
] | ||
Cheesesteak limbo would be a more apt name for this establishment, at least for the full customer experience. First of all I have to say the cheesesteaks are quite tasty; the steak is tender and flavored nicely and the other ingredients come together to form a fine sanwich. It is lacking in size however. For ten bones ... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.10098 | 0.080857 | 0.5 | 945 | [
{
"phrase": "the cheesesteaks are quite tasty",
"sentiment_intensity": "friendly"
},
{
"phrase": "the steak is tender and flavored nicely and the other ingredients come together to form a fine sanwich",
"sentiment_intensity": "friendly"
}
] | ||
Who doesn't love Elmo?! Everyone loves Elmo. Well, As a parent, I saw a selfish Elmo that cared only for himself and hurt others and cause much pain. This is a terrible movie for kids to learn from. No one even addressed the selfishness that got Elmo in grouchland in the first place. Not a good thing for kids to watch.... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.098649 | 0.046247 | 0.5 | 351 | [
{
"phrase": "Who doesn't love Elmo?! Everyone loves Elmo.",
"sentiment_intensity": "extremely friendly"
}
] | ||
Movie was just too irritating . If you have n't watched ` There 's something about Mary ' , you might enjoy the story of the movie . Only the ` scientist ' variable is an added component to the story . Akshay was too irritating after the first 15 minutes of his intro . But I was impressed with Johnny Lever 's comedy ra... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.091405 | 0.085389 | 0.52381 | 611 | [
{
"phrase": "you might enjoy the story of the movie",
"sentiment_intensity": "slightly friendly"
},
{
"phrase": "I was impressed with Johnny Lever's comedy rather than Paresh Rawal",
"sentiment_intensity": "friendly"
},
{
"phrase": "I was impressed that they did not take a complete rip o... | ||
RT @USER : Well on the bright side I am undefeated in 2018 and that is all that really matters . Happy New Years my dudes . | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.996353 | 0.070088 | 0.535271 | 123 | [
{
"phrase": "Happy New Years my dudes",
"sentiment_intensity": "friendly"
}
] | ||
Exceeded my expectations . This movie sounded like one I would like , and turned out to be one I \* love \* . I found the main characters both believable and surprisingly likable . Movie-makers like to portray the couples in this type of movie as bickering , selfish , loveless bastards who in some way deserved it . The... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.106372 | 0.243471 | 0.542857 | 836 | [
{
"phrase": "Exceeded my expectations",
"sentiment_intensity": "very friendly"
},
{
"phrase": "turned out to be one I * love *",
"sentiment_intensity": "extremely friendly"
},
{
"phrase": "I found the main characters both believable and surprisingly likable",
"sentiment_intensity": "... | ||
Can't believe I haven't reviewed this place yet. Love this place, a couple previous restaurants have come and go in this location but I really hope Bento Hut sticks around. Sushi is great and good lunch specials (even without specials though it's still extremely affordable especially for the quality). I seem to always ... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.229697 | 0.213432 | 0.652778 | 1,537 | [
{
"phrase": "Love this place",
"sentiment_intensity": "very friendly"
},
{
"phrase": "Sushi is great and good lunch specials",
"sentiment_intensity": "friendly"
},
{
"phrase": "Inside is really nicely done",
"sentiment_intensity": "friendly"
},
{
"phrase": "Win",
"sentime... | ||
S t A r B u C k S . The Starbucks Coffee Company up in Seattle Washington was established the year this sweet movie was made . BUT since it was filmed no where near Seattle where the Starbucks headquarters is currently located , we see no Starbucks coffee houses anywhere in the movie , making this comment completely ir... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.179348 | 0.063077 | 0.654279 | 699 | [
{
"phrase": "forgive my ignorance and enjoy the wonderful beverages that starbucks has to offer",
"sentiment_intensity": "friendly"
},
{
"phrase": "oh and A Clockwork Orange is awesome cuz the guy who did the production design also did it for Star Wars episode 4",
"sentiment_intensity": "friendl... | ||
