TR-OOLONG v0.10.0

DATACARD.md

@@ -1,8 +1,8 @@
 # TR-OOLONG datacard
 
-Covers **v0.9.0**: 11
-subsets, 353 documents, 2,521 questions, 7 question types, Turkish and English
-(1,746 / 775 questions). Every answer is computed
+Covers **v0.10.0**: 11
+subsets, 347 documents, 2,362 questions, 7 question types, Turkish and English
+(1,615 / 747 questions). Every answer is computed
 from the source dataset's labels, twice, by two independent pieces of code
 that must agree. No answer was written by hand.
 
@@ -16,6 +16,7 @@ For what the benchmark is and how to use it, see the [README](README.md).
 | 0.7.1 | 2026-09-20 | adds `uid`, `dataset`, `haystack_uid`. No question, answer, document or grade changed |
 | 0.8.0 | 2026-09-28 | adds 174 close comparisons (`core`); removes the 137 brand questions; adds `role` to every question; stops releasing difficulty grades. Documents and all other questions unchanged |
 | 0.9.0 | 2026-09-29 | adds 158 core documents with 244 close comparisons (418 core in total), 91 core questions identical in Turkish and English. All earlier documents and questions unchanged |
+| 0.10.0 | 2026-10-01 | cleaning (newspaper mastheads, reviews that write their score, HTML); close comparisons only between label pairs that can be told apart from the text; film ratings converted to sentiment. 309 core questions, 77 identical across languages. Six datasets' documents changed; the intent and complaint documents did not |
 
 ## Fields
 
@@ -56,11 +57,12 @@ and record counts, and the question-type counts.
 | `amazon_hpc_en` | McAuley-Lab/Amazon-Reviews-2023, Health & Personal Care | product reviews + brand | sentiment from the writer's own stars | 3 |
 | `musteri_tr` | turkish-nlp-suite/MusteriYorumlari (Hepsiburada, Trendyol) | shopping reviews | sentiment from the writer's own stars | 3 |
 | `marc_en` | SetFit/amazon_reviews_multi_en (MARC) | shopping reviews | sentiment from the writer's own stars | 3 |
-| `sinema_tr` | turkish-nlp-suite/BuyukSinema | film reviews | the writer's own 1-10 rating | 10 |
+| `sinema_tr` | turkish-nlp-suite/BuyukSinema | film reviews | sentiment from the writer's own 1-10 rating | 3 |
 | `sikayet_tr` | Kaggle savasy/multiclass-classification-data-for-turkish-tc32 | consumer complaints | product category, chosen by the person filing | 29 |
 | `interpress_tr` | Interpress Turkish news, 270k | news articles | newspaper section, set by the publisher | 16 |
 
-Star ratings map to sentiment as 1-2 negative, 3 neutral, 4-5 positive.
+Star ratings map to sentiment as 1-2 negative, 3 neutral, 4-5 positive; the
+film set's 10-point ratings as 1-4 negative, 5-6 neutral, 7-10 positive.
 
 Notes per source, only where something is not obvious:
 
@@ -78,14 +80,20 @@ Notes per source, only where something is not obvious:
   (`kargo-nakliyat` 84.7%, `cep-telefon-kategori` 73.4%, `anne-bebek` 36.9%).
   The build must run with `leak_label_words: true`. Kaggle needs an account, so
   the fetch script takes a path to a local copy of `ticaret-yorum.csv`.
-- **`interpress_tr`.** Records are full articles (median 1,650 characters), so a
+- **`interpress_tr`.** 2.1% of source records were a newspaper's masthead
+  (publisher, editors, printing house) rather than an article; from v0.10.0 a
+  record with 3 or more masthead job titles, or 2 plus an issue header ("Yıl: 3
+  Sayı: 1531"), is removed. Records are full articles (median 1,650 characters), so a
   100K-token document holds about 200 of them. 35.1% of articles were removed
   because they name their own section, so the shipped articles are a filtered
   subset of the corpus. `savunma` fell below the size floor, leaving 16 of 17
   sections. The source has daily publication dates (2010-2017); they are kept
   but not used by any question.
-- **`sinema_tr`.** Ratings are uneven (2.4% at 3 stars, 24.4% at 8 stars), so
-  some labels are naturally rare in a document.
+- **`sinema_tr`.** Used with all 10 ratings as labels until v0.9.0. An exact
+  rating cannot be read from the text ("loved it" can be a 7 or a 10), and every
+  question type depended on it, so from v0.10.0 the ratings are mapped to
+  sentiment. 11.2% of reviews wrote their score in the text ("6/10", "80/100")
+  and are removed.
 - **`vitamins_tr` / `amazon_hpc_en`.** The only pair with a brand per record,
   printed in the text as `[[Brand]]`. Questions about brands were removed in
   v0.8.0 because searching for the printed name answers them; the markers stay
@@ -95,7 +103,10 @@ Notes per source, only where something is not obvious:
   ordinary words. Questions use the printed marker, so answers are unaffected.
 
 Records removed for containing a label name: MASSIVE 0.0%, `sinema_tr` 0.3%,
-`sikayet_tr` 24.6%, `interpress_tr` 35.1%.
+`sikayet_tr` 24.6%, `interpress_tr` 35.1%. Reviews removed because they write
+their score ("7/10", "5 yıldız veriyorum", "1 star"), from v0.10.0: film 6,129,
+MARC 3,570, Amazon 1,488, shopping 639, supplements 449. HTML ("<br />", in
+12.7% of Amazon reviews) is replaced by a space.
 
 ## Licences
 
@@ -156,8 +167,8 @@ the data.
 7. Add **core documents** (v0.9.0). Random label mixes rarely put two large
    labels close together, so these documents are built from exact counts: one
    or two pairs of labels each get a set share of the document (40% each for
-   positive and negative on the review sets, 15% on complaints, news and
-   films, 7% on intent) with a gap inside the step 6 window; the other labels
+   positive and negative on the review sets, 35% on the film set, 15% on
+   complaints and news, 7% on intent) with a gap inside the step 6 window; the other labels
    share the rest. Sizes are in records (2,500 / 6,000 / 12,000 on the review
    sets, 3,000 on intent, 1,000 to 10,000 elsewhere, reaching about 1M tokens
    on complaints, news and films). The two counts differ by at least 5
@@ -169,6 +180,16 @@ the data.
    label more common in the source and the less common one, or between
    positive and negative), and whether the answer is the first-named label
    (every other question).
+8. **Only separable label pairs** (v0.10.0). A close comparison may only use two
+   labels that a word-count classifier tells apart at 0.8 or more AND that it
+   recognises among all labels at least 60% of the time
+   (`experiments/separable_pairs/`, one file per dataset; the Turkish and
+   English intent sets share the pairs allowed in both languages). This rules
+   out pairs decided by which box an editor, reviewer or filer ticked: news
+   "turizm" vs "seyahat", or complaints filed under "alışveriş" about a refund
+   with no product named. Labels kept for core pairs: 32 of 48 intents, 23 of 29
+   complaint categories, 5 of 16 news sections (health, technology, celebrity,
+   food, sport), and positive vs negative on every sentiment set.
 
 ## Label quality
 
@@ -200,7 +221,7 @@ label of each record they read. Full setup and examples:
 
 | role | questions | strongest shortcut found | reader of everything |
 |---|---:|---|---|
-| core: close comparisons | 418 | sampling half the document 0.63; top 10% by topic search 0.56; guessing 0.50 | 0.85 at 95% labelling accuracy, 0.76 at 90% |
+| core: close comparisons | 309 | sampling half the document 0.62; top 10% by topic search 0.54; guessing 0.50 | 0.86 at 95% labelling accuracy, 0.76 at 90% |
 | retrieval: rare-label counts | 265 | topic search reading 5%: 0.53 under `partial` | depends strongly on labelling accuracy |
 | control: counts, proportions, rankings, wide comparisons | 1,838 | sampling 5%: about 0.8 under `relative` | about 0.9 or more |
 
@@ -209,14 +230,13 @@ label of each record they read. Full setup and examples:
   of reading. Under `relative`, a reader that opens nothing and divides the
   record count by the number of labels already scores 0.43 to 0.55 on counts,
   and longer documents are not harder.
-- Core questions come in two groups. The **244 from core documents**
+- Core questions come in two groups. The **214 from core documents**
   (`core_document: true`) are fully balanced: picking the first-named label
-  gets 0.50, the label more common in the source dataset 0.49, topic search
-  0.49, sampling half the document 0.64. The **174 from ordinary documents**
-  (v0.8.0) are slightly exposed: the label more common in the source dataset
-  wins 61% of the time, and on the paired intent sets, where the compared
-  categories hold about 60 records, topic search reaches 0.74 to 0.79. Report
-  the two groups separately.
+  gets 0.50, the label more common in the source dataset 0.50, topic search
+  0.49, sampling half the document 0.62. The **95 from ordinary documents** are
+  slightly exposed: the label more common in the source dataset wins 62% of the
+  time and topic search reaches 0.67, because the compared categories are
+  sometimes small. Report the two groups separately.
 - Core scores have a ceiling below 1.0: wrong labels in the source data decide
   some close comparisons. With 5% of labels wrong a perfect model scores about
   0.85. A native-speaker check of the Turkish core categories is prepared.
@@ -236,7 +256,7 @@ as the unit of evidence.
 
 | subset | missing types | why |
 |---|---|---|
-| `interpress_tr`, `sinema_tr` | `most_common`, `least_common`, `second_most` | section shares and adjacent ratings are too close to rank with the required 15% margin |
+| `interpress_tr` | `most_common`, `least_common`, `second_most` | section shares are too close to rank with the required 15% margin |
 | `vitamins_tr`, `musteri_tr`, `amazon_hpc_en`, `marc_en` | close comparisons on random-mix documents | with "neutral" excluded, the two remaining labels are rarely close; these sets' core questions come from core documents |
 
 **One flagged tier.** On `en_intent` at 100K tokens, a reader using only the

README.md

@@ -71,27 +71,27 @@ them. Every answer is computed from the
 source dataset's own labels, twice, by independent code. The construction
 follows [OOLONG](https://arxiv.org/abs/2511.02817) (Bertsch et al., 2025).
 
-Built with [`tr-oolong`](https://github.com/yigitates17/tr-oolong) v0.9.0. Developed at the Institute for
+Built with [`tr-oolong`](https://github.com/yigitates17/tr-oolong) v0.10.0. Developed at the Institute for
 Data Science & Artificial Intelligence (DSAI), Boğaziçi University, as MSc
 thesis work. Full details: `DATACARD.md` in this repository.
 
 ## At a glance
 
-**11 subsets · 353 documents · 2,521 questions · 106.4M tokens · 7 question types**
+**11 subsets · 347 documents · 2,362 questions · 102.5M tokens · 7 question types**
 
 | subset | lang | classes | docs | questions | shortest | longest | max records in one doc |
 |---|---|---:|---:|---:|---:|---:|---:|
-| `interpress_tr` | tr | **16** | 43 | 345 | 99,267 | 1,045,425 | 2,890 |
-| `sikayet_tr` | tr | **29** | 49 | 381 | 99,861 | 1,017,683 | 10,000 |
-| `sinema_tr` | tr | **10** | 44 | 290 | 99,179 | 1,000,973 | 8,000 |
-| `amazon_hpc_en` | en | 3 | 43 | 240 | 98,559 | 987,623 | 16,116 |
-| `marc_en` | en | 3 | 38 | 212 | 98,232 | 536,428 | 12,000 |
-| `vitamins_tr` | tr | 3 | 38 | 187 | 99,204 | 496,526 | 12,898 |
-| `musteri_tr` | tr | 3 | 38 | 217 | 91,242 | 496,238 | 13,618 |
-| `tr_intent` | tr | **48** | 10 | 151 | 49,981 | 99,998 | 6,169 |
-| `en_intent` | en | **48** | 10 | 148 | 49,921 | 99,871 | 8,122 |
-| `tr_intent_paired` | tr | **48** | 20 | 175 | 47,598 | 99,057 | 6,000 |
-| `en_intent_paired` | en | **48** | 20 | 175 | 36,224 | 75,187 | 6,000 |
+| `sikayet_tr` | tr | **29** | 49 | 366 | 99,861 | 1,016,268 | 10,000 |
+| `interpress_tr` | tr | **16** | 43 | 337 | 99,091 | 998,305 | 2,941 |
+| `amazon_hpc_en` | en | 3 | 43 | 237 | 98,426 | 987,208 | 17,087 |
+| `sinema_tr` | tr | 3 | 38 | 212 | 99,176 | 720,375 | 6,000 |
+| `marc_en` | en | 3 | 38 | 216 | 98,244 | 522,220 | 12,000 |
+| `vitamins_tr` | tr | 3 | 38 | 187 | 99,223 | 496,474 | 12,946 |
+| `musteri_tr` | tr | 3 | 38 | 213 | 88,854 | 496,144 | 13,825 |
+| `tr_intent` | tr | **48** | 10 | 139 | 49,981 | 99,998 | 6,169 |
+| `en_intent` | en | **48** | 10 | 133 | 49,921 | 99,871 | 8,122 |
+| `tr_intent_paired` | tr | **48** | 20 | 161 | 47,629 | 99,057 | 6,000 |
+| `en_intent_paired` | en | **48** | 20 | 161 | 36,250 | 75,187 | 6,000 |
 
 Lengths are tokens under `Qwen/Qwen3-8B`. `classes` is the number of labels.
 
@@ -99,19 +99,19 @@ Lengths are tokens under `Qwen/Qwen3-8B`. `classes` is the number of labels.
 
 | role | question types | questions | what it shows |
 |---|---|---:|---|
-| `core` | `close_comparison`: which are there more of, A or B? (both frequent, counts very close) | 418 | that the model read and judged the whole document |
-| `retrieval` | `count` with `"rare": true` (answer 5 to 30) | 265 | that it can find a few records by meaning |
-| `control` | `count` 656, `proportion` 492, `label_vs_label` 292, `most_common` 138, `least_common` 132, `second_most` 128 | 1,838 | that it can classify the records at all |
+| `core` | `close_comparison`: which are there more of, A or B? (both frequent, told apart from the text, counts very close) | 309 | that the model read and judged the whole document |
+| `retrieval` | `count` with `"rare": true` (answer 5 to 30) | 237 | that it can find a few records by meaning |
+| `control` | `count` 587, `proportion` 483, `label_vs_label` 299, `most_common` 156, `least_common` 147, `second_most` 144 | 1,816 | that it can classify the records at all |
 
 **Turkish/English pairs.** Compare languages within a pair, never by totals
 (three subsets are Turkish only). `tr_intent_paired` and `en_intent_paired`
 contain the same utterances (MASSIVE is a human translation) in the same order,
-so all 175 questions, 55 of them core, are identical in both languages.
+so all 161 questions, 41 of them core, are identical in both languages.
 `tr_intent`/`en_intent` match on token budget instead. `musteri_tr`/`marc_en`
 and `vitamins_tr`/`amazon_hpc_en` are different corpora with the same task;
 their core documents are built with identical sizes and identical
 positive/negative counts, so their 18 core questions each are identical too.
-In total 91 core questions are identical across the two languages.
+In total 77 core questions are identical across the two languages.
 `sikayet_tr`, `interpress_tr` and `sinema_tr` are Turkish only.
 