It 's about hustling , indeed . According to the movie slang , a `` slide '' is a guy so dumb he even has no idea he 's being framed on . I wish I know someone that would knowingly allow to be framed , so the term is kind of a lame one to me . Anyway , when you finish watching this movie-crap you feel that someone got ... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.257646 | 0.035526 | 0.666667 | 831 | [
{
"phrase": "But I have to agree with those who still find something enjoyable on it",
"sentiment_intensity": "slightly friendly"
}
] | ||
Eddie does it again in this breakthrough original motion picture . This is what needs to happen on most reality shows . . then they might actually be watchable . Eddie Furlong captured the essence of a lunatic with great passion and enthusiasm . I 've seen this movie about a hundred times now , and will probably watch ... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.066322 | 0.143406 | 0.666667 | 820 | [
{
"phrase": "Eddie does it again in this breakthrough original motion picture",
"sentiment_intensity": "very friendly"
},
{
"phrase": "Eddie Furlong captured the essence of a lunatic with great passion and enthusiasm",
"sentiment_intensity": "very friendly"
},
{
"phrase": "I 've seen thi... | ||
I'd been wanting a way to add casters to my coffee table (the Tofteryd from IKEA). These worked perfectly for that! The legs of the coffee table are exactly 1" tubes, but these sockets were a bit loose. I just wrapped a bit of tape around them, and then they fit snugly inside the legs. I used them with Sheperd casters ... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.171821 | 0.099761 | 0.666667 | 476 | [
{
"phrase": "These worked perfectly for that!",
"sentiment_intensity": "extremely friendly"
},
{
"phrase": "I used them with Sheperd casters that had the same dimensions as these sockets",
"sentiment_intensity": "friendly"
},
{
"phrase": "now everything is secure and working well",
"... | ||
Salt lick was the best bbq I have ever had. It was $25 per person for family style all you can eat. I have had BBQ after this place and it never compares. Some restaurants charge you for leftovers but not salt lick! They let you take it home and enjoy it for breakfast. You may not have room for it but make sure you get... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.185154 | 0.174674 | 0.6875 | 411 | [
{
"phrase": "Salt lick was the best bbq I have ever had.",
"sentiment_intensity": "very friendly"
},
{
"phrase": "They let you take it home and enjoy it for breakfast.",
"sentiment_intensity": "friendly"
},
{
"phrase": "make sure you get the cobbler.",
"sentiment_intensity": "friendl... | ||
Keep the magic in our world ! #FightFor1Point5 #MakeParisReal #ClimateActionNow #FaceTheClimateEmergency #StrongerTogether #FridaysForFuture #globalclimatejustice ! | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.209182 | 0.048439 | 0.75 | 164 | [
{
"phrase": "Keep the magic in our world!",
"sentiment_intensity": "friendly"
}
] | ||
Put The Matrix on Acid and you get Dark City -LRB- Minor Spoilers -RRB- . It 's so coincidental it 's scary . Dark City is a Sci-Fi thriller that is like the Matrix if only it was clone and on acid . It stars Rufus Sewell as John Murdoch , a man who wakes up only to find out he has a wife that he never met , the cops w... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.134379 | 0.184054 | 0.75 | 1,240 | [
{
"phrase": "The special effects are incredible",
"sentiment_intensity": "very friendly"
},
{
"phrase": "The look of this film is amazing and will have you in a daze",
"sentiment_intensity": "extremely friendly"
},
{
"phrase": "The performances are all right",
"sentiment_intensity": ... | ||
The space is kind of cool with some funky chairs and décor and a big open layout, but not the best venue for all concert types. I went to see Rhye, a more low key performance, and the vibe was ruined by all the talking that was going on that sounded more like being in bar. The singer even tried to get people to mellow ... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.172938 | 0.052559 | 0.75 | 957 | [
{
"phrase": "The space is kind of cool with some funky chairs and décor and a big open layout",
"sentiment_intensity": "slightly friendly"
},
{
"phrase": "I would go back if it was a louder type band",