 ### Why the Turkish intent questions use English label names
@@ -136,7 +136,7 @@ text being classified; the label only names the bucket.
 | questions over dates | yes, its hardest group | none yet (`interpress_tr` has dates) |
 | same question, same answer in two languages | no | yes (`*_intent_paired`) |
 | numeric score | `partial` (0.75 per unit of error) | `partial`, plus `relative` for large answers |
-| questions that resist sampling and search | none found | 418 close comparisons |
+| questions that resist sampling and search | none found | 309 close comparisons |
 | published shortcut checks | none | yes, in the GitHub repository |
 
 OOLONG's construction code was not released; this is an independent
@@ -179,17 +179,17 @@ Each subset keeps the licence of its source.
 
 | subset | lang | source | license | text included | questions | note |
 |---|---|---|---|---|---|---|
-| `tr_intent` | tr | AmazonScience/massive (tr-TR) | `cc-by-4.0` | yes | 151 | MASSIVE is CC-BY-4.0: redistribution of derived data is permitted with attribution and a statement of changes. |
-| `en_intent` | en | AmazonScience/massive (en-US) | `cc-by-4.0` | yes | 148 | As above; this is the parallel English twin. |
-| `tr_intent_paired` | tr | AmazonScience/massive (tr-TR), record-matched | `cc-by-4.0` | yes | 175 | Record-matched with en_intent_paired: the same utterances in the same order, so all 175 questions (55 core) have the same answer in both languages (word answers via answer_key). |
-| `en_intent_paired` | en | AmazonScience/massive (en-US), record-matched | `cc-by-4.0` | yes | 175 | The English half of the record-matched pair. Sized in records, not tokens, so its documents are shorter in tokens than the Turkish half (1.34x under the Qwen3-8B tokenizer). |
+| `tr_intent` | tr | AmazonScience/massive (tr-TR) | `cc-by-4.0` | yes | 139 | MASSIVE is CC-BY-4.0: redistribution of derived data is permitted with attribution and a statement of changes. |
+| `en_intent` | en | AmazonScience/massive (en-US) | `cc-by-4.0` | yes | 133 | As above; this is the parallel English twin. |
+| `tr_intent_paired` | tr | AmazonScience/massive (tr-TR), record-matched | `cc-by-4.0` | yes | 161 | Record-matched with en_intent_paired: the same utterances in the same order, so all 161 questions (41 core) have the same answer in both languages (word answers via answer_key). |
+| `en_intent_paired` | en | AmazonScience/massive (en-US), record-matched | `cc-by-4.0` | yes | 161 | The English half of the record-matched pair. Sized in records, not tokens, so its documents are shorter in tokens than the Turkish half (1.34x under the Qwen3-8B tokenizer). |
 | `vitamins_tr` | tr | turkish-nlp-suite/vitamins-supplements-reviews (Vitaminler.com) | `cc-by-sa-4.0` | yes | 187 | CC-BY-SA-4.0 is SHARE-ALIKE: this subset and anything derived from it must stay CC-BY-SA-4.0. Cite Altinok (ACL 2023). |
-| `musteri_tr` | tr | turkish-nlp-suite/MusteriYorumlari (Hepsiburada, Trendyol) | `cc-by-sa-4.0` | yes | 217 | CC-BY-SA-4.0 is SHARE-ALIKE: this subset and anything derived from it must stay CC-BY-SA-4.0. Labels are the customer's own 1-5 star rating. |
-| `marc_en` | en | SetFit/amazon_reviews_multi_en (Multilingual Amazon Reviews Corpus) | `apache-2.0` | yes | 212 | Apache-2.0: redistribution permitted. The English half of the cleanest pair; labels are the reviewer's own 1-5 star rating. |
-| `amazon_hpc_en` | en | McAuley-Lab/Amazon-Reviews-2023 (Health_and_Personal_Care) | `other` | **no** | 240 | Review text is governed by Amazon's Conditions of Use, not by the repository's license. Text withheld; rebuild locally with scripts/health.py + configs/amazon_hpc_en.json. |
-| `sikayet_tr` | tr | Kaggle savasy/multiclass-classification-data-for-turkish-tc32 | `other` | **no** | 381 | 29 categories of Turkish consumer complaints. The uploader declares NO license and the text is scraped from a complaints site, so the text is withheld. Rebuild locally with scripts/sikayet_tr.py, which takes the path to a copy of ticaret-yorum.csv downloaded from Kaggle. Three of the original 32 categories were dropped because their names appear in over 30% of their complaints. |
-| `interpress_tr` | tr | Interpress Turkish news category corpus, 270k | `other` | **no** | 345 | 16 sections of Turkish news (17 in the source), with daily publication dates. The upstream card declares no license. Text withheld; rebuild locally with scripts/interpress_tr.py, which fetches the archive the Hugging Face loading script points at and verifies its sha256. The Apache header on that loading script covers the SCRIPT, not the data, and must not be cited as the data's license. |
-| `sinema_tr` | tr | turkish-nlp-suite/BuyukSinema | `cc-by-sa-4.0` | yes | 290 | Turkish film reviews labelled with the reviewer's own 10-point rating. The only large-label-space Turkish source found with a declared license, so unlike the other two v0.7.0 additions its text ships. Share-alike: anything derived from this subset stays CC-BY-SA-4.0. |
+| `musteri_tr` | tr | turkish-nlp-suite/MusteriYorumlari (Hepsiburada, Trendyol) | `cc-by-sa-4.0` | yes | 213 | CC-BY-SA-4.0 is SHARE-ALIKE: this subset and anything derived from it must stay CC-BY-SA-4.0. Labels are the customer's own 1-5 star rating. |
+| `marc_en` | en | SetFit/amazon_reviews_multi_en (Multilingual Amazon Reviews Corpus) | `apache-2.0` | yes | 216 | Apache-2.0: redistribution permitted. The English half of the cleanest pair; labels are the reviewer's own 1-5 star rating. |
+| `amazon_hpc_en` | en | McAuley-Lab/Amazon-Reviews-2023 (Health_and_Personal_Care) | `other` | **no** | 237 | Review text is governed by Amazon's Conditions of Use, not by the repository's license. Text withheld; rebuild locally with scripts/health.py + configs/amazon_hpc_en.json. |
+| `sikayet_tr` | tr | Kaggle savasy/multiclass-classification-data-for-turkish-tc32 | `other` | **no** | 366 | 29 categories of Turkish consumer complaints. The uploader declares NO license and the text is scraped from a complaints site, so the text is withheld. Rebuild locally with scripts/sikayet_tr.py, which takes the path to a copy of ticaret-yorum.csv downloaded from Kaggle. Three of the original 32 categories were dropped because their names appear in over 30% of their complaints. |
+| `interpress_tr` | tr | Interpress Turkish news category corpus, 270k | `other` | **no** | 337 | 16 sections of Turkish news (17 in the source), with daily publication dates. The upstream card declares no license. Text withheld; rebuild locally with scripts/interpress_tr.py, which fetches the archive the Hugging Face loading script points at and verifies its sha256. The Apache header on that loading script covers the SCRIPT, not the data, and must not be cited as the data's license. |
+| `sinema_tr` | tr | turkish-nlp-suite/BuyukSinema | `cc-by-sa-4.0` | yes | 212 | Turkish film reviews; sentiment from the reviewer's own 10-point rating (1-4 negative, 5-6 neutral, 7-10 positive; exact ratings cannot be read from the text). Share-alike: anything derived from this subset stays CC-BY-SA-4.0. |
 
 ### Subsets without text
 
@@ -214,7 +214,7 @@ they read, so they show what a reading strategy can achieve.
 
 | role | strongest shortcut found | reader of everything |
 |---|---|---|
-| core | sampling half the document 0.63; the 10% most relevant records by topic search 0.56; guessing 0.50 | 0.85 at 95% labelling accuracy, 0.76 at 90% |
+| core | sampling half the document 0.62; the 10% most relevant records by topic search 0.54; guessing 0.50 | 0.86 at 95% labelling accuracy, 0.76 at 90% |
 | retrieval | topic search reading 5%: 0.53 under `partial` | depends strongly on labelling accuracy |
 | control | sampling 5%: about 0.8 under `relative` | about 0.9 or more |
 
@@ -225,13 +225,16 @@ they read, so they show what a reading strategy can achieve.
 - Control questions are not evidence of reading: a 5% sample answers them
   almost as well as the whole document, and under `relative` a longer document
   is not harder.
-- 418 questions are core (margin of error about ±0.05 on a model's core score).
-  The 244 from core documents (`core_document: true`), built from exact label
-  counts with which label is larger and which is named first balanced by
-  design, are the cleaner group: first-named 0.50, source-dataset shares 0.49,
-  topic search 0.49. The 174 from ordinary documents are slightly exposed
-  (source-dataset shares 0.61; topic search up to 0.79 on the paired intent
-  sets). Report the two groups separately.
+- 309 questions are core (margin of error about ±0.06 on a model's core score).
+  They only compare label pairs that can be told apart from the text (lists in
+  the GitHub repository). The 214 from core documents (`core_document: true`),
+  built from exact label counts with which label is larger and which is named
+  first balanced by design, are the cleaner group: first-named 0.50,
+  source-dataset shares 0.50, topic search 0.49. The 95 from ordinary
+  documents are slightly exposed (source-dataset shares 0.62, topic search
+  0.67). Report the two groups separately.
+- Cleaning (v0.10.0): newspaper mastheads, reviews that write their score
+  ("7/10", "5 stars") and HTML are removed.
 - Wrong labels in the source data decide some close comparisons, so even a
   perfect model scores below 1.0 on core questions (about 0.85 with 5% of
   labels wrong).

amazon_hpc_en/manifest.json

@@ -1,6 +1,6 @@
 {
-  "version": "0.9.0",
-  "date": "2026-09-29",
+  "version": "0.10.0",
+  "date": "2026-10-01",
   "environment": {
     "python": "3.12.9",
     "polars": "1.42.0",
@@ -17,6 +17,15 @@
     "drop_label_leakage": true,
     "leak_label_words": false,
     "min_class_support": 0,
+    "boilerplate_keywords": [],
+    "boilerplate_min_keywords": 3,
+    "boilerplate_weak_min": 2,
+    "boilerplate_regex": "",
+    "drop_text_regex": [
+      "\\b\\d{1,3}([.,]\\d)?\\s*/\\s*(10|5|100)\\b",
+      "\\b(\\d|zero|one|two|three|four|five)[\\s-]+stars?\\b"
+    ],
+    "strip_html": true,
     "seed": 42,
     "haystack_target_tokens": [
       100000,
@@ -76,6 +85,7 @@
       "neutral"
     ],
     "close_ordinal_gap": 0,
+    "close_allowed_pairs_file": "experiments/separable_pairs/amazon_hpc_en.json",
     "core_doc_records": [
       2500,
       6000,
@@ -113,63 +123,64 @@
     "out_dir": "amazon_hpc_en_out",
     "dataset": ""
   },
-  "source_hash_first1000": "d181a0e1a6f5d612",
-  "rows_after_cleaning": 55352,
+  "source_hash_first1000": "c81b07217a148b11",
+  "rows_after_cleaning": 53930,
   "cleaning": {
-    "rows_dropped_label_leakage": 557,
-    "label_leakage_rate": 0.009963,
+    "rows_dropped_text_regex": 1488,
+    "rows_dropped_label_leakage": 511,
+    "label_leakage_rate": 0.009386,
     "classes_dropped_low_support": [],
     "min_class_support": 0
   },
   "ranking_feasibility": {
     "n_classes": 3,
-    "min_class_pool": 17774,
-    "max_records": 53322,
+    "min_class_pool": 17433,
+    "max_records": 52299,
     "binding": true,
-    "ranking_length_ceiling_tokens": 2889092
+    "ranking_length_ceiling_tokens": 2815987
   },
   "tier_overlap": {
     "100000": {
       "n_haystacks": 10,
-      "mean_jaccard": 0.0153,
-      "max_jaccard": 0.0277,
-      "pool_fraction_per_haystack": 0.0295
+      "mean_jaccard": 0.0164,
+      "max_jaccard": 0.0305,
+      "pool_fraction_per_haystack": 0.0314
     },
     "1000000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.1939,
-      "max_jaccard": 0.3299,
-      "pool_fraction_per_haystack": 0.2847
+      "mean_jaccard": 0.2071,
+      "max_jaccard": 0.3455,
+      "pool_fraction_per_haystack": 0.3049
     },
     "250000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.033,
-      "max_jaccard": 0.0498,
-      "pool_fraction_per_haystack": 0.0722
+      "mean_jaccard": 0.0352,
+      "max_jaccard": 0.0594,
+      "pool_fraction_per_haystack": 0.0767
     },
     "500000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.0721,
-      "max_jaccard": 0.0879,
-      "pool_fraction_per_haystack": 0.1441
+      "mean_jaccard": 0.0791,
+      "max_jaccard": 0.0949,
+      "pool_fraction_per_haystack": 0.1575
     },
     "core12000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.1329,
-      "max_jaccard": 0.1353,
-      "pool_fraction_per_haystack": 0.2168
+      "mean_jaccard": 0.1363,
+      "max_jaccard": 0.1386,
+      "pool_fraction_per_haystack": 0.2225
     },
     "core2500": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0259,
-      "max_jaccard": 0.0303,
-      "pool_fraction_per_haystack": 0.0452
+      "mean_jaccard": 0.0265,
+      "max_jaccard": 0.0301,
+      "pool_fraction_per_haystack": 0.0464
     },
     "core6000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0633,
-      "max_jaccard": 0.0689,
-      "pool_fraction_per_haystack": 0.1084
+      "mean_jaccard": 0.0648,
+      "max_jaccard": 0.0706,
+      "pool_fraction_per_haystack": 0.1113
     }
   },
   "max_pool_fraction": 0.35,
@@ -177,21 +188,21 @@
   "tiers_built_over_pool_cap": [
     {
       "tier": 1000000,
-      "pool_fraction": 0.3501
+      "pool_fraction": 0.3616
     }
   ],
   "label_space": 3,
   "proportion_unit": "percent",
   "top_k_k": 3,
   "tokenizer": "Qwen/Qwen3-8B",
-  "questions_written": 240,
+  "questions_written": 237,
   "kind_distribution": {
-    "close_comparison": 18,
+    "close_comparison": 19,
     "count": 69,
-    "label_vs_label": 26,
-    "least_common": 21,
-    "most_common": 23,
-    "proportion": 63,
+    "label_vs_label": 25,
+    "least_common": 22,
+    "most_common": 22,
+    "proportion": 60,
     "second_most": 20
   },
   "rare_count_questions": 0,
@@ -202,234 +213,234 @@
   "haystacks": [
     {
       "haystack_id": "en-100000-0",
-      "n_examples": 1622,
-      "n_tokens": 98707,
-      "n_chars": 405814,
-      "drift_target": "neutral",
+      "n_examples": 1728,
+      "n_tokens": 98668,
+      "n_chars": 407916,
+      "drift_target": "negative",
       "drift_ok": true,
-      "n_questions": 11
+      "n_questions": 10
     },
     {
       "haystack_id": "en-100000-1",
-      "n_examples": 1685,
-      "n_tokens": 98726,
-      "n_chars": 405059,
-      "drift_target": "neutral",
+      "n_examples": 1702,
+      "n_tokens": 98426,
+      "n_chars": 407012,
+      "drift_target": "positive",
       "drift_ok": true,
-      "n_questions": 10
+      "n_questions": 9
     },
     {
       "haystack_id": "en-100000-2",
-      "n_examples": 1637,
-      "n_tokens": 98722,
-      "n_chars": 407683,
+      "n_examples": 1662,
+      "n_tokens": 98695,
+      "n_chars": 409244,
       "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 10
     },
     {
       "haystack_id": "en-100000-3",
-      "n_examples": 1630,
-      "n_tokens": 98672,
-      "n_chars": 408405,
+      "n_examples": 1734,
+      "n_tokens": 98648,
+      "n_chars": 407516,
       "drift_target": "neutral",
       "drift_ok": true,
       "n_questions": 10
     },
     {
       "haystack_id": "en-100000-4",
-      "n_examples": 1663,
-      "n_tokens": 98676,
-      "n_chars": 406543,
+      "n_examples": 1679,
+      "n_tokens": 98683,
+      "n_chars": 408325,
       "drift_target": "negative",
       "drift_ok": true,
-      "n_questions": 8
+      "n_questions": 9
     },
     {
       "haystack_id": "en-100000-5",
-      "n_examples": 1649,
-      "n_tokens": 98694,
-      "n_chars": 406610,
+      "n_examples": 1661,
+      "n_tokens": 98696,
+      "n_chars": 407686,
       "drift_target": "negative",
       "drift_ok": true,
       "n_questions": 7
     },
     {
       "haystack_id": "en-100000-6",
-      "n_examples": 1571,
-      "n_tokens": 98780,
-      "n_chars": 408070,
-      "drift_target": "positive",
+      "n_examples": 1702,
+      "n_tokens": 98634,
+      "n_chars": 409296,
+      "drift_target": "neutral",
       "drift_ok": true,
       "n_questions": 10
     },
     {
       "haystack_id": "en-100000-7",
-      "n_examples": 1680,
-      "n_tokens": 98679,
-      "n_chars": 405081,
-      "drift_target": "positive",
+      "n_examples": 1664,
+      "n_tokens": 98537,
+      "n_chars": 407879,
+      "drift_target": "neutral",
       "drift_ok": true,
-      "n_questions": 10
+      "n_questions": 9
     },
     {
       "haystack_id": "en-100000-8",
-      "n_examples": 1596,
-      "n_tokens": 98618,
-      "n_chars": 408600,
+      "n_examples": 1663,
+      "n_tokens": 98688,
+      "n_chars": 408818,
       "drift_target": "negative",
       "drift_ok": true,
       "n_questions": 10
     },
     {
       "haystack_id": "en-100000-9",
-      "n_examples": 1580,
-      "n_tokens": 98559,
-      "n_chars": 407320,
+      "n_examples": 1732,
+      "n_tokens": 98620,
+      "n_chars": 407021,
       "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "en-250000-0",
-      "n_examples": 4045,
-      "n_tokens": 246871,
-      "n_chars": 1017000,
-      "drift_target": "positive",
+      "n_examples": 4268,
+      "n_tokens": 246620,
+      "n_chars": 1020385,
+      "drift_target": "negative",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "en-250000-1",
-      "n_examples": 4163,
-      "n_tokens": 246749,
-      "n_chars": 1010901,
+      "n_examples": 4137,
+      "n_tokens": 246797,
+      "n_chars": 1022546,
       "drift_target": "neutral",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "en-250000-2",
-      "n_examples": 3933,
-      "n_tokens": 246839,
-      "n_chars": 1022161,
-      "drift_target": "positive",
+      "n_examples": 4070,
+      "n_tokens": 246780,
+      "n_chars": 1024146,
+      "drift_target": "neutral",
       "drift_ok": true,
-      "n_questions": 10
+      "n_questions": 9
     },
     {
       "haystack_id": "en-250000-3",
-      "n_examples": 3937,
-      "n_tokens": 246881,
-      "n_chars": 1021066,
-      "drift_target": "neutral",
+      "n_examples": 4155,
+      "n_tokens": 246722,
+      "n_chars": 1024544,
+      "drift_target": "negative",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "en-250000-4",
-      "n_examples": 3891,
-      "n_tokens": 246867,
-      "n_chars": 1024978,
-      "drift_target": "negative",
+      "n_examples": 4059,
+      "n_tokens": 246725,
+      "n_chars": 1029456,
+      "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "en-500000-0",
-      "n_examples": 8020,
-      "n_tokens": 493665,
-      "n_chars": 2040041,
-      "drift_target": "negative",
+      "n_examples": 8390,
+      "n_tokens": 493387,
+      "n_chars": 2045447,
+      "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "en-500000-1",
-      "n_examples": 7969,
-      "n_tokens": 493671,
-      "n_chars": 2030753,
-      "drift_target": "positive",
+      "n_examples": 8590,
+      "n_tokens": 493331,
+      "n_chars": 2039417,
+      "drift_target": "negative",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "en-500000-2",
-      "n_examples": 7679,
-      "n_tokens": 493862,
-      "n_chars": 2047190,
+      "n_examples": 8393,
+      "n_tokens": 493362,
+      "n_chars": 2055964,
       "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "en-500000-3",
-      "n_examples": 7876,
-      "n_tokens": 493716,
-      "n_chars": 2039828,
-      "drift_target": "positive",
+      "n_examples": 8474,
+      "n_tokens": 493413,
+      "n_chars": 2043385,
+      "drift_target": "negative",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "en-500000-4",
-      "n_examples": 8331,
-      "n_tokens": 493458,
-      "n_chars": 2026688,
+      "n_examples": 8620,
+      "n_tokens": 493325,
+      "n_chars": 2036511,
       "drift_target": "neutral",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "en-1000000-0",
-      "n_examples": 16036,
-      "n_tokens": 987423,
-      "n_chars": 4080221,
+      "n_examples": 16460,
+      "n_tokens": 987012,
+      "n_chars": 4101522,
       "drift_target": "negative",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "en-1000000-1",
-      "n_examples": 15616,
-      "n_tokens": 987618,
-      "n_chars": 4106313,
-      "drift_target": "positive",
+      "n_examples": 16282,
+      "n_tokens": 987029,
+      "n_chars": 4124959,
+      "drift_target": "negative",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "en-1000000-2",
-      "n_examples": 16116,
-      "n_tokens": 987513,
-      "n_chars": 4065892,
+      "n_examples": 17087,
+      "n_tokens": 986750,
+      "n_chars": 4080688,
       "drift_target": "neutral",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "en-1000000-3",
-      "n_examples": 15485,
-      "n_tokens": 987623,
-      "n_chars": 4100384,
-      "drift_target": "positive",
+      "n_examples": 16163,
+      "n_tokens": 987208,
+      "n_chars": 4114037,
+      "drift_target": "neutral",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "en-1000000-4",
-      "n_examples": 15551,
-      "n_tokens": 987580,
-      "n_chars": 4101762,
-      "drift_target": "positive",
+      "n_examples": 16217,
+      "n_tokens": 987137,
+      "n_chars": 4121588,
+      "drift_target": "neutral",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "en-core2500-0",
       "n_examples": 2500,
-      "n_tokens": 153287,
-      "n_chars": 632563,
+      "n_tokens": 152087,
+      "n_chars": 632739,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -443,8 +454,8 @@
     {
       "haystack_id": "en-core2500-1",
       "n_examples": 2500,
-      "n_tokens": 154387,
-      "n_chars": 638087,
+      "n_tokens": 144811,
+      "n_chars": 599961,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -458,8 +469,8 @@
     {
       "haystack_id": "en-core2500-2",
       "n_examples": 2500,
-      "n_tokens": 153047,
-      "n_chars": 630599,
+      "n_tokens": 149147,
+      "n_chars": 620176,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -473,8 +484,8 @@
     {
       "haystack_id": "en-core2500-3",
       "n_examples": 2500,
-      "n_tokens": 161139,
-      "n_chars": 666084,
+      "n_tokens": 144373,
+      "n_chars": 598492,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -488,8 +499,8 @@
     {
       "haystack_id": "en-core2500-4",
       "n_examples": 2500,
-      "n_tokens": 157431,
-      "n_chars": 653787,
+      "n_tokens": 147247,
+      "n_chars": 613027,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -503,8 +514,8 @@
     {
       "haystack_id": "en-core2500-5",
       "n_examples": 2500,
-      "n_tokens": 156141,
-      "n_chars": 645975,
+      "n_tokens": 147832,
+      "n_chars": 614787,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -518,8 +529,8 @@
     {
       "haystack_id": "en-core6000-0",
       "n_examples": 6000,
-      "n_tokens": 380154,
-      "n_chars": 1573308,
+      "n_tokens": 357689,
+      "n_chars": 1485396,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -533,8 +544,8 @@
     {
       "haystack_id": "en-core6000-1",
       "n_examples": 6000,
-      "n_tokens": 369027,
-      "n_chars": 1523889,
+      "n_tokens": 350745,
+      "n_chars": 1454994,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -548,8 +559,8 @@
     {
       "haystack_id": "en-core6000-2",
       "n_examples": 6000,
-      "n_tokens": 373791,
-      "n_chars": 1542181,
+      "n_tokens": 353804,
+      "n_chars": 1471236,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -563,8 +574,8 @@
     {
       "haystack_id": "en-core6000-3",
       "n_examples": 6000,
-      "n_tokens": 379845,
-      "n_chars": 1568638,
+      "n_tokens": 346817,
+      "n_chars": 1438439,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -578,8 +589,8 @@
     {
       "haystack_id": "en-core6000-4",
       "n_examples": 6000,
-      "n_tokens": 375033,
-      "n_chars": 1552170,
+      "n_tokens": 362289,
+      "n_chars": 1504659,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -593,8 +604,8 @@
     {
       "haystack_id": "en-core6000-5",
       "n_examples": 6000,
-      "n_tokens": 375087,
-      "n_chars": 1547697,
+      "n_tokens": 354489,
+      "n_chars": 1472688,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -608,8 +619,8 @@
     {
       "haystack_id": "en-core12000-0",
       "n_examples": 12000,
-      "n_tokens": 748879,
-      "n_chars": 3098197,
+      "n_tokens": 709067,
+      "n_chars": 2943838,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -623,8 +634,8 @@
     {
       "haystack_id": "en-core12000-1",
       "n_examples": 12000,
-      "n_tokens": 749542,
-      "n_chars": 3095570,
+      "n_tokens": 704723,
+      "n_chars": 2926690,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -638,8 +649,8 @@
     {
       "haystack_id": "en-core12000-2",
       "n_examples": 12000,
-      "n_tokens": 751215,
-      "n_chars": 3103103,
+      "n_tokens": 711782,
+      "n_chars": 2955536,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -653,8 +664,8 @@
     {
       "haystack_id": "en-core12000-3",
       "n_examples": 12000,
-      "n_tokens": 736030,
-      "n_chars": 3039015,
+      "n_tokens": 714844,
+      "n_chars": 2964008,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -668,8 +679,8 @@
     {
       "haystack_id": "en-core12000-4",
       "n_examples": 12000,
-      "n_tokens": 752625,
-      "n_chars": 3110742,
+      "n_tokens": 719402,
+      "n_chars": 2991820,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -683,8 +694,8 @@
     {
       "haystack_id": "en-core12000-5",
       "n_examples": 12000,
-      "n_tokens": 745171,
-      "n_chars": 3080164,
+      "n_tokens": 717933,
+      "n_chars": 2982111,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,