"sentiment_intensity": "friendly"
}
] | ||
I got a vegan thali delivered. It was really good. Full of flavor and spicy. I ordered the chole, the aloo gobi, and the bengan bharta. All very tasty. I will definitely order again. I wish the naan was vegan but otherwise I was very happy with the choices. Delicious! | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.065878 | 0.303671 | 0.75 | 268 | [
{
"phrase": "It was really good.",
"sentiment_intensity": "friendly"
},
{
"phrase": "Full of flavor and spicy.",
"sentiment_intensity": "friendly"
},
{
"phrase": "All very tasty.",
"sentiment_intensity": "friendly"
},
{
"phrase": "I will definitely order again.",
"sentime... | ||
A Waste of Time . Admittedly , I watched this movie primarily because I 'm nuts about Nicole Kidman - more to the point , I just like looking at Nicole Kidman . As I read the summary for the movie , it seemed intriguing because of my previous experience with Kidman 's performance in the suspenseful thriller , `` The Ot... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.349394 | 0.117401 | 0.8 | 858 | [
{
"phrase": "it was visually stunning and beautiful",
"sentiment_intensity": "friendly"
},
{
"phrase": "The cast did an adequate job of bringing the audience into the lives of the characters they portrayed",
"sentiment_intensity": "slightly friendly"
}
] | ||
Distortion of jury system . This movie starts off by not defining the concept of reasonable doubt as opposed to any doubt at all . Having been a grand juror , I can state that reasonable doubt means that the evidence can not point to guilt and only guilt . Gut feelings and emotion have no place in this . I give it an e... | 92 | Friendly Sentiment Extraction | Extract sentences or phrases from the input text that convey a friendly sentiment and provide the corresponding sentiment intensity. The output should be a list of extracted phrases with their sentiment intensity scores.
Expected Output Format:
The output should be a JSON list of objects, where each object contains th... | 1.119052 | 0.078706 | 1 | 1,276 | [
{
"phrase": "I give it an eight because it is still great drama",
"sentiment_intensity": "friendly"
},
{
"phrase": "I loved this movie until I actually served on a jury",
"sentiment_intensity": "friendly"
}
] | ||
The first time I went to Magic Kingdom, a few years ago, it was probably one of the worst experiences of my life. My friend and I were constantly having parents trying to cut us in the line because we weren't children or had any children with us. We didn't know we could ask for buttons, nor did we know about Fast Passe... | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 1.392315 | 0.380085 | 0.559295 | 1,105 | [
{
"original_sentence": "The first time I went to Magic Kingdom, a few years ago, it was probably one of the worst experiences of my life.",
"suggested_improvement": "My initial visit to Magic Kingdom a few years ago didn't quite live up to my expectations, but I'm glad I gave it another chance."
},
{
... | ||
A delightful animated movie for adults . I only bought this video to hear the voice of Bradley Cole who does Will , but to my surprise , I thoroughly enjoyed the movie . They are the most interesting little animated characters I 've ever seen . It was easy to forget that they were animated at all . Sci-fi is not on my ... | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 1.2913 | 0.328704 | 0.568044 | 380 | [
{
"original_sentence": "A delightful animated movie for adults.",
"suggested_improvement": "I was thoroughly entertained by this delightful animated movie, which I found to be a great choice for adults."
},
{
"original_sentence": "I only bought this video to hear the voice of Bradley Cole who does W... | ||
One of the best . Clearly this is one of the classic examples of a 60 's and 70 's Movie adaptation of a book that is an unsung masterpiece of how you can as a producer , Director of Photography , Set Designer or Special Effects coordinator relay the textual concept , intention and feel of the original author in to a v... | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 1.376602 | 0.231623 | 0.574166 | 1,444 | [
{
"original_sentence": "One of the best.",
"suggested_improvement": "I highly recommend this movie, it's truly exceptional."