en_intent/manifest.json

@@ -1,6 +1,6 @@
 {
-  "version": "0.9.0",
-  "date": "2026-09-28",
+  "version": "0.10.0",
+  "date": "2026-10-01",
   "environment": {
     "python": "3.12.9",
     "polars": "1.42.0",
@@ -17,6 +17,12 @@
     "drop_label_leakage": true,
     "leak_label_words": false,
     "min_class_support": 0,
+    "boilerplate_keywords": [],
+    "boilerplate_min_keywords": 3,
+    "boilerplate_weak_min": 2,
+    "boilerplate_regex": "",
+    "drop_text_regex": [],
+    "strip_html": false,
     "seed": 42,
     "haystack_target_tokens": [
       50000,
@@ -62,12 +68,14 @@
     "close_min_diff": 2,
     "close_exclude_labels": [],
     "close_ordinal_gap": 0,
+    "close_allowed_pairs_file": "experiments/separable_pairs/intent_shared.json",
     "core_doc_records": [],
     "core_docs_per_tier": 0,
     "core_share": 0.3,
     "core_pairs": [],
     "core_pairs_per_doc": 1,
     "core_seed_key": "",
+    "core_min_diff": 5,
     "entity_band_tol": 0.05,
     "label_band_tol": 0.15,
     "entity_jitter_sigma": 1.0,
@@ -121,9 +129,9 @@
   "proportion_unit": "per_mille",
   "top_k_k": 3,
   "tokenizer": "Qwen/Qwen3-8B",
-  "questions_written": 148,
+  "questions_written": 133,
   "kind_distribution": {
-    "close_comparison": 28,
+    "close_comparison": 13,
     "count": 60,
     "label_vs_label": 10,
     "least_common": 10,
@@ -144,7 +152,7 @@
       "n_chars": 191324,
       "drift_target": "qa_definition",
       "drift_ok": true,
-      "n_questions": 15
+      "n_questions": 14
     },
     {
       "haystack_id": "en-50000-1",
@@ -153,7 +161,7 @@
       "n_chars": 194364,
       "drift_target": "calendar_remove",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 13
     },
     {
       "haystack_id": "en-50000-2",
@@ -162,7 +170,7 @@
       "n_chars": 193634,
       "drift_target": "takeaway_query",
       "drift_ok": true,
-      "n_questions": 14
+      "n_questions": 13
     },
     {
       "haystack_id": "en-50000-3",
@@ -171,7 +179,7 @@
       "n_chars": 193124,
       "drift_target": "alarm_query",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 14
     },
     {
       "haystack_id": "en-50000-4",
@@ -180,7 +188,7 @@
       "n_chars": 193170,
       "drift_target": "general_quirky",
       "drift_ok": true,
-      "n_questions": 14
+      "n_questions": 12
     },
     {
       "haystack_id": "en-100000-0",
@@ -198,7 +206,7 @@
       "n_chars": 387072,
       "drift_target": "recommendation_events",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 14
     },
     {
       "haystack_id": "en-100000-2",
@@ -207,7 +215,7 @@
       "n_chars": 383372,
       "drift_target": "qa_factoid",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 15
     },
     {
       "haystack_id": "en-100000-3",
@@ -216,7 +224,7 @@
       "n_chars": 387492,
       "drift_target": "calendar_set",
       "drift_ok": true,
-      "n_questions": 13
+      "n_questions": 12
     },
     {
       "haystack_id": "en-100000-4",
@@ -225,7 +233,7 @@
       "n_chars": 385331,
       "drift_target": "cooking_recipe",
       "drift_ok": true,
-      "n_questions": 15
+      "n_questions": 13
     }
   ]
 }