},
{
"original_sentence": "Clearly this is one of the classic examples of a 60 's and 70 's Movie adaptation of a book that is an unsung masterpiece of how you can as a produc... | ||
Great piece ! However Justice Isa made a cardinal mistake : he took on the country 's mendacious , corrupt and ill-meaning #DeepState . ' Abbu Kissi kay nahein ; video sab ki baneingi ' ! C . c . @USER The SC will turn cartwheels , and even do headstands to please the #DeepState ! | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 1.822792 | 0.174657 | 0.596123 | 281 | [
{
"original_sentence": "Great piece!",
"suggested_improvement": "I thoroughly enjoyed this article and appreciate the author's insights."
},
{
"original_sentence": "However Justice Isa made a cardinal mistake : he took on the country 's mendacious, corrupt and ill-meaning #DeepState.",
"suggeste... | ||
Retropective review ! . Noticing that the inevitable sequel to Underworld was about to be sprung upon us , I went looking for a review of it . Unfortunately , all I have to go on is the initial small tide of test audiences -LRB- who are probably fans of the 1st -RRB- . The critics have not been allowed a bite of the ch... | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 1.389916 | 0.183451 | 0.59969 | 1,480 | [
{
"original_sentence": "Retropective review!. Noticing that the inevitable sequel to Underworld was about to be sprung upon us, I went looking for a review of it.",
"suggested_improvement": "I recently took a look back at the original Underworld film, as the sequel is about to be released, and I wanted to s... | ||
A series of anti-climatic events . I was expecting this movie to be great from the reviews and , I 'm not going to lie , the interest-drawing title . The movie started out just fine with a good opening scene and was even going fine but was totally ruined by the time we reached the middle . There was n't a main struggle... | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 1.446491 | 0.311742 | 0.608474 | 1,087 | [
{
"original_sentence": "A series of anti-climatic events.",
"suggested_improvement": "Unfortunately, the movie was marred by a series of underwhelming events that failed to live up to my expectations."
},
{
"original_sentence": "I was expecting this movie to be great from the reviews and, I 'm not g... | ||
@USER I believe that 's a photo of your breakfast . You just burnt the white of the egg ! | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 1.35596 | 0.106034 | 0.638889 | 89 | [
{
"original_sentence": "@USER I believe that 's a photo of your breakfast.",
"suggested_improvement": "@USER I think that might be a photo of your breakfast?"
},
{
"original_sentence": "You just burnt the white of the egg!",
"suggested_improvement": "It looks like the egg might be slightly overc... | ||
ON THE ONES AND TWOS HTTPURL #dj #music #becreative HTTPURL | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 2.373543 | 0.170619 | 0.640771 | 59 | [
{
"original_sentence": "ON THE ONES AND TWOS HTTPURL #dj #music #becreative HTTPURL",
"suggested_improvement": "Get ready to groove with the latest beats on the ones and twos! Check out this amazing DJ set at HTTPURL and let the music inspire your creativity! #dj #music #becreative"
}
] | ||
RT @USER : Nag-eenjoy talaga si Maine noh . She must have missed barangay and JoWaPao so much too ! I can listen to her hagikgik na pa ... | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 1.645444 | 0.087892 | 0.644218 | 138 | [
{
"original_sentence": "RT @USER : Nag-eenjoy talaga si Maine noh.",
"suggested_improvement": "I think Maine really enjoyed herself, don't you agree?"
},
{
"original_sentence": "She must have missed barangay and JoWaPao so much too!",
"suggested_improvement": "I'm sure she must have missed her f... | ||
That voice in your head (no, not an actual voice) that is the self talk that makes you feel bad. We all have it. Rick is an expert at getting you to see what you are doing to yourself that is preventing you from enjoying life. I've recommended this book to many friends and all have found it to be life changing. This bo... | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 1.39162 | 0.340253 | 0.650914 | 460 | [
{
"original_sentence": "That voice in your head (no, not an actual voice) that is the self talk that makes you feel bad.",
"suggested_improvement": "We all experience that internal dialogue that can sometimes be quite critical and affect our well-being."