en_intent/questions.jsonl

@@ -10,9 +10,8 @@
 {"uid": "en_intent:en-50000-0-q9", "dataset": "en_intent", "id": "en-50000-0-q9", "haystack_uid": "en_intent:en-50000-0", "haystack_id": "en-50000-0", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "weather_query", "answer": "27", "question": "How many utterances have the intent 'weather_query'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-50000-0-q10", "dataset": "en_intent", "id": "en-50000-0-q10", "haystack_uid": "en_intent:en-50000-0", "haystack_id": "en-50000-0", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "recommendation_events", "answer": "18", "question": "How many utterances have the intent 'recommendation_events'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-50000-0-q11", "dataset": "en_intent", "id": "en-50000-0-q11", "haystack_uid": "en_intent:en-50000-0", "haystack_id": "en-50000-0", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "calendar_set", "answer": "25", "question": "How many utterances have the intent 'calendar_set'? Answer with the number only.", "role": "retrieval"}
-{"uid": "en_intent:en-50000-0-q12", "dataset": "en_intent", "id": "en-50000-0-q12", "haystack_uid": "en_intent:en-50000-0", "haystack_id": "en-50000-0", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "lists_remove", "label_b": "lists_createoradd", "candidates": ["lists_remove", "lists_createoradd"], "answer": "lists_createoradd", "answer_key": "lists_createoradd", "question": "Which are there more of in these records: records labeled 'lists_remove' or records labeled 'lists_createoradd'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-50000-0-q13", "dataset": "en_intent", "id": "en-50000-0-q13", "haystack_uid": "en_intent:en-50000-0", "haystack_id": "en-50000-0", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "play_podcasts", "candidates": ["qa_factoid", "play_podcasts"], "answer": "qa_factoid", "answer_key": "qa_factoid", "question": "Which are there more of in these records: records labeled 'qa_factoid' or records labeled 'play_podcasts'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-50000-0-q14", "dataset": "en_intent", "id": "en-50000-0-q14", "haystack_uid": "en_intent:en-50000-0", "haystack_id": "en-50000-0", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "email_querycontact", "label_b": "takeaway_query", "candidates": ["email_querycontact", "takeaway_query"], "answer": "takeaway_query", "answer_key": "takeaway_query", "question": "Which are there more of in these records: records labeled 'email_querycontact' or records labeled 'takeaway_query'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent:en-50000-0-q12", "dataset": "en_intent", "id": "en-50000-0-q12", "haystack_uid": "en_intent:en-50000-0", "haystack_id": "en-50000-0", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "play_podcasts", "candidates": ["qa_factoid", "play_podcasts"], "answer": "qa_factoid", "answer_key": "qa_factoid", "question": "Which are there more of in these records: records labeled 'qa_factoid' or records labeled 'play_podcasts'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent:en-50000-0-q13", "dataset": "en_intent", "id": "en-50000-0-q13", "haystack_uid": "en_intent:en-50000-0", "haystack_id": "en-50000-0", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "lists_createoradd", "label_b": "lists_remove", "candidates": ["lists_createoradd", "lists_remove"], "answer": "lists_createoradd", "answer_key": "lists_createoradd", "question": "Which are there more of in these records: records labeled 'lists_createoradd' or records labeled 'lists_remove'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent:en-50000-1-q0", "dataset": "en_intent", "id": "en-50000-1-q0", "haystack_uid": "en_intent:en-50000-1", "haystack_id": "en-50000-1", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "transport_ticket", "answer": "42", "question": "How many utterances have the intent 'transport_ticket'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-50000-1-q1", "dataset": "en_intent", "id": "en-50000-1-q1", "haystack_uid": "en_intent:en-50000-1", "haystack_id": "en-50000-1", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "recommendation_movies", "answer": "62", "question": "How many utterances have the intent 'recommendation_movies'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-50000-1-q2", "dataset": "en_intent", "id": "en-50000-1-q2", "haystack_uid": "en_intent:en-50000-1", "haystack_id": "en-50000-1", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "datetime_query", "answer": "38", "question": "How many utterances have the intent 'datetime_query'? Answer with the number only.", "role": "control"}
@@ -25,10 +24,7 @@
 {"uid": "en_intent:en-50000-1-q9", "dataset": "en_intent", "id": "en-50000-1-q9", "haystack_uid": "en_intent:en-50000-1", "haystack_id": "en-50000-1", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "qa_factoid", "answer": "27", "question": "How many utterances have the intent 'qa_factoid'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-50000-1-q10", "dataset": "en_intent", "id": "en-50000-1-q10", "haystack_uid": "en_intent:en-50000-1", "haystack_id": "en-50000-1", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "qa_definition", "answer": "22", "question": "How many utterances have the intent 'qa_definition'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-50000-1-q11", "dataset": "en_intent", "id": "en-50000-1-q11", "haystack_uid": "en_intent:en-50000-1", "haystack_id": "en-50000-1", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "audio_volume_up", "answer": "9", "question": "How many utterances have the intent 'audio_volume_up'? Answer with the number only.", "role": "retrieval"}
-{"uid": "en_intent:en-50000-1-q12", "dataset": "en_intent", "id": "en-50000-1-q12", "haystack_uid": "en_intent:en-50000-1", "haystack_id": "en-50000-1", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "news_query", "label_b": "music_query", "candidates": ["news_query", "music_query"], "answer": "news_query", "answer_key": "news_query", "question": "Which are there more of in these records: records labeled 'news_query' or records labeled 'music_query'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-50000-1-q13", "dataset": "en_intent", "id": "en-50000-1-q13", "haystack_uid": "en_intent:en-50000-1", "haystack_id": "en-50000-1", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "calendar_set", "label_b": "recommendation_locations", "candidates": ["calendar_set", "recommendation_locations"], "answer": "calendar_set", "answer_key": "calendar_set", "question": "Which are there more of in these records: records labeled 'calendar_set' or records labeled 'recommendation_locations'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-50000-1-q14", "dataset": "en_intent", "id": "en-50000-1-q14", "haystack_uid": "en_intent:en-50000-1", "haystack_id": "en-50000-1", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "lists_remove", "label_b": "lists_createoradd", "candidates": ["lists_remove", "lists_createoradd"], "answer": "lists_createoradd", "answer_key": "lists_createoradd", "question": "Which are there more of in these records: records labeled 'lists_remove' or records labeled 'lists_createoradd'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-50000-1-q15", "dataset": "en_intent", "id": "en-50000-1-q15", "haystack_uid": "en_intent:en-50000-1", "haystack_id": "en-50000-1", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "recommendation_events", "label_b": "play_podcasts", "candidates": ["recommendation_events", "play_podcasts"], "answer": "recommendation_events", "answer_key": "recommendation_events", "question": "Which are there more of in these records: records labeled 'recommendation_events' or records labeled 'play_podcasts'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent:en-50000-1-q12", "dataset": "en_intent", "id": "en-50000-1-q12", "haystack_uid": "en_intent:en-50000-1", "haystack_id": "en-50000-1", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "lists_remove", "label_b": "lists_createoradd", "candidates": ["lists_remove", "lists_createoradd"], "answer": "lists_createoradd", "answer_key": "lists_createoradd", "question": "Which are there more of in these records: records labeled 'lists_remove' or records labeled 'lists_createoradd'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent:en-50000-2-q0", "dataset": "en_intent", "id": "en-50000-2-q0", "haystack_uid": "en_intent:en-50000-2", "haystack_id": "en-50000-2", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "general_joke", "answer": "25", "question": "How many utterances have the intent 'general_joke'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-50000-2-q1", "dataset": "en_intent", "id": "en-50000-2-q1", "haystack_uid": "en_intent:en-50000-2", "haystack_id": "en-50000-2", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "iot_hue_lightoff", "answer": "36", "question": "How many utterances have the intent 'iot_hue_lightoff'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-50000-2-q2", "dataset": "en_intent", "id": "en-50000-2-q2", "haystack_uid": "en_intent:en-50000-2", "haystack_id": "en-50000-2", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "calendar_set", "answer": "171", "question": "How many utterances have the intent 'calendar_set'? Answer with the number only.", "role": "control"}
@@ -42,7 +38,6 @@
 {"uid": "en_intent:en-50000-2-q10", "dataset": "en_intent", "id": "en-50000-2-q10", "haystack_uid": "en_intent:en-50000-2", "haystack_id": "en-50000-2", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "play_radio", "answer": "14", "question": "How many utterances have the intent 'play_radio'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-50000-2-q11", "dataset": "en_intent", "id": "en-50000-2-q11", "haystack_uid": "en_intent:en-50000-2", "haystack_id": "en-50000-2", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "email_sendemail", "answer": "28", "question": "How many utterances have the intent 'email_sendemail'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-50000-2-q12", "dataset": "en_intent", "id": "en-50000-2-q12", "haystack_uid": "en_intent:en-50000-2", "haystack_id": "en-50000-2", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_stock", "label_b": "qa_currency", "candidates": ["qa_stock", "qa_currency"], "answer": "qa_currency", "answer_key": "qa_currency", "question": "Which are there more of in these records: records labeled 'qa_stock' or records labeled 'qa_currency'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-50000-2-q13", "dataset": "en_intent", "id": "en-50000-2-q13", "haystack_uid": "en_intent:en-50000-2", "haystack_id": "en-50000-2", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "social_post", "label_b": "recommendation_events", "candidates": ["social_post", "recommendation_events"], "answer": "social_post", "answer_key": "social_post", "question": "Which are there more of in these records: records labeled 'social_post' or records labeled 'recommendation_events'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent:en-50000-3-q0", "dataset": "en_intent", "id": "en-50000-3-q0", "haystack_uid": "en_intent:en-50000-3", "haystack_id": "en-50000-3", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "recommendation_events", "answer": "63", "question": "How many utterances have the intent 'recommendation_events'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-50000-3-q1", "dataset": "en_intent", "id": "en-50000-3-q1", "haystack_uid": "en_intent:en-50000-3", "haystack_id": "en-50000-3", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "social_query", "answer": "149", "question": "How many utterances have the intent 'social_query'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-50000-3-q2", "dataset": "en_intent", "id": "en-50000-3-q2", "haystack_uid": "en_intent:en-50000-3", "haystack_id": "en-50000-3", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "email_querycontact", "answer": "65", "question": "How many utterances have the intent 'email_querycontact'? Answer with the number only.", "role": "control"}
@@ -55,10 +50,8 @@
 {"uid": "en_intent:en-50000-3-q9", "dataset": "en_intent", "id": "en-50000-3-q9", "haystack_uid": "en_intent:en-50000-3", "haystack_id": "en-50000-3", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "audio_volume_up", "answer": "24", "question": "How many utterances have the intent 'audio_volume_up'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-50000-3-q10", "dataset": "en_intent", "id": "en-50000-3-q10", "haystack_uid": "en_intent:en-50000-3", "haystack_id": "en-50000-3", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "play_music", "answer": "10", "question": "How many utterances have the intent 'play_music'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-50000-3-q11", "dataset": "en_intent", "id": "en-50000-3-q11", "haystack_uid": "en_intent:en-50000-3", "haystack_id": "en-50000-3", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "takeaway_query", "answer": "28", "question": "How many utterances have the intent 'takeaway_query'? Answer with the number only.", "role": "retrieval"}
-{"uid": "en_intent:en-50000-3-q12", "dataset": "en_intent", "id": "en-50000-3-q12", "haystack_uid": "en_intent:en-50000-3", "haystack_id": "en-50000-3", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "cooking_recipe", "label_b": "alarm_query", "candidates": ["cooking_recipe", "alarm_query"], "answer": "cooking_recipe", "answer_key": "cooking_recipe", "question": "Which are there more of in these records: records labeled 'cooking_recipe' or records labeled 'alarm_query'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-50000-3-q13", "dataset": "en_intent", "id": "en-50000-3-q13", "haystack_uid": "en_intent:en-50000-3", "haystack_id": "en-50000-3", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "transport_traffic", "label_b": "iot_coffee", "candidates": ["transport_traffic", "iot_coffee"], "answer": "transport_traffic", "answer_key": "transport_traffic", "question": "Which are there more of in these records: records labeled 'transport_traffic' or records labeled 'iot_coffee'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-50000-3-q14", "dataset": "en_intent", "id": "en-50000-3-q14", "haystack_uid": "en_intent:en-50000-3", "haystack_id": "en-50000-3", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_maths", "label_b": "qa_stock", "candidates": ["qa_maths", "qa_stock"], "answer": "qa_maths", "answer_key": "qa_maths", "question": "Which are there more of in these records: records labeled 'qa_maths' or records labeled 'qa_stock'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-50000-3-q15", "dataset": "en_intent", "id": "en-50000-3-q15", "haystack_uid": "en_intent:en-50000-3", "haystack_id": "en-50000-3", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "qa_definition", "candidates": ["qa_factoid", "qa_definition"], "answer": "qa_definition", "answer_key": "qa_definition", "question": "Which are there more of in these records: records labeled 'qa_factoid' or records labeled 'qa_definition'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent:en-50000-3-q12", "dataset": "en_intent", "id": "en-50000-3-q12", "haystack_uid": "en_intent:en-50000-3", "haystack_id": "en-50000-3", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_definition", "label_b": "qa_factoid", "candidates": ["qa_definition", "qa_factoid"], "answer": "qa_definition", "answer_key": "qa_definition", "question": "Which are there more of in these records: records labeled 'qa_definition' or records labeled 'qa_factoid'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent:en-50000-3-q13", "dataset": "en_intent", "id": "en-50000-3-q13", "haystack_uid": "en_intent:en-50000-3", "haystack_id": "en-50000-3", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "transport_ticket", "label_b": "calendar_set", "candidates": ["transport_ticket", "calendar_set"], "answer": "calendar_set", "answer_key": "calendar_set", "question": "Which are there more of in these records: records labeled 'transport_ticket' or records labeled 'calendar_set'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent:en-50000-4-q0", "dataset": "en_intent", "id": "en-50000-4-q0", "haystack_uid": "en_intent:en-50000-4", "haystack_id": "en-50000-4", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "social_post", "answer": "49", "question": "How many utterances have the intent 'social_post'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-50000-4-q1", "dataset": "en_intent", "id": "en-50000-4-q1", "haystack_uid": "en_intent:en-50000-4", "haystack_id": "en-50000-4", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "lists_query", "answer": "108", "question": "How many utterances have the intent 'lists_query'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-50000-4-q2", "dataset": "en_intent", "id": "en-50000-4-q2", "haystack_uid": "en_intent:en-50000-4", "haystack_id": "en-50000-4", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "calendar_remove", "answer": "261", "question": "How many utterances have the intent 'calendar_remove'? Answer with the number only.", "role": "control"}
@@ -71,8 +64,6 @@
 {"uid": "en_intent:en-50000-4-q9", "dataset": "en_intent", "id": "en-50000-4-q9", "haystack_uid": "en_intent:en-50000-4", "haystack_id": "en-50000-4", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "alarm_remove", "answer": "14", "question": "How many utterances have the intent 'alarm_remove'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-50000-4-q10", "dataset": "en_intent", "id": "en-50000-4-q10", "haystack_uid": "en_intent:en-50000-4", "haystack_id": "en-50000-4", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "music_likeness", "answer": "17", "question": "How many utterances have the intent 'music_likeness'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-50000-4-q11", "dataset": "en_intent", "id": "en-50000-4-q11", "haystack_uid": "en_intent:en-50000-4", "haystack_id": "en-50000-4", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "alarm_query", "answer": "6", "question": "How many utterances have the intent 'alarm_query'? Answer with the number only.", "role": "retrieval"}
-{"uid": "en_intent:en-50000-4-q12", "dataset": "en_intent", "id": "en-50000-4-q12", "haystack_uid": "en_intent:en-50000-4", "haystack_id": "en-50000-4", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "calendar_query", "label_b": "lists_query", "candidates": ["calendar_query", "lists_query"], "answer": "calendar_query", "answer_key": "calendar_query", "question": "Which are there more of in these records: records labeled 'calendar_query' or records labeled 'lists_query'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-50000-4-q13", "dataset": "en_intent", "id": "en-50000-4-q13", "haystack_uid": "en_intent:en-50000-4", "haystack_id": "en-50000-4", "language": "en", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "recommendation_events", "label_b": "qa_definition", "candidates": ["recommendation_events", "qa_definition"], "answer": "qa_definition", "answer_key": "qa_definition", "question": "Which are there more of in these records: records labeled 'recommendation_events' or records labeled 'qa_definition'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent:en-100000-0-q0", "dataset": "en_intent", "id": "en-100000-0-q0", "haystack_uid": "en_intent:en-100000-0", "haystack_id": "en-100000-0", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "iot_hue_lightup", "answer": "48", "question": "How many utterances have the intent 'iot_hue_lightup'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-100000-0-q1", "dataset": "en_intent", "id": "en-100000-0-q1", "haystack_uid": "en_intent:en-100000-0", "haystack_id": "en-100000-0", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "lists_query", "answer": "66", "question": "How many utterances have the intent 'lists_query'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-100000-0-q2", "dataset": "en_intent", "id": "en-100000-0-q2", "haystack_uid": "en_intent:en-100000-0", "haystack_id": "en-100000-0", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "news_query", "answer": "102", "question": "How many utterances have the intent 'news_query'? Answer with the number only.", "role": "control"}
@@ -85,7 +76,7 @@
 {"uid": "en_intent:en-100000-0-q9", "dataset": "en_intent", "id": "en-100000-0-q9", "haystack_uid": "en_intent:en-100000-0", "haystack_id": "en-100000-0", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "email_sendemail", "answer": "13", "question": "How many utterances have the intent 'email_sendemail'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-100000-0-q10", "dataset": "en_intent", "id": "en-100000-0-q10", "haystack_uid": "en_intent:en-100000-0", "haystack_id": "en-100000-0", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "general_joke", "answer": "15", "question": "How many utterances have the intent 'general_joke'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-100000-0-q11", "dataset": "en_intent", "id": "en-100000-0-q11", "haystack_uid": "en_intent:en-100000-0", "haystack_id": "en-100000-0", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "recommendation_movies", "answer": "29", "question": "How many utterances have the intent 'recommendation_movies'? Answer with the number only.", "role": "retrieval"}
-{"uid": "en_intent:en-100000-0-q12", "dataset": "en_intent", "id": "en-100000-0-q12", "haystack_uid": "en_intent:en-100000-0", "haystack_id": "en-100000-0", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "alarm_set", "label_b": "music_query", "candidates": ["alarm_set", "music_query"], "answer": "music_query", "answer_key": "music_query", "question": "Which are there more of in these records: records labeled 'alarm_set' or records labeled 'music_query'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent:en-100000-0-q12", "dataset": "en_intent", "id": "en-100000-0-q12", "haystack_uid": "en_intent:en-100000-0", "haystack_id": "en-100000-0", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "alarm_set", "label_b": "lists_remove", "candidates": ["alarm_set", "lists_remove"], "answer": "alarm_set", "answer_key": "alarm_set", "question": "Which are there more of in these records: records labeled 'alarm_set' or records labeled 'lists_remove'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent:en-100000-1-q0", "dataset": "en_intent", "id": "en-100000-1-q0", "haystack_uid": "en_intent:en-100000-1", "haystack_id": "en-100000-1", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "music_likeness", "answer": "161", "question": "How many utterances have the intent 'music_likeness'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-100000-1-q1", "dataset": "en_intent", "id": "en-100000-1-q1", "haystack_uid": "en_intent:en-100000-1", "haystack_id": "en-100000-1", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "recommendation_events", "answer": "252", "question": "How many utterances have the intent 'recommendation_events'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-100000-1-q2", "dataset": "en_intent", "id": "en-100000-1-q2", "haystack_uid": "en_intent:en-100000-1", "haystack_id": "en-100000-1", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "transport_ticket", "answer": "104", "question": "How many utterances have the intent 'transport_ticket'? Answer with the number only.", "role": "control"}
@@ -98,10 +89,8 @@
 {"uid": "en_intent:en-100000-1-q9", "dataset": "en_intent", "id": "en-100000-1-q9", "haystack_uid": "en_intent:en-100000-1", "haystack_id": "en-100000-1", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "audio_volume_up", "answer": "30", "question": "How many utterances have the intent 'audio_volume_up'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-100000-1-q10", "dataset": "en_intent", "id": "en-100000-1-q10", "haystack_uid": "en_intent:en-100000-1", "haystack_id": "en-100000-1", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "audio_volume_mute", "answer": "25", "question": "How many utterances have the intent 'audio_volume_mute'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-100000-1-q11", "dataset": "en_intent", "id": "en-100000-1-q11", "haystack_uid": "en_intent:en-100000-1", "haystack_id": "en-100000-1", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "iot_hue_lightup", "answer": "95", "question": "How many utterances have the intent 'iot_hue_lightup'? Answer with the number only.", "role": "control"}
-{"uid": "en_intent:en-100000-1-q12", "dataset": "en_intent", "id": "en-100000-1-q12", "haystack_uid": "en_intent:en-100000-1", "haystack_id": "en-100000-1", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "recommendation_locations", "label_b": "lists_createoradd", "candidates": ["recommendation_locations", "lists_createoradd"], "answer": "recommendation_locations", "answer_key": "recommendation_locations", "question": "Which are there more of in these records: records labeled 'recommendation_locations' or records labeled 'lists_createoradd'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-100000-1-q13", "dataset": "en_intent", "id": "en-100000-1-q13", "haystack_uid": "en_intent:en-100000-1", "haystack_id": "en-100000-1", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "email_querycontact", "label_b": "play_music", "candidates": ["email_querycontact", "play_music"], "answer": "email_querycontact", "answer_key": "email_querycontact", "question": "Which are there more of in these records: records labeled 'email_querycontact' or records labeled 'play_music'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-100000-1-q14", "dataset": "en_intent", "id": "en-100000-1-q14", "haystack_uid": "en_intent:en-100000-1", "haystack_id": "en-100000-1", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "lists_remove", "label_b": "alarm_set", "candidates": ["lists_remove", "alarm_set"], "answer": "lists_remove", "answer_key": "lists_remove", "question": "Which are there more of in these records: records labeled 'lists_remove' or records labeled 'alarm_set'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-100000-1-q15", "dataset": "en_intent", "id": "en-100000-1-q15", "haystack_uid": "en_intent:en-100000-1", "haystack_id": "en-100000-1", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "calendar_query", "label_b": "music_query", "candidates": ["calendar_query", "music_query"], "answer": "music_query", "answer_key": "music_query", "question": "Which are there more of in these records: records labeled 'calendar_query' or records labeled 'music_query'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent:en-100000-1-q12", "dataset": "en_intent", "id": "en-100000-1-q12", "haystack_uid": "en_intent:en-100000-1", "haystack_id": "en-100000-1", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "play_music", "label_b": "news_query", "candidates": ["play_music", "news_query"], "answer": "news_query", "answer_key": "news_query", "question": "Which are there more of in these records: records labeled 'play_music' or records labeled 'news_query'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent:en-100000-1-q13", "dataset": "en_intent", "id": "en-100000-1-q13", "haystack_uid": "en_intent:en-100000-1", "haystack_id": "en-100000-1", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "alarm_set", "label_b": "lists_remove", "candidates": ["alarm_set", "lists_remove"], "answer": "lists_remove", "answer_key": "lists_remove", "question": "Which are there more of in these records: records labeled 'alarm_set' or records labeled 'lists_remove'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent:en-100000-2-q0", "dataset": "en_intent", "id": "en-100000-2-q0", "haystack_uid": "en_intent:en-100000-2", "haystack_id": "en-100000-2", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "email_querycontact", "answer": "168", "question": "How many utterances have the intent 'email_querycontact'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-100000-2-q1", "dataset": "en_intent", "id": "en-100000-2-q1", "haystack_uid": "en_intent:en-100000-2", "haystack_id": "en-100000-2", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "transport_taxi", "answer": "126", "question": "How many utterances have the intent 'transport_taxi'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-100000-2-q2", "dataset": "en_intent", "id": "en-100000-2-q2", "haystack_uid": "en_intent:en-100000-2", "haystack_id": "en-100000-2", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "qa_definition", "answer": "245", "question": "How many utterances have the intent 'qa_definition'? Answer with the number only.", "role": "control"}
@@ -114,10 +103,9 @@
 {"uid": "en_intent:en-100000-2-q9", "dataset": "en_intent", "id": "en-100000-2-q9", "haystack_uid": "en_intent:en-100000-2", "haystack_id": "en-100000-2", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "cooking_recipe", "answer": "26", "question": "How many utterances have the intent 'cooking_recipe'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-100000-2-q10", "dataset": "en_intent", "id": "en-100000-2-q10", "haystack_uid": "en_intent:en-100000-2", "haystack_id": "en-100000-2", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "audio_volume_up", "answer": "19", "question": "How many utterances have the intent 'audio_volume_up'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-100000-2-q11", "dataset": "en_intent", "id": "en-100000-2-q11", "haystack_uid": "en_intent:en-100000-2", "haystack_id": "en-100000-2", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "general_quirky", "answer": "13", "question": "How many utterances have the intent 'general_quirky'? Answer with the number only.", "role": "retrieval"}
-{"uid": "en_intent:en-100000-2-q12", "dataset": "en_intent", "id": "en-100000-2-q12", "haystack_uid": "en_intent:en-100000-2", "haystack_id": "en-100000-2", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_currency", "label_b": "play_audiobook", "candidates": ["qa_currency", "play_audiobook"], "answer": "play_audiobook", "answer_key": "play_audiobook", "question": "Which are there more of in these records: records labeled 'qa_currency' or records labeled 'play_audiobook'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent:en-100000-2-q12", "dataset": "en_intent", "id": "en-100000-2-q12", "haystack_uid": "en_intent:en-100000-2", "haystack_id": "en-100000-2", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "transport_query", "label_b": "lists_remove", "candidates": ["transport_query", "lists_remove"], "answer": "transport_query", "answer_key": "transport_query", "question": "Which are there more of in these records: records labeled 'transport_query' or records labeled 'lists_remove'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent:en-100000-2-q13", "dataset": "en_intent", "id": "en-100000-2-q13", "haystack_uid": "en_intent:en-100000-2", "haystack_id": "en-100000-2", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "email_sendemail", "label_b": "weather_query", "candidates": ["email_sendemail", "weather_query"], "answer": "weather_query", "answer_key": "weather_query", "question": "Which are there more of in these records: records labeled 'email_sendemail' or records labeled 'weather_query'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-100000-2-q14", "dataset": "en_intent", "id": "en-100000-2-q14", "haystack_uid": "en_intent:en-100000-2", "haystack_id": "en-100000-2", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "lists_remove", "label_b": "transport_query", "candidates": ["lists_remove", "transport_query"], "answer": "transport_query", "answer_key": "transport_query", "question": "Which are there more of in these records: records labeled 'lists_remove' or records labeled 'transport_query'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-100000-2-q15", "dataset": "en_intent", "id": "en-100000-2-q15", "haystack_uid": "en_intent:en-100000-2", "haystack_id": "en-100000-2", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "lists_query", "label_b": "email_querycontact", "candidates": ["lists_query", "email_querycontact"], "answer": "lists_query", "answer_key": "lists_query", "question": "Which are there more of in these records: records labeled 'lists_query' or records labeled 'email_querycontact'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent:en-100000-2-q14", "dataset": "en_intent", "id": "en-100000-2-q14", "haystack_uid": "en_intent:en-100000-2", "haystack_id": "en-100000-2", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_currency", "label_b": "play_audiobook", "candidates": ["qa_currency", "play_audiobook"], "answer": "play_audiobook", "answer_key": "play_audiobook", "question": "Which are there more of in these records: records labeled 'qa_currency' or records labeled 'play_audiobook'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent:en-100000-3-q0", "dataset": "en_intent", "id": "en-100000-3-q0", "haystack_uid": "en_intent:en-100000-3", "haystack_id": "en-100000-3", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "qa_maths", "answer": "115", "question": "How many utterances have the intent 'qa_maths'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-100000-3-q1", "dataset": "en_intent", "id": "en-100000-3-q1", "haystack_uid": "en_intent:en-100000-3", "haystack_id": "en-100000-3", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "email_querycontact", "answer": "168", "question": "How many utterances have the intent 'email_querycontact'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-100000-3-q2", "dataset": "en_intent", "id": "en-100000-3-q2", "haystack_uid": "en_intent:en-100000-3", "haystack_id": "en-100000-3", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "play_podcasts", "answer": "75", "question": "How many utterances have the intent 'play_podcasts'? Answer with the number only.", "role": "control"}
@@ -130,7 +118,6 @@
 {"uid": "en_intent:en-100000-3-q9", "dataset": "en_intent", "id": "en-100000-3-q9", "haystack_uid": "en_intent:en-100000-3", "haystack_id": "en-100000-3", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "general_joke", "answer": "15", "question": "How many utterances have the intent 'general_joke'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-100000-3-q10", "dataset": "en_intent", "id": "en-100000-3-q10", "haystack_uid": "en_intent:en-100000-3", "haystack_id": "en-100000-3", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "general_quirky", "answer": "27", "question": "How many utterances have the intent 'general_quirky'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-100000-3-q11", "dataset": "en_intent", "id": "en-100000-3-q11", "haystack_uid": "en_intent:en-100000-3", "haystack_id": "en-100000-3", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "calendar_set", "answer": "816", "question": "How many utterances have the intent 'calendar_set'? Answer with the number only.", "role": "control"}
-{"uid": "en_intent:en-100000-3-q12", "dataset": "en_intent", "id": "en-100000-3-q12", "haystack_uid": "en_intent:en-100000-3", "haystack_id": "en-100000-3", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "music_query", "label_b": "qa_stock", "candidates": ["music_query", "qa_stock"], "answer": "qa_stock", "answer_key": "qa_stock", "question": "Which are there more of in these records: records labeled 'music_query' or records labeled 'qa_stock'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent:en-100000-4-q0", "dataset": "en_intent", "id": "en-100000-4-q0", "haystack_uid": "en_intent:en-100000-4", "haystack_id": "en-100000-4", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "news_query", "answer": "567", "question": "How many utterances have the intent 'news_query'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-100000-4-q1", "dataset": "en_intent", "id": "en-100000-4-q1", "haystack_uid": "en_intent:en-100000-4", "haystack_id": "en-100000-4", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "iot_hue_lightoff", "answer": "157", "question": "How many utterances have the intent 'iot_hue_lightoff'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent:en-100000-4-q2", "dataset": "en_intent", "id": "en-100000-4-q2", "haystack_uid": "en_intent:en-100000-4", "haystack_id": "en-100000-4", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "cooking_recipe", "answer": "248", "question": "How many utterances have the intent 'cooking_recipe'? Answer with the number only.", "role": "control"}
@@ -143,6 +130,4 @@
 {"uid": "en_intent:en-100000-4-q9", "dataset": "en_intent", "id": "en-100000-4-q9", "haystack_uid": "en_intent:en-100000-4", "haystack_id": "en-100000-4", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "email_query", "answer": "28", "question": "How many utterances have the intent 'email_query'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-100000-4-q10", "dataset": "en_intent", "id": "en-100000-4-q10", "haystack_uid": "en_intent:en-100000-4", "haystack_id": "en-100000-4", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "qa_factoid", "answer": "8", "question": "How many utterances have the intent 'qa_factoid'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent:en-100000-4-q11", "dataset": "en_intent", "id": "en-100000-4-q11", "haystack_uid": "en_intent:en-100000-4", "haystack_id": "en-100000-4", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "recommendation_events", "answer": "219", "question": "How many utterances have the intent 'recommendation_events'? Answer with the number only.", "role": "control"}
-{"uid": "en_intent:en-100000-4-q12", "dataset": "en_intent", "id": "en-100000-4-q12", "haystack_uid": "en_intent:en-100000-4", "haystack_id": "en-100000-4", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "general_quirky", "label_b": "calendar_remove", "candidates": ["general_quirky", "calendar_remove"], "answer": "general_quirky", "answer_key": "general_quirky", "question": "Which are there more of in these records: records labeled 'general_quirky' or records labeled 'calendar_remove'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-100000-4-q13", "dataset": "en_intent", "id": "en-100000-4-q13", "haystack_uid": "en_intent:en-100000-4", "haystack_id": "en-100000-4", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "takeaway_query", "label_b": "iot_hue_lightchange", "candidates": ["takeaway_query", "iot_hue_lightchange"], "answer": "takeaway_query", "answer_key": "takeaway_query", "question": "Which are there more of in these records: records labeled 'takeaway_query' or records labeled 'iot_hue_lightchange'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent:en-100000-4-q14", "dataset": "en_intent", "id": "en-100000-4-q14", "haystack_uid": "en_intent:en-100000-4", "haystack_id": "en-100000-4", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "play_audiobook", "label_b": "recommendation_locations", "candidates": ["play_audiobook", "recommendation_locations"], "answer": "play_audiobook", "answer_key": "play_audiobook", "question": "Which are there more of in these records: records labeled 'play_audiobook' or records labeled 'recommendation_locations'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent:en-100000-4-q12", "dataset": "en_intent", "id": "en-100000-4-q12", "haystack_uid": "en_intent:en-100000-4", "haystack_id": "en-100000-4", "language": "en", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "iot_hue_lightchange", "label_b": "takeaway_query", "candidates": ["iot_hue_lightchange", "takeaway_query"], "answer": "takeaway_query", "answer_key": "takeaway_query", "question": "Which are there more of in these records: records labeled 'iot_hue_lightchange' or records labeled 'takeaway_query'? Answer with the label name only.", "role": "core"}

en_intent_paired/manifest.json

@@ -1,6 +1,6 @@
 {
-  "version": "0.9.0",
-  "date": "2026-09-29",
+  "version": "0.10.0",
+  "date": "2026-10-01",
   "environment": {
     "python": "3.12.9",
     "polars": "1.42.0",
@@ -17,6 +17,12 @@
     "drop_label_leakage": true,
     "leak_label_words": false,
     "min_class_support": 0,
+    "boilerplate_keywords": [],
+    "boilerplate_min_keywords": 3,
+    "boilerplate_weak_min": 2,
+    "boilerplate_regex": "",
+    "drop_text_regex": [],
+    "strip_html": false,
     "seed": 42,
     "haystack_target_tokens": [
       50000,
@@ -65,6 +71,7 @@
     "close_min_diff": 2,
     "close_exclude_labels": [],
     "close_ordinal_gap": 0,
+    "close_allowed_pairs_file": "experiments/separable_pairs/intent_paired_shared.json",
     "core_doc_records": [
       3000
     ],
@@ -116,8 +123,8 @@
     },
     "core3000": {
       "n_haystacks": 10,
-      "mean_jaccard": 0.1514,
-      "max_jaccard": 0.1978,
+      "mean_jaccard": 0.1473,
+      "max_jaccard": 0.1795,
       "pool_fraction_per_haystack": 0.199
     }
   },
@@ -133,9 +140,9 @@
   "proportion_unit": "per_mille",
   "top_k_k": 3,
   "tokenizer": "Qwen/Qwen3-8B",
-  "questions_written": 175,
+  "questions_written": 161,
   "kind_distribution": {
-    "close_comparison": 55,
+    "close_comparison": 41,
     "count": 58,
     "label_vs_label": 11,
     "least_common": 10,
@@ -165,7 +172,7 @@
       "n_chars": 142197,
       "drift_target": "transport_taxi",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 14
     },
     {
       "haystack_id": "en-r3000-2",
@@ -174,7 +181,7 @@
       "n_chars": 140732,
       "drift_target": "music_likeness",
       "drift_ok": true,
-      "n_questions": 14
+      "n_questions": 12
     },
     {
       "haystack_id": "en-r3000-3",
@@ -183,7 +190,7 @@
       "n_chars": 139886,
       "drift_target": "iot_hue_lightoff",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 14
     },
     {
       "haystack_id": "en-r3000-4",
@@ -192,7 +199,7 @@
       "n_chars": 139371,
       "drift_target": "email_querycontact",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 14
     },
     {
       "haystack_id": "en-r6000-0",
@@ -201,7 +208,7 @@
       "n_chars": 287041,
       "drift_target": "lists_query",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 15
     },
     {
       "haystack_id": "en-r6000-1",
@@ -210,7 +217,7 @@
       "n_chars": 288635,
       "drift_target": "social_post",
       "drift_ok": true,
-      "n_questions": 15
+      "n_questions": 14
     },
     {
       "haystack_id": "en-r6000-2",
@@ -219,7 +226,7 @@
       "n_chars": 291739,
       "drift_target": "email_query",
       "drift_ok": true,
-      "n_questions": 15
+      "n_questions": 13
     },
     {
       "haystack_id": "en-r6000-3",
@@ -228,7 +235,7 @@
       "n_chars": 290276,
       "drift_target": "transport_query",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 15
     },
     {
       "haystack_id": "en-r6000-4",
@@ -237,13 +244,13 @@
       "n_chars": 279670,
       "drift_target": "lists_query",
       "drift_ok": true,
-      "n_questions": 15
+      "n_questions": 14
     },
     {
       "haystack_id": "en-core3000-0",
       "n_examples": 3000,
-      "n_tokens": 36401,
-      "n_chars": 138594,
+      "n_tokens": 37041,
+      "n_chars": 142659,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -259,7 +266,7 @@
         "calendar_set": 31,
         "cooking_recipe": 20,
         "datetime_query": 31,
-        "email_query": 204,
+        "email_query": 210,
         "email_querycontact": 40,
         "email_sendemail": 29,
         "general_joke": 56,
@@ -270,25 +277,25 @@
         "iot_hue_lightoff": 68,
         "iot_hue_lightup": 66,
         "lists_createoradd": 202,
-        "lists_query": 210,
-        "lists_remove": 40,
-        "music_likeness": 45,
-        "music_query": 97,
-        "news_query": 64,
-        "play_audiobook": 49,
-        "play_game": 56,
-        "play_music": 61,
-        "play_podcasts": 31,
-        "play_radio": 77,
-        "qa_currency": 26,
-        "qa_definition": 45,
+        "lists_query": 40,
+        "lists_remove": 45,
+        "music_likeness": 97,
+        "music_query": 64,
+        "news_query": 49,
+        "play_audiobook": 56,
+        "play_game": 61,
+        "play_music": 31,
+        "play_podcasts": 77,
+        "play_radio": 26,
+        "qa_currency": 45,
+        "qa_definition": 43,
         "qa_factoid": 43,
-        "qa_maths": 43,
-        "qa_stock": 41,
-        "recommendation_events": 66,
-        "recommendation_locations": 24,
-        "recommendation_movies": 63,
-        "social_post": 53,
+        "qa_maths": 41,
+        "qa_stock": 66,
+        "recommendation_events": 24,
+        "recommendation_locations": 63,
+        "recommendation_movies": 53,
+        "social_post": 204,
         "social_query": 26,
         "takeaway_order": 57,
         "takeaway_query": 53,
@@ -302,8 +309,8 @@
     {
       "haystack_id": "en-core3000-1",
       "n_examples": 3000,
-      "n_tokens": 37184,
-      "n_chars": 143017,
+      "n_tokens": 37353,
+      "n_chars": 144720,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -314,27 +321,27 @@
         "alarm_set": 110,
         "audio_volume_mute": 56,
         "audio_volume_up": 45,
-        "calendar_query": 210,
-        "calendar_remove": 28,
+        "calendar_query": 203,
+        "calendar_remove": 210,
         "calendar_set": 203,
-        "cooking_recipe": 61,
-        "datetime_query": 42,
-        "email_query": 17,
-        "email_querycontact": 47,
-        "email_sendemail": 16,
-        "general_joke": 33,
-        "general_quirky": 46,
-        "iot_cleaning": 85,
-        "iot_coffee": 11,
-        "iot_hue_lightchange": 45,
-        "iot_hue_lightoff": 36,
-        "iot_hue_lightup": 86,
-        "lists_createoradd": 34,
-        "lists_query": 210,
+        "cooking_recipe": 28,
+        "datetime_query": 61,
+        "email_query": 42,
+        "email_querycontact": 17,
+        "email_sendemail": 47,
+        "general_joke": 16,
+        "general_quirky": 33,
+        "iot_cleaning": 46,
+        "iot_coffee": 85,
+        "iot_hue_lightchange": 11,
+        "iot_hue_lightoff": 45,
+        "iot_hue_lightup": 36,
+        "lists_createoradd": 86,
+        "lists_query": 34,
         "lists_remove": 42,
         "music_likeness": 40,
         "music_query": 88,
-        "news_query": 203,
+        "news_query": 210,
         "play_audiobook": 21,
         "play_game": 32,
         "play_music": 77,
@@ -362,8 +369,8 @@
     {
       "haystack_id": "en-core3000-2",
       "n_examples": 3000,
-      "n_tokens": 36677,
-      "n_chars": 140259,
+      "n_tokens": 37772,
+      "n_chars": 145684,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -376,21 +383,21 @@
         "audio_volume_up": 73,
         "calendar_query": 89,
         "calendar_remove": 64,
-        "calendar_set": 37,
-        "cooking_recipe": 11,
-        "datetime_query": 39,
-        "email_query": 34,
-        "email_querycontact": 49,
-        "email_sendemail": 56,
-        "general_joke": 69,
-        "general_quirky": 56,
-        "iot_cleaning": 17,
-        "iot_coffee": 74,
-        "iot_hue_lightchange": 69,
-        "iot_hue_lightoff": 83,
-        "iot_hue_lightup": 41,
-        "lists_createoradd": 210,
-        "lists_query": 203,
+        "calendar_set": 210,
+        "cooking_recipe": 37,
+        "datetime_query": 11,
+        "email_query": 39,
+        "email_querycontact": 34,
+        "email_sendemail": 49,
+        "general_joke": 56,
+        "general_quirky": 69,
+        "iot_cleaning": 56,
+        "iot_coffee": 17,
+        "iot_hue_lightchange": 74,
+        "iot_hue_lightoff": 69,
+        "iot_hue_lightup": 83,
+        "lists_createoradd": 204,
+        "lists_query": 41,
         "lists_remove": 30,
         "music_likeness": 42,
         "music_query": 66,
@@ -399,14 +406,14 @@
         "play_game": 59,
         "play_music": 58,
         "play_podcasts": 15,
-        "play_radio": 58,
-        "qa_currency": 21,
-        "qa_definition": 72,
-        "qa_factoid": 48,
-        "qa_maths": 61,
-        "qa_stock": 132,
-        "recommendation_events": 41,
-        "recommendation_locations": 204,
+        "play_radio": 203,
+        "qa_currency": 58,
+        "qa_definition": 21,
+        "qa_factoid": 72,
+        "qa_maths": 48,
+        "qa_stock": 61,
+        "recommendation_events": 132,
+        "recommendation_locations": 41,
         "recommendation_movies": 22,
         "social_post": 21,
         "social_query": 19,
@@ -422,8 +429,8 @@
     {
       "haystack_id": "en-core3000-3",
       "n_examples": 3000,
-      "n_tokens": 36622,
-      "n_chars": 140646,
+      "n_tokens": 36612,
+      "n_chars": 140460,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -434,38 +441,38 @@
         "alarm_set": 32,
         "audio_volume_mute": 40,
         "audio_volume_up": 51,
-        "calendar_query": 55,
-        "calendar_remove": 69,
-        "calendar_set": 36,
-        "cooking_recipe": 46,
+        "calendar_query": 203,
+        "calendar_remove": 55,
+        "calendar_set": 69,
+        "cooking_recipe": 210,
         "datetime_query": 204,
-        "email_query": 35,
-        "email_querycontact": 64,
-        "email_sendemail": 49,
-        "general_joke": 60,
-        "general_quirky": 33,
-        "iot_cleaning": 53,
-        "iot_coffee": 43,
-        "iot_hue_lightchange": 81,
-        "iot_hue_lightoff": 84,
-        "iot_hue_lightup": 34,
-        "lists_createoradd": 117,
-        "lists_query": 210,
-        "lists_remove": 24,
-        "music_likeness": 31,
-        "music_query": 74,
-        "news_query": 54,
-        "play_audiobook": 40,
-        "play_game": 35,
-        "play_music": 30,
-        "play_podcasts": 67,
-        "play_radio": 203,
-        "qa_currency": 37,
-        "qa_definition": 51,
-        "qa_factoid": 14,
-        "qa_maths": 115,
-        "qa_stock": 45,
-        "recommendation_events": 210,
+        "email_query": 36,
+        "email_querycontact": 46,
+        "email_sendemail": 35,
+        "general_joke": 64,
+        "general_quirky": 49,
+        "iot_cleaning": 60,
+        "iot_coffee": 33,
+        "iot_hue_lightchange": 53,
+        "iot_hue_lightoff": 43,
+        "iot_hue_lightup": 81,
+        "lists_createoradd": 84,
+        "lists_query": 34,
+        "lists_remove": 103,
+        "music_likeness": 24,
+        "music_query": 31,
+        "news_query": 74,
+        "play_audiobook": 54,
+        "play_game": 40,
+        "play_music": 35,
+        "play_podcasts": 30,
+        "play_radio": 210,
+        "qa_currency": 67,
+        "qa_definition": 37,
+        "qa_factoid": 51,
+        "qa_maths": 14,
+        "qa_stock": 129,
+        "recommendation_events": 45,
         "recommendation_locations": 49,
         "recommendation_movies": 34,
         "social_post": 34,
@@ -602,8 +609,8 @@
     {
       "haystack_id": "en-core3000-6",
       "n_examples": 3000,
-      "n_tokens": 36224,
-      "n_chars": 139550,
+      "n_tokens": 36480,
+      "n_chars": 141150,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -623,25 +630,25 @@
         "email_querycontact": 37,
         "email_sendemail": 20,
         "general_joke": 49,
-        "general_quirky": 210,
-        "iot_cleaning": 45,
-        "iot_coffee": 66,
-        "iot_hue_lightchange": 48,
-        "iot_hue_lightoff": 33,
-        "iot_hue_lightup": 34,
-        "lists_createoradd": 21,
-        "lists_query": 210,
-        "lists_remove": 16,
-        "music_likeness": 90,
-        "music_query": 29,
-        "news_query": 33,
+        "general_quirky": 45,
+        "iot_cleaning": 66,
+        "iot_coffee": 48,
+        "iot_hue_lightchange": 33,
+        "iot_hue_lightoff": 34,
+        "iot_hue_lightup": 21,
+        "lists_createoradd": 203,
+        "lists_query": 16,
+        "lists_remove": 90,
+        "music_likeness": 29,
+        "music_query": 33,
+        "news_query": 210,
         "play_audiobook": 52,
         "play_game": 28,
         "play_music": 39,
-        "play_podcasts": 203,
+        "play_podcasts": 204,
         "play_radio": 61,
         "qa_currency": 24,
-        "qa_definition": 204,
+        "qa_definition": 210,
         "qa_factoid": 26,
         "qa_maths": 115,
         "qa_stock": 35,

en_intent_paired/questions.jsonl

@@ -11,9 +11,9 @@
 {"uid": "en_intent_paired:en-r3000-0-q10", "dataset": "en_intent_paired", "id": "en-r3000-0-q10", "haystack_uid": "en_intent_paired:en-r3000-0", "haystack_id": "en-r3000-0", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "lists_remove", "answer": "7", "question": "How many utterances have the intent 'lists_remove'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r3000-0-q11", "dataset": "en_intent_paired", "id": "en-r3000-0-q11", "haystack_uid": "en_intent_paired:en-r3000-0", "haystack_id": "en-r3000-0", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "general_quirky", "answer": "21", "question": "How many utterances have the intent 'general_quirky'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r3000-0-q12", "dataset": "en_intent_paired", "id": "en-r3000-0-q12", "haystack_uid": "en_intent_paired:en-r3000-0", "haystack_id": "en-r3000-0", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "qa_stock", "label_b": "iot_hue_lightchange", "candidates": ["qa_stock", "iot_hue_lightchange"], "answer": "qa_stock", "answer_key": "qa_stock", "question": "Which are there more of in these records: records labeled 'qa_stock' or records labeled 'iot_hue_lightchange'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r3000-0-q13", "dataset": "en_intent_paired", "id": "en-r3000-0-q13", "haystack_uid": "en_intent_paired:en-r3000-0", "haystack_id": "en-r3000-0", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "iot_coffee", "label_b": "takeaway_query", "candidates": ["iot_coffee", "takeaway_query"], "answer": "takeaway_query", "answer_key": "takeaway_query", "question": "Which are there more of in these records: records labeled 'iot_coffee' or records labeled 'takeaway_query'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r3000-0-q14", "dataset": "en_intent_paired", "id": "en-r3000-0-q14", "haystack_uid": "en_intent_paired:en-r3000-0", "haystack_id": "en-r3000-0", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_query", "label_b": "cooking_recipe", "candidates": ["calendar_query", "cooking_recipe"], "answer": "calendar_query", "answer_key": "calendar_query", "question": "Which are there more of in these records: records labeled 'calendar_query' or records labeled 'cooking_recipe'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r3000-0-q15", "dataset": "en_intent_paired", "id": "en-r3000-0-q15", "haystack_uid": "en_intent_paired:en-r3000-0", "haystack_id": "en-r3000-0", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "social_post", "label_b": "iot_hue_lightoff", "candidates": ["social_post", "iot_hue_lightoff"], "answer": "social_post", "answer_key": "social_post", "question": "Which are there more of in these records: records labeled 'social_post' or records labeled 'iot_hue_lightoff'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r3000-0-q13", "dataset": "en_intent_paired", "id": "en-r3000-0-q13", "haystack_uid": "en_intent_paired:en-r3000-0", "haystack_id": "en-r3000-0", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "social_post", "label_b": "iot_hue_lightoff", "candidates": ["social_post", "iot_hue_lightoff"], "answer": "social_post", "answer_key": "social_post", "question": "Which are there more of in these records: records labeled 'social_post' or records labeled 'iot_hue_lightoff'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r3000-0-q14", "dataset": "en_intent_paired", "id": "en-r3000-0-q14", "haystack_uid": "en_intent_paired:en-r3000-0", "haystack_id": "en-r3000-0", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "takeaway_query", "label_b": "iot_coffee", "candidates": ["takeaway_query", "iot_coffee"], "answer": "takeaway_query", "answer_key": "takeaway_query", "question": "Which are there more of in these records: records labeled 'takeaway_query' or records labeled 'iot_coffee'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r3000-0-q15", "dataset": "en_intent_paired", "id": "en-r3000-0-q15", "haystack_uid": "en_intent_paired:en-r3000-0", "haystack_id": "en-r3000-0", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "cooking_recipe", "label_b": "takeaway_order", "candidates": ["cooking_recipe", "takeaway_order"], "answer": "cooking_recipe", "answer_key": "cooking_recipe", "question": "Which are there more of in these records: records labeled 'cooking_recipe' or records labeled 'takeaway_order'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent_paired:en-r3000-1-q0", "dataset": "en_intent_paired", "id": "en-r3000-1-q0", "haystack_uid": "en_intent_paired:en-r3000-1", "haystack_id": "en-r3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "calendar_query", "answer": "27", "question": "How many utterances have the intent 'calendar_query'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r3000-1-q1", "dataset": "en_intent_paired", "id": "en-r3000-1-q1", "haystack_uid": "en_intent_paired:en-r3000-1", "haystack_id": "en-r3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "social_query", "answer": "138", "question": "How many utterances have the intent 'social_query'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r3000-1-q2", "dataset": "en_intent_paired", "id": "en-r3000-1-q2", "haystack_uid": "en_intent_paired:en-r3000-1", "haystack_id": "en-r3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "cooking_recipe", "answer": "35", "question": "How many utterances have the intent 'cooking_recipe'? Answer with the number only.", "role": "control"}
@@ -26,10 +26,8 @@
 {"uid": "en_intent_paired:en-r3000-1-q9", "dataset": "en_intent_paired", "id": "en-r3000-1-q9", "haystack_uid": "en_intent_paired:en-r3000-1", "haystack_id": "en-r3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "iot_coffee", "answer": "14", "question": "How many utterances have the intent 'iot_coffee'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r3000-1-q10", "dataset": "en_intent_paired", "id": "en-r3000-1-q10", "haystack_uid": "en_intent_paired:en-r3000-1", "haystack_id": "en-r3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "recommendation_locations", "answer": "5", "question": "How many utterances have the intent 'recommendation_locations'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r3000-1-q11", "dataset": "en_intent_paired", "id": "en-r3000-1-q11", "haystack_uid": "en_intent_paired:en-r3000-1", "haystack_id": "en-r3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "calendar_remove", "answer": "30", "question": "How many utterances have the intent 'calendar_remove'? Answer with the number only.", "role": "retrieval"}
-{"uid": "en_intent_paired:en-r3000-1-q12", "dataset": "en_intent_paired", "id": "en-r3000-1-q12", "haystack_uid": "en_intent_paired:en-r3000-1", "haystack_id": "en-r3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "datetime_query", "label_b": "social_post", "candidates": ["datetime_query", "social_post"], "answer": "social_post", "answer_key": "social_post", "question": "Which are there more of in these records: records labeled 'datetime_query' or records labeled 'social_post'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r3000-1-q13", "dataset": "en_intent_paired", "id": "en-r3000-1-q13", "haystack_uid": "en_intent_paired:en-r3000-1", "haystack_id": "en-r3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "transport_ticket", "label_b": "weather_query", "candidates": ["transport_ticket", "weather_query"], "answer": "transport_ticket", "answer_key": "transport_ticket", "question": "Which are there more of in these records: records labeled 'transport_ticket' or records labeled 'weather_query'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r3000-1-q14", "dataset": "en_intent_paired", "id": "en-r3000-1-q14", "haystack_uid": "en_intent_paired:en-r3000-1", "haystack_id": "en-r3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "email_sendemail", "label_b": "music_likeness", "candidates": ["email_sendemail", "music_likeness"], "answer": "music_likeness", "answer_key": "music_likeness", "question": "Which are there more of in these records: records labeled 'email_sendemail' or records labeled 'music_likeness'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r3000-1-q15", "dataset": "en_intent_paired", "id": "en-r3000-1-q15", "haystack_uid": "en_intent_paired:en-r3000-1", "haystack_id": "en-r3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "transport_traffic", "label_b": "transport_query", "candidates": ["transport_traffic", "transport_query"], "answer": "transport_traffic", "answer_key": "transport_traffic", "question": "Which are there more of in these records: records labeled 'transport_traffic' or records labeled 'transport_query'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r3000-1-q12", "dataset": "en_intent_paired", "id": "en-r3000-1-q12", "haystack_uid": "en_intent_paired:en-r3000-1", "haystack_id": "en-r3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "social_post", "label_b": "datetime_query", "candidates": ["social_post", "datetime_query"], "answer": "social_post", "answer_key": "social_post", "question": "Which are there more of in these records: records labeled 'social_post' or records labeled 'datetime_query'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r3000-1-q13", "dataset": "en_intent_paired", "id": "en-r3000-1-q13", "haystack_uid": "en_intent_paired:en-r3000-1", "haystack_id": "en-r3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "email_query", "label_b": "weather_query", "candidates": ["email_query", "weather_query"], "answer": "email_query", "answer_key": "email_query", "question": "Which are there more of in these records: records labeled 'email_query' or records labeled 'weather_query'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent_paired:en-r3000-2-q0", "dataset": "en_intent_paired", "id": "en-r3000-2-q0", "haystack_uid": "en_intent_paired:en-r3000-2", "haystack_id": "en-r3000-2", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "qa_definition", "answer": "195", "question": "How many utterances have the intent 'qa_definition'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r3000-2-q1", "dataset": "en_intent_paired", "id": "en-r3000-2-q1", "haystack_uid": "en_intent_paired:en-r3000-2", "haystack_id": "en-r3000-2", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "iot_hue_lightchange", "answer": "36", "question": "How many utterances have the intent 'iot_hue_lightchange'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r3000-2-q2", "dataset": "en_intent_paired", "id": "en-r3000-2-q2", "haystack_uid": "en_intent_paired:en-r3000-2", "haystack_id": "en-r3000-2", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "email_query", "answer": "16", "question": "How many utterances have the intent 'email_query'? Answer with the number only.", "role": "control"}
@@ -42,8 +40,6 @@
 {"uid": "en_intent_paired:en-r3000-2-q9", "dataset": "en_intent_paired", "id": "en-r3000-2-q9", "haystack_uid": "en_intent_paired:en-r3000-2", "haystack_id": "en-r3000-2", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "play_audiobook", "answer": "13", "question": "How many utterances have the intent 'play_audiobook'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r3000-2-q10", "dataset": "en_intent_paired", "id": "en-r3000-2-q10", "haystack_uid": "en_intent_paired:en-r3000-2", "haystack_id": "en-r3000-2", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "play_radio", "answer": "10", "question": "How many utterances have the intent 'play_radio'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r3000-2-q11", "dataset": "en_intent_paired", "id": "en-r3000-2-q11", "haystack_uid": "en_intent_paired:en-r3000-2", "haystack_id": "en-r3000-2", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "transport_taxi", "answer": "28", "question": "How many utterances have the intent 'transport_taxi'? Answer with the number only.", "role": "retrieval"}
-{"uid": "en_intent_paired:en-r3000-2-q12", "dataset": "en_intent_paired", "id": "en-r3000-2-q12", "haystack_uid": "en_intent_paired:en-r3000-2", "haystack_id": "en-r3000-2", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "music_query", "label_b": "qa_stock", "candidates": ["music_query", "qa_stock"], "answer": "qa_stock", "answer_key": "qa_stock", "question": "Which are there more of in these records: records labeled 'music_query' or records labeled 'qa_stock'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r3000-2-q13", "dataset": "en_intent_paired", "id": "en-r3000-2-q13", "haystack_uid": "en_intent_paired:en-r3000-2", "haystack_id": "en-r3000-2", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "news_query", "label_b": "qa_maths", "candidates": ["news_query", "qa_maths"], "answer": "qa_maths", "answer_key": "qa_maths", "question": "Which are there more of in these records: records labeled 'news_query' or records labeled 'qa_maths'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent_paired:en-r3000-3-q0", "dataset": "en_intent_paired", "id": "en-r3000-3-q0", "haystack_uid": "en_intent_paired:en-r3000-3", "haystack_id": "en-r3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "iot_hue_lightup", "answer": "96", "question": "How many utterances have the intent 'iot_hue_lightup'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r3000-3-q1", "dataset": "en_intent_paired", "id": "en-r3000-3-q1", "haystack_uid": "en_intent_paired:en-r3000-3", "haystack_id": "en-r3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "music_query", "answer": "15", "question": "How many utterances have the intent 'music_query'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r3000-3-q2", "dataset": "en_intent_paired", "id": "en-r3000-3-q2", "haystack_uid": "en_intent_paired:en-r3000-3", "haystack_id": "en-r3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "transport_taxi", "answer": "89", "question": "How many utterances have the intent 'transport_taxi'? Answer with the number only.", "role": "control"}
@@ -56,10 +52,8 @@
 {"uid": "en_intent_paired:en-r3000-3-q9", "dataset": "en_intent_paired", "id": "en-r3000-3-q9", "haystack_uid": "en_intent_paired:en-r3000-3", "haystack_id": "en-r3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "takeaway_query", "answer": "18", "question": "How many utterances have the intent 'takeaway_query'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r3000-3-q10", "dataset": "en_intent_paired", "id": "en-r3000-3-q10", "haystack_uid": "en_intent_paired:en-r3000-3", "haystack_id": "en-r3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "play_radio", "answer": "15", "question": "How many utterances have the intent 'play_radio'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r3000-3-q11", "dataset": "en_intent_paired", "id": "en-r3000-3-q11", "haystack_uid": "en_intent_paired:en-r3000-3", "haystack_id": "en-r3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "audio_volume_up", "answer": "24", "question": "How many utterances have the intent 'audio_volume_up'? Answer with the number only.", "role": "retrieval"}
-{"uid": "en_intent_paired:en-r3000-3-q12", "dataset": "en_intent_paired", "id": "en-r3000-3-q12", "haystack_uid": "en_intent_paired:en-r3000-3", "haystack_id": "en-r3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "general_joke", "label_b": "qa_maths", "candidates": ["general_joke", "qa_maths"], "answer": "qa_maths", "answer_key": "qa_maths", "question": "Which are there more of in these records: records labeled 'general_joke' or records labeled 'qa_maths'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r3000-3-q13", "dataset": "en_intent_paired", "id": "en-r3000-3-q13", "haystack_uid": "en_intent_paired:en-r3000-3", "haystack_id": "en-r3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "alarm_remove", "label_b": "iot_coffee", "candidates": ["alarm_remove", "iot_coffee"], "answer": "iot_coffee", "answer_key": "iot_coffee", "question": "Which are there more of in these records: records labeled 'alarm_remove' or records labeled 'iot_coffee'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r3000-3-q14", "dataset": "en_intent_paired", "id": "en-r3000-3-q14", "haystack_uid": "en_intent_paired:en-r3000-3", "haystack_id": "en-r3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_createoradd", "label_b": "iot_hue_lightchange", "candidates": ["lists_createoradd", "iot_hue_lightchange"], "answer": "lists_createoradd", "answer_key": "lists_createoradd", "question": "Which are there more of in these records: records labeled 'lists_createoradd' or records labeled 'iot_hue_lightchange'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r3000-3-q15", "dataset": "en_intent_paired", "id": "en-r3000-3-q15", "haystack_uid": "en_intent_paired:en-r3000-3", "haystack_id": "en-r3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "weather_query", "label_b": "general_quirky", "candidates": ["weather_query", "general_quirky"], "answer": "weather_query", "answer_key": "weather_query", "question": "Which are there more of in these records: records labeled 'weather_query' or records labeled 'general_quirky'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r3000-3-q12", "dataset": "en_intent_paired", "id": "en-r3000-3-q12", "haystack_uid": "en_intent_paired:en-r3000-3", "haystack_id": "en-r3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "iot_coffee", "label_b": "qa_factoid", "candidates": ["iot_coffee", "qa_factoid"], "answer": "qa_factoid", "answer_key": "qa_factoid", "question": "Which are there more of in these records: records labeled 'iot_coffee' or records labeled 'qa_factoid'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r3000-3-q13", "dataset": "en_intent_paired", "id": "en-r3000-3-q13", "haystack_uid": "en_intent_paired:en-r3000-3", "haystack_id": "en-r3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_createoradd", "label_b": "iot_hue_lightchange", "candidates": ["lists_createoradd", "iot_hue_lightchange"], "answer": "lists_createoradd", "answer_key": "lists_createoradd", "question": "Which are there more of in these records: records labeled 'lists_createoradd' or records labeled 'iot_hue_lightchange'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent_paired:en-r3000-4-q0", "dataset": "en_intent_paired", "id": "en-r3000-4-q0", "haystack_uid": "en_intent_paired:en-r3000-4", "haystack_id": "en-r3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "iot_coffee", "answer": "60", "question": "How many utterances have the intent 'iot_coffee'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r3000-4-q1", "dataset": "en_intent_paired", "id": "en-r3000-4-q1", "haystack_uid": "en_intent_paired:en-r3000-4", "haystack_id": "en-r3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "lists_createoradd", "answer": "33", "question": "How many utterances have the intent 'lists_createoradd'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r3000-4-q2", "dataset": "en_intent_paired", "id": "en-r3000-4-q2", "haystack_uid": "en_intent_paired:en-r3000-4", "haystack_id": "en-r3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "iot_hue_lightoff", "answer": "55", "question": "How many utterances have the intent 'iot_hue_lightoff'? Answer with the number only.", "role": "control"}
@@ -72,10 +66,8 @@
 {"uid": "en_intent_paired:en-r3000-4-q9", "dataset": "en_intent_paired", "id": "en-r3000-4-q9", "haystack_uid": "en_intent_paired:en-r3000-4", "haystack_id": "en-r3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "calendar_remove", "answer": "6", "question": "How many utterances have the intent 'calendar_remove'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r3000-4-q10", "dataset": "en_intent_paired", "id": "en-r3000-4-q10", "haystack_uid": "en_intent_paired:en-r3000-4", "haystack_id": "en-r3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "social_query", "answer": "15", "question": "How many utterances have the intent 'social_query'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r3000-4-q11", "dataset": "en_intent_paired", "id": "en-r3000-4-q11", "haystack_uid": "en_intent_paired:en-r3000-4", "haystack_id": "en-r3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "lists_query", "answer": "19", "question": "How many utterances have the intent 'lists_query'? Answer with the number only.", "role": "retrieval"}
-{"uid": "en_intent_paired:en-r3000-4-q12", "dataset": "en_intent_paired", "id": "en-r3000-4-q12", "haystack_uid": "en_intent_paired:en-r3000-4", "haystack_id": "en-r3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "iot_coffee", "candidates": ["qa_factoid", "iot_coffee"], "answer": "qa_factoid", "answer_key": "qa_factoid", "question": "Which are there more of in these records: records labeled 'qa_factoid' or records labeled 'iot_coffee'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r3000-4-q13", "dataset": "en_intent_paired", "id": "en-r3000-4-q13", "haystack_uid": "en_intent_paired:en-r3000-4", "haystack_id": "en-r3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "email_querycontact", "label_b": "play_radio", "candidates": ["email_querycontact", "play_radio"], "answer": "play_radio", "answer_key": "play_radio", "question": "Which are there more of in these records: records labeled 'email_querycontact' or records labeled 'play_radio'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r3000-4-q14", "dataset": "en_intent_paired", "id": "en-r3000-4-q14", "haystack_uid": "en_intent_paired:en-r3000-4", "haystack_id": "en-r3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "cooking_recipe", "label_b": "play_audiobook", "candidates": ["cooking_recipe", "play_audiobook"], "answer": "cooking_recipe", "answer_key": "cooking_recipe", "question": "Which are there more of in these records: records labeled 'cooking_recipe' or records labeled 'play_audiobook'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r3000-4-q15", "dataset": "en_intent_paired", "id": "en-r3000-4-q15", "haystack_uid": "en_intent_paired:en-r3000-4", "haystack_id": "en-r3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "play_music", "label_b": "recommendation_movies", "candidates": ["play_music", "recommendation_movies"], "answer": "play_music", "answer_key": "play_music", "question": "Which are there more of in these records: records labeled 'play_music' or records labeled 'recommendation_movies'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r3000-4-q12", "dataset": "en_intent_paired", "id": "en-r3000-4-q12", "haystack_uid": "en_intent_paired:en-r3000-4", "haystack_id": "en-r3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "play_audiobook", "candidates": ["qa_factoid", "play_audiobook"], "answer": "qa_factoid", "answer_key": "qa_factoid", "question": "Which are there more of in these records: records labeled 'qa_factoid' or records labeled 'play_audiobook'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r3000-4-q13", "dataset": "en_intent_paired", "id": "en-r3000-4-q13", "haystack_uid": "en_intent_paired:en-r3000-4", "haystack_id": "en-r3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "cooking_recipe", "label_b": "iot_coffee", "candidates": ["cooking_recipe", "iot_coffee"], "answer": "cooking_recipe", "answer_key": "cooking_recipe", "question": "Which are there more of in these records: records labeled 'cooking_recipe' or records labeled 'iot_coffee'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent_paired:en-r6000-0-q0", "dataset": "en_intent_paired", "id": "en-r6000-0-q0", "haystack_uid": "en_intent_paired:en-r6000-0", "haystack_id": "en-r6000-0", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "iot_cleaning", "answer": "108", "question": "How many utterances have the intent 'iot_cleaning'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r6000-0-q1", "dataset": "en_intent_paired", "id": "en-r6000-0-q1", "haystack_uid": "en_intent_paired:en-r6000-0", "haystack_id": "en-r6000-0", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "takeaway_query", "answer": "14", "question": "How many utterances have the intent 'takeaway_query'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r6000-0-q2", "dataset": "en_intent_paired", "id": "en-r6000-0-q2", "haystack_uid": "en_intent_paired:en-r6000-0", "haystack_id": "en-r6000-0", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "play_podcasts", "answer": "74", "question": "How many utterances have the intent 'play_podcasts'? Answer with the number only.", "role": "control"}
@@ -88,10 +80,9 @@
 {"uid": "en_intent_paired:en-r6000-0-q9", "dataset": "en_intent_paired", "id": "en-r6000-0-q9", "haystack_uid": "en_intent_paired:en-r6000-0", "haystack_id": "en-r6000-0", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "play_radio", "answer": "20", "question": "How many utterances have the intent 'play_radio'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r6000-0-q10", "dataset": "en_intent_paired", "id": "en-r6000-0-q10", "haystack_uid": "en_intent_paired:en-r6000-0", "haystack_id": "en-r6000-0", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "email_sendemail", "answer": "213", "question": "How many utterances have the intent 'email_sendemail'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r6000-0-q11", "dataset": "en_intent_paired", "id": "en-r6000-0-q11", "haystack_uid": "en_intent_paired:en-r6000-0", "haystack_id": "en-r6000-0", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "recommendation_movies", "answer": "98", "question": "How many utterances have the intent 'recommendation_movies'? Answer with the number only.", "role": "control"}
-{"uid": "en_intent_paired:en-r6000-0-q12", "dataset": "en_intent_paired", "id": "en-r6000-0-q12", "haystack_uid": "en_intent_paired:en-r6000-0", "haystack_id": "en-r6000-0", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "social_query", "label_b": "qa_currency", "candidates": ["social_query", "qa_currency"], "answer": "qa_currency", "answer_key": "qa_currency", "question": "Which are there more of in these records: records labeled 'social_query' or records labeled 'qa_currency'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r6000-0-q13", "dataset": "en_intent_paired", "id": "en-r6000-0-q13", "haystack_uid": "en_intent_paired:en-r6000-0", "haystack_id": "en-r6000-0", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "music_query", "label_b": "iot_hue_lightoff", "candidates": ["music_query", "iot_hue_lightoff"], "answer": "iot_hue_lightoff", "answer_key": "iot_hue_lightoff", "question": "Which are there more of in these records: records labeled 'music_query' or records labeled 'iot_hue_lightoff'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r6000-0-q14", "dataset": "en_intent_paired", "id": "en-r6000-0-q14", "haystack_uid": "en_intent_paired:en-r6000-0", "haystack_id": "en-r6000-0", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "recommendation_events", "label_b": "alarm_set", "candidates": ["recommendation_events", "alarm_set"], "answer": "recommendation_events", "answer_key": "recommendation_events", "question": "Which are there more of in these records: records labeled 'recommendation_events' or records labeled 'alarm_set'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r6000-0-q15", "dataset": "en_intent_paired", "id": "en-r6000-0-q15", "haystack_uid": "en_intent_paired:en-r6000-0", "haystack_id": "en-r6000-0", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "qa_definition", "label_b": "music_likeness", "candidates": ["qa_definition", "music_likeness"], "answer": "qa_definition", "answer_key": "qa_definition", "question": "Which are there more of in these records: records labeled 'qa_definition' or records labeled 'music_likeness'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r6000-0-q12", "dataset": "en_intent_paired", "id": "en-r6000-0-q12", "haystack_uid": "en_intent_paired:en-r6000-0", "haystack_id": "en-r6000-0", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "calendar_set", "label_b": "transport_query", "candidates": ["calendar_set", "transport_query"], "answer": "transport_query", "answer_key": "transport_query", "question": "Which are there more of in these records: records labeled 'calendar_set' or records labeled 'transport_query'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r6000-0-q13", "dataset": "en_intent_paired", "id": "en-r6000-0-q13", "haystack_uid": "en_intent_paired:en-r6000-0", "haystack_id": "en-r6000-0", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "weather_query", "candidates": ["qa_factoid", "weather_query"], "answer": "weather_query", "answer_key": "weather_query", "question": "Which are there more of in these records: records labeled 'qa_factoid' or records labeled 'weather_query'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r6000-0-q14", "dataset": "en_intent_paired", "id": "en-r6000-0-q14", "haystack_uid": "en_intent_paired:en-r6000-0", "haystack_id": "en-r6000-0", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "email_query", "label_b": "qa_currency", "candidates": ["email_query", "qa_currency"], "answer": "email_query", "answer_key": "email_query", "question": "Which are there more of in these records: records labeled 'email_query' or records labeled 'qa_currency'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent_paired:en-r6000-1-q0", "dataset": "en_intent_paired", "id": "en-r6000-1-q0", "haystack_uid": "en_intent_paired:en-r6000-1", "haystack_id": "en-r6000-1", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "audio_volume_mute", "answer": "104", "question": "How many utterances have the intent 'audio_volume_mute'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r6000-1-q1", "dataset": "en_intent_paired", "id": "en-r6000-1-q1", "haystack_uid": "en_intent_paired:en-r6000-1", "haystack_id": "en-r6000-1", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "audio_volume_up", "answer": "112", "question": "How many utterances have the intent 'audio_volume_up'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r6000-1-q2", "dataset": "en_intent_paired", "id": "en-r6000-1-q2", "haystack_uid": "en_intent_paired:en-r6000-1", "haystack_id": "en-r6000-1", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "iot_coffee", "answer": "148", "question": "How many utterances have the intent 'iot_coffee'? Answer with the number only.", "role": "control"}
@@ -104,9 +95,8 @@
 {"uid": "en_intent_paired:en-r6000-1-q9", "dataset": "en_intent_paired", "id": "en-r6000-1-q9", "haystack_uid": "en_intent_paired:en-r6000-1", "haystack_id": "en-r6000-1", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "takeaway_query", "answer": "21", "question": "How many utterances have the intent 'takeaway_query'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r6000-1-q10", "dataset": "en_intent_paired", "id": "en-r6000-1-q10", "haystack_uid": "en_intent_paired:en-r6000-1", "haystack_id": "en-r6000-1", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "takeaway_order", "answer": "9", "question": "How many utterances have the intent 'takeaway_order'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r6000-1-q11", "dataset": "en_intent_paired", "id": "en-r6000-1-q11", "haystack_uid": "en_intent_paired:en-r6000-1", "haystack_id": "en-r6000-1", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "news_query", "answer": "13", "question": "How many utterances have the intent 'news_query'? Answer with the number only.", "role": "retrieval"}
-{"uid": "en_intent_paired:en-r6000-1-q12", "dataset": "en_intent_paired", "id": "en-r6000-1-q12", "haystack_uid": "en_intent_paired:en-r6000-1", "haystack_id": "en-r6000-1", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "iot_hue_lightoff", "label_b": "weather_query", "candidates": ["iot_hue_lightoff", "weather_query"], "answer": "iot_hue_lightoff", "answer_key": "iot_hue_lightoff", "question": "Which are there more of in these records: records labeled 'iot_hue_lightoff' or records labeled 'weather_query'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r6000-1-q13", "dataset": "en_intent_paired", "id": "en-r6000-1-q13", "haystack_uid": "en_intent_paired:en-r6000-1", "haystack_id": "en-r6000-1", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "alarm_query", "label_b": "music_likeness", "candidates": ["alarm_query", "music_likeness"], "answer": "alarm_query", "answer_key": "alarm_query", "question": "Which are there more of in these records: records labeled 'alarm_query' or records labeled 'music_likeness'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r6000-1-q14", "dataset": "en_intent_paired", "id": "en-r6000-1-q14", "haystack_uid": "en_intent_paired:en-r6000-1", "haystack_id": "en-r6000-1", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "recommendation_locations", "label_b": "email_query", "candidates": ["recommendation_locations", "email_query"], "answer": "email_query", "answer_key": "email_query", "question": "Which are there more of in these records: records labeled 'recommendation_locations' or records labeled 'email_query'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r6000-1-q12", "dataset": "en_intent_paired", "id": "en-r6000-1-q12", "haystack_uid": "en_intent_paired:en-r6000-1", "haystack_id": "en-r6000-1", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "email_query", "label_b": "transport_ticket", "candidates": ["email_query", "transport_ticket"], "answer": "transport_ticket", "answer_key": "transport_ticket", "question": "Which are there more of in these records: records labeled 'email_query' or records labeled 'transport_ticket'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r6000-1-q13", "dataset": "en_intent_paired", "id": "en-r6000-1-q13", "haystack_uid": "en_intent_paired:en-r6000-1", "haystack_id": "en-r6000-1", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "weather_query", "label_b": "iot_hue_lightoff", "candidates": ["weather_query", "iot_hue_lightoff"], "answer": "iot_hue_lightoff", "answer_key": "iot_hue_lightoff", "question": "Which are there more of in these records: records labeled 'weather_query' or records labeled 'iot_hue_lightoff'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent_paired:en-r6000-2-q0", "dataset": "en_intent_paired", "id": "en-r6000-2-q0", "haystack_uid": "en_intent_paired:en-r6000-2", "haystack_id": "en-r6000-2", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "email_querycontact", "answer": "134", "question": "How many utterances have the intent 'email_querycontact'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r6000-2-q1", "dataset": "en_intent_paired", "id": "en-r6000-2-q1", "haystack_uid": "en_intent_paired:en-r6000-2", "haystack_id": "en-r6000-2", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "qa_stock", "answer": "92", "question": "How many utterances have the intent 'qa_stock'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r6000-2-q2", "dataset": "en_intent_paired", "id": "en-r6000-2-q2", "haystack_uid": "en_intent_paired:en-r6000-2", "haystack_id": "en-r6000-2", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "takeaway_order", "answer": "100", "question": "How many utterances have the intent 'takeaway_order'? Answer with the number only.", "role": "control"}
@@ -119,9 +109,7 @@
 {"uid": "en_intent_paired:en-r6000-2-q9", "dataset": "en_intent_paired", "id": "en-r6000-2-q9", "haystack_uid": "en_intent_paired:en-r6000-2", "haystack_id": "en-r6000-2", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "second_most", "label": null, "candidates": ["alarm_set", "music_query", "play_audiobook", "qa_currency", "qa_stock"], "answer": "qa_stock", "question": "Which label is the second most common in these records? Labels: 'alarm_set', 'music_query', 'play_audiobook', 'qa_currency', 'qa_stock'. Answer with the label name only.", "role": "control"}
 {"uid": "en_intent_paired:en-r6000-2-q10", "dataset": "en_intent_paired", "id": "en-r6000-2-q10", "haystack_uid": "en_intent_paired:en-r6000-2", "haystack_id": "en-r6000-2", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "label_vs_label", "label": null, "label_a": "qa_currency", "label_b": "audio_volume_mute", "candidates": ["qa_currency", "audio_volume_mute"], "answer": "less common", "answer_key": "less", "question": "Are records labeled 'qa_currency' more common, less common, or the same frequency as records labeled 'audio_volume_mute'? Answer 'more common', 'less common', or 'the same'.", "role": "control"}
 {"uid": "en_intent_paired:en-r6000-2-q11", "dataset": "en_intent_paired", "id": "en-r6000-2-q11", "haystack_uid": "en_intent_paired:en-r6000-2", "haystack_id": "en-r6000-2", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "label_vs_label", "label": null, "label_a": "qa_definition", "label_b": "lists_createoradd", "candidates": ["qa_definition", "lists_createoradd"], "answer": "less common", "answer_key": "less", "question": "Are records labeled 'qa_definition' more common, less common, or the same frequency as records labeled 'lists_createoradd'? Answer 'more common', 'less common', or 'the same'.", "role": "control"}
-{"uid": "en_intent_paired:en-r6000-2-q12", "dataset": "en_intent_paired", "id": "en-r6000-2-q12", "haystack_uid": "en_intent_paired:en-r6000-2", "haystack_id": "en-r6000-2", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "social_query", "label_b": "datetime_query", "candidates": ["social_query", "datetime_query"], "answer": "datetime_query", "answer_key": "datetime_query", "question": "Which are there more of in these records: records labeled 'social_query' or records labeled 'datetime_query'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r6000-2-q13", "dataset": "en_intent_paired", "id": "en-r6000-2-q13", "haystack_uid": "en_intent_paired:en-r6000-2", "haystack_id": "en-r6000-2", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "news_query", "label_b": "audio_volume_mute", "candidates": ["news_query", "audio_volume_mute"], "answer": "audio_volume_mute", "answer_key": "audio_volume_mute", "question": "Which are there more of in these records: records labeled 'news_query' or records labeled 'audio_volume_mute'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r6000-2-q14", "dataset": "en_intent_paired", "id": "en-r6000-2-q14", "haystack_uid": "en_intent_paired:en-r6000-2", "haystack_id": "en-r6000-2", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "alarm_set", "candidates": ["qa_factoid", "alarm_set"], "answer": "alarm_set", "answer_key": "alarm_set", "question": "Which are there more of in these records: records labeled 'qa_factoid' or records labeled 'alarm_set'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r6000-2-q12", "dataset": "en_intent_paired", "id": "en-r6000-2-q12", "haystack_uid": "en_intent_paired:en-r6000-2", "haystack_id": "en-r6000-2", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "alarm_set", "candidates": ["qa_factoid", "alarm_set"], "answer": "alarm_set", "answer_key": "alarm_set", "question": "Which are there more of in these records: records labeled 'qa_factoid' or records labeled 'alarm_set'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent_paired:en-r6000-3-q0", "dataset": "en_intent_paired", "id": "en-r6000-3-q0", "haystack_uid": "en_intent_paired:en-r6000-3", "haystack_id": "en-r6000-3", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "qa_currency", "answer": "72", "question": "How many utterances have the intent 'qa_currency'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r6000-3-q1", "dataset": "en_intent_paired", "id": "en-r6000-3-q1", "haystack_uid": "en_intent_paired:en-r6000-3", "haystack_id": "en-r6000-3", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "general_quirky", "answer": "304", "question": "How many utterances have the intent 'general_quirky'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r6000-3-q2", "dataset": "en_intent_paired", "id": "en-r6000-3-q2", "haystack_uid": "en_intent_paired:en-r6000-3", "haystack_id": "en-r6000-3", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "transport_traffic", "answer": "138", "question": "How many utterances have the intent 'transport_traffic'? Answer with the number only.", "role": "control"}
@@ -134,10 +122,9 @@
 {"uid": "en_intent_paired:en-r6000-3-q9", "dataset": "en_intent_paired", "id": "en-r6000-3-q9", "haystack_uid": "en_intent_paired:en-r6000-3", "haystack_id": "en-r6000-3", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "play_audiobook", "answer": "22", "question": "How many utterances have the intent 'play_audiobook'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r6000-3-q10", "dataset": "en_intent_paired", "id": "en-r6000-3-q10", "haystack_uid": "en_intent_paired:en-r6000-3", "haystack_id": "en-r6000-3", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "play_podcasts", "answer": "14", "question": "How many utterances have the intent 'play_podcasts'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r6000-3-q11", "dataset": "en_intent_paired", "id": "en-r6000-3-q11", "haystack_uid": "en_intent_paired:en-r6000-3", "haystack_id": "en-r6000-3", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "music_likeness", "answer": "24", "question": "How many utterances have the intent 'music_likeness'? Answer with the number only.", "role": "retrieval"}
-{"uid": "en_intent_paired:en-r6000-3-q12", "dataset": "en_intent_paired", "id": "en-r6000-3-q12", "haystack_uid": "en_intent_paired:en-r6000-3", "haystack_id": "en-r6000-3", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "iot_coffee", "label_b": "social_query", "candidates": ["iot_coffee", "social_query"], "answer": "iot_coffee", "answer_key": "iot_coffee", "question": "Which are there more of in these records: records labeled 'iot_coffee' or records labeled 'social_query'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r6000-3-q13", "dataset": "en_intent_paired", "id": "en-r6000-3-q13", "haystack_uid": "en_intent_paired:en-r6000-3", "haystack_id": "en-r6000-3", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "calendar_query", "label_b": "calendar_set", "candidates": ["calendar_query", "calendar_set"], "answer": "calendar_set", "answer_key": "calendar_set", "question": "Which are there more of in these records: records labeled 'calendar_query' or records labeled 'calendar_set'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r6000-3-q14", "dataset": "en_intent_paired", "id": "en-r6000-3-q14", "haystack_uid": "en_intent_paired:en-r6000-3", "haystack_id": "en-r6000-3", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "transport_taxi", "label_b": "calendar_remove", "candidates": ["transport_taxi", "calendar_remove"], "answer": "transport_taxi", "answer_key": "transport_taxi", "question": "Which are there more of in these records: records labeled 'transport_taxi' or records labeled 'calendar_remove'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r6000-3-q15", "dataset": "en_intent_paired", "id": "en-r6000-3-q15", "haystack_uid": "en_intent_paired:en-r6000-3", "haystack_id": "en-r6000-3", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "takeaway_query", "label_b": "email_query", "candidates": ["takeaway_query", "email_query"], "answer": "email_query", "answer_key": "email_query", "question": "Which are there more of in these records: records labeled 'takeaway_query' or records labeled 'email_query'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r6000-3-q12", "dataset": "en_intent_paired", "id": "en-r6000-3-q12", "haystack_uid": "en_intent_paired:en-r6000-3", "haystack_id": "en-r6000-3", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "calendar_set", "label_b": "calendar_query", "candidates": ["calendar_set", "calendar_query"], "answer": "calendar_set", "answer_key": "calendar_set", "question": "Which are there more of in these records: records labeled 'calendar_set' or records labeled 'calendar_query'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r6000-3-q13", "dataset": "en_intent_paired", "id": "en-r6000-3-q13", "haystack_uid": "en_intent_paired:en-r6000-3", "haystack_id": "en-r6000-3", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "transport_taxi", "label_b": "calendar_remove", "candidates": ["transport_taxi", "calendar_remove"], "answer": "transport_taxi", "answer_key": "transport_taxi", "question": "Which are there more of in these records: records labeled 'transport_taxi' or records labeled 'calendar_remove'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r6000-3-q14", "dataset": "en_intent_paired", "id": "en-r6000-3-q14", "haystack_uid": "en_intent_paired:en-r6000-3", "haystack_id": "en-r6000-3", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "email_query", "label_b": "takeaway_query", "candidates": ["email_query", "takeaway_query"], "answer": "email_query", "answer_key": "email_query", "question": "Which are there more of in these records: records labeled 'email_query' or records labeled 'takeaway_query'? Answer with the label name only.", "role": "core"}
 {"uid": "en_intent_paired:en-r6000-4-q0", "dataset": "en_intent_paired", "id": "en-r6000-4-q0", "haystack_uid": "en_intent_paired:en-r6000-4", "haystack_id": "en-r6000-4", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "social_query", "answer": "45", "question": "How many utterances have the intent 'social_query'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r6000-4-q1", "dataset": "en_intent_paired", "id": "en-r6000-4-q1", "haystack_uid": "en_intent_paired:en-r6000-4", "haystack_id": "en-r6000-4", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "iot_hue_lightchange", "answer": "54", "question": "How many utterances have the intent 'iot_hue_lightchange'? Answer with the number only.", "role": "control"}
 {"uid": "en_intent_paired:en-r6000-4-q2", "dataset": "en_intent_paired", "id": "en-r6000-4-q2", "haystack_uid": "en_intent_paired:en-r6000-4", "haystack_id": "en-r6000-4", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "iot_hue_lightoff", "answer": "15", "question": "How many utterances have the intent 'iot_hue_lightoff'? Answer with the number only.", "role": "control"}
@@ -150,23 +137,22 @@
 {"uid": "en_intent_paired:en-r6000-4-q9", "dataset": "en_intent_paired", "id": "en-r6000-4-q9", "haystack_uid": "en_intent_paired:en-r6000-4", "haystack_id": "en-r6000-4", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "iot_hue_lightup", "answer": "30", "question": "How many utterances have the intent 'iot_hue_lightup'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r6000-4-q10", "dataset": "en_intent_paired", "id": "en-r6000-4-q10", "haystack_uid": "en_intent_paired:en-r6000-4", "haystack_id": "en-r6000-4", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "cooking_recipe", "answer": "29", "question": "How many utterances have the intent 'cooking_recipe'? Answer with the number only.", "role": "retrieval"}
 {"uid": "en_intent_paired:en-r6000-4-q11", "dataset": "en_intent_paired", "id": "en-r6000-4-q11", "haystack_uid": "en_intent_paired:en-r6000-4", "haystack_id": "en-r6000-4", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "calendar_set", "answer": "22", "question": "How many utterances have the intent 'calendar_set'? Answer with the number only.", "role": "retrieval"}
-{"uid": "en_intent_paired:en-r6000-4-q12", "dataset": "en_intent_paired", "id": "en-r6000-4-q12", "haystack_uid": "en_intent_paired:en-r6000-4", "haystack_id": "en-r6000-4", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "qa_currency", "label_b": "play_game", "candidates": ["qa_currency", "play_game"], "answer": "qa_currency", "answer_key": "qa_currency", "question": "Which are there more of in these records: records labeled 'qa_currency' or records labeled 'play_game'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r6000-4-q13", "dataset": "en_intent_paired", "id": "en-r6000-4-q13", "haystack_uid": "en_intent_paired:en-r6000-4", "haystack_id": "en-r6000-4", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "play_music", "label_b": "weather_query", "candidates": ["play_music", "weather_query"], "answer": "play_music", "answer_key": "play_music", "question": "Which are there more of in these records: records labeled 'play_music' or records labeled 'weather_query'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-r6000-4-q14", "dataset": "en_intent_paired", "id": "en-r6000-4-q14", "haystack_uid": "en_intent_paired:en-r6000-4", "haystack_id": "en-r6000-4", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "lists_query", "label_b": "lists_remove", "candidates": ["lists_query", "lists_remove"], "answer": "lists_query", "answer_key": "lists_query", "question": "Which are there more of in these records: records labeled 'lists_query' or records labeled 'lists_remove'? Answer with the label name only.", "role": "core"}
-{"uid": "en_intent_paired:en-core3000-0-q0", "dataset": "en_intent_paired", "id": "en-core3000-0-q0", "haystack_uid": "en_intent_paired:en-core3000-0", "haystack_id": "en-core3000-0", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_query", "label_b": "lists_createoradd", "candidates": ["lists_query", "lists_createoradd"], "answer": "lists_query", "answer_key": "lists_query", "question": "Which are there more of in these records: records labeled 'lists_query' or records labeled 'lists_createoradd'? Answer with the label name only.", "role": "core", "core_document": true}
-{"uid": "en_intent_paired:en-core3000-0-q1", "dataset": "en_intent_paired", "id": "en-core3000-0-q1", "haystack_uid": "en_intent_paired:en-core3000-0", "haystack_id": "en-core3000-0", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "email_query", "label_b": "weather_query", "candidates": ["email_query", "weather_query"], "answer": "weather_query", "answer_key": "weather_query", "question": "Which are there more of in these records: records labeled 'email_query' or records labeled 'weather_query'? Answer with the label name only.", "role": "core", "core_document": true}
-{"uid": "en_intent_paired:en-core3000-1-q0", "dataset": "en_intent_paired", "id": "en-core3000-1-q0", "haystack_uid": "en_intent_paired:en-core3000-1", "haystack_id": "en-core3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_query", "label_b": "news_query", "candidates": ["lists_query", "news_query"], "answer": "lists_query", "answer_key": "lists_query", "question": "Which are there more of in these records: records labeled 'lists_query' or records labeled 'news_query'? Answer with the label name only.", "role": "core", "core_document": true}
-{"uid": "en_intent_paired:en-core3000-1-q1", "dataset": "en_intent_paired", "id": "en-core3000-1-q1", "haystack_uid": "en_intent_paired:en-core3000-1", "haystack_id": "en-core3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_set", "label_b": "calendar_query", "candidates": ["calendar_set", "calendar_query"], "answer": "calendar_query", "answer_key": "calendar_query", "question": "Which are there more of in these records: records labeled 'calendar_set' or records labeled 'calendar_query'? Answer with the label name only.", "role": "core", "core_document": true}
-{"uid": "en_intent_paired:en-core3000-2-q0", "dataset": "en_intent_paired", "id": "en-core3000-2-q0", "haystack_uid": "en_intent_paired:en-core3000-2", "haystack_id": "en-core3000-2", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_createoradd", "label_b": "recommendation_locations", "candidates": ["lists_createoradd", "recommendation_locations"], "answer": "lists_createoradd", "answer_key": "lists_createoradd", "question": "Which are there more of in these records: records labeled 'lists_createoradd' or records labeled 'recommendation_locations'? Answer with the label name only.", "role": "core", "core_document": true}
-{"uid": "en_intent_paired:en-core3000-2-q1", "dataset": "en_intent_paired", "id": "en-core3000-2-q1", "haystack_uid": "en_intent_paired:en-core3000-2", "haystack_id": "en-core3000-2", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_query", "label_b": "transport_query", "candidates": ["lists_query", "transport_query"], "answer": "transport_query", "answer_key": "transport_query", "question": "Which are there more of in these records: records labeled 'lists_query' or records labeled 'transport_query'? Answer with the label name only.", "role": "core", "core_document": true}
-{"uid": "en_intent_paired:en-core3000-3-q0", "dataset": "en_intent_paired", "id": "en-core3000-3-q0", "haystack_uid": "en_intent_paired:en-core3000-3", "haystack_id": "en-core3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "recommendation_events", "label_b": "datetime_query", "candidates": ["recommendation_events", "datetime_query"], "answer": "recommendation_events", "answer_key": "recommendation_events", "question": "Which are there more of in these records: records labeled 'recommendation_events' or records labeled 'datetime_query'? Answer with the label name only.", "role": "core", "core_document": true}
-{"uid": "en_intent_paired:en-core3000-3-q1", "dataset": "en_intent_paired", "id": "en-core3000-3-q1", "haystack_uid": "en_intent_paired:en-core3000-3", "haystack_id": "en-core3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "play_radio", "label_b": "lists_query", "candidates": ["play_radio", "lists_query"], "answer": "lists_query", "answer_key": "lists_query", "question": "Which are there more of in these records: records labeled 'play_radio' or records labeled 'lists_query'? Answer with the label name only.", "role": "core", "core_document": true}
+{"uid": "en_intent_paired:en-r6000-4-q12", "dataset": "en_intent_paired", "id": "en-r6000-4-q12", "haystack_uid": "en_intent_paired:en-r6000-4", "haystack_id": "en-r6000-4", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "weather_query", "label_b": "play_music", "candidates": ["weather_query", "play_music"], "answer": "play_music", "answer_key": "play_music", "question": "Which are there more of in these records: records labeled 'weather_query' or records labeled 'play_music'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-r6000-4-q13", "dataset": "en_intent_paired", "id": "en-r6000-4-q13", "haystack_uid": "en_intent_paired:en-r6000-4", "haystack_id": "en-r6000-4", "language": "en", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "qa_currency", "label_b": "email_sendemail", "