},
{
"original_sentence": "We all have it.",
... | ||
Another Great Clips in the area. This place has a great deal for the summer of 2016. Every weekend they have $5.99 haircuts until the end of July. I received prompt service and actually had someone who knows how to cut. I believe I had the shift manager as she had to be pulled away to help with another customer's hair.... | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 1.560084 | 0.354984 | 0.653274 | 690 | [
{
"original_sentence": "Another Great Clips in the area.",
"suggested_improvement": "I'm excited to see another Great Clips location in the area, offering more convenience for customers."
},
{
"original_sentence": "This place has a great deal for the summer of 2016.",
"suggested_improvement": "I... | ||
Great piece of equipment !! The documentation could be written a little more clearly, but I believe all the information is there to get connected. For me, setup took less than twenty minutes. I've had one print problem with it in 3 months. All I had to do was launch the administration website and reboot the switch. The... | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 1.422068 | 0.322302 | 0.678333 | 565 | [
{
"original_sentence": "Great piece of equipment!!",
"suggested_improvement": "I'm extremely satisfied with this equipment, it's been a great addition to my setup."
},
{
"original_sentence": "The documentation could be written a little more clearly, but I believe all the information is there to get ... | ||
Congrats @USER savor the moment . Ignore the static . Keep moving forward . #MAGA | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 1.41423 | 0.18911 | 0.747917 | 81 | [
{
"original_sentence": "Congrats @USER savor the moment.",
"suggested_improvement": "Congratulations @USER, I hope you're enjoying this moment."
},
{
"original_sentence": "Ignore the static.",
"suggested_improvement": "Please disregard any negative comments."
},
{
"original_sentence": "K... | ||
@USER @USER @USER @USER Agreed , but it 's that winning the primaries part that gives me worry . I will occasionally venture out into political Twitter land beyond the YangGang , and Let 's just say I 'll give it mixed reviews at best . ( Being very generous there ) . | 168 | Politeness Improvement Suggestions | Provide suggestions for improving the politeness of a given text by rephrasing sentences or adding polite language. The output should be a list of suggested improvements in a structured JSON format.
Expected Output Format:
The output should be a JSON object with the following structure:
[
{
"original_sentence":... | 2.077108 | 0.144906 | 0.749226 | 268 | [
{
"original_sentence": "@USER @USER @USER @USER Agreed, but it 's that winning the primaries part that gives me worry.",
"suggested_improvement": "I appreciate your input, @USER @USER @USER @USER. However, I do have some concerns regarding the primary elections."
},
{
"original_sentence": "I will oc... |
End of preview.
Comprehensive and Efficient Distillation for Lightweight Sentiment Analysis Models
Recent efforts leverage knowledge distillation techniques to develop lightweight and practical sentiment analysis models. These methods are grounded in human-written instructions and large-scale user texts. Despite the promising results, two key challenges remain: (1) manually written instructions are limited in diversity and quantity, making them insufficient to ensure comprehensive coverage of distilled knowledge; (2) large-scale user texts incur high computational cost, hindering the practicality of these methods. To this end, we introduce CompEffDist, a comprehensive and efficient distillation framework for sentiment analysis. Our framework consists of two key modules: attribute-based automatic instruction construction and difficulty-based data filtering, which correspondingly tackle the aforementioned challenges. Applying our method across multiple model series (Llama-3, Qwen-3, and Gemma-3), we enable 3B student models to match the performance of 20x larger teacher models on most tasks. In addition, our approach greatly outperforms baseline methods in data efficiency, attaining the same performance level with only 10% of the data.
@inproceedings{xie-etal-2025-comprehensive,
title = "Comprehensive and Efficient Distillation for Lightweight Sentiment Analysis Models",
author = "Xie, Guangyu and
Zhang, Yice and
Bao, Jianzhu and
Wang, Qianlong and
Sun, Yang and
Wang, Bingbing and
Xu, Ruifeng",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.emnlp-main.1122/",
doi = "10.18653/v1/2025.emnlp-main.1122",
pages = "22081--22102",
ISBN = "979-8-89176-332-6",
}
- Downloads last month
- 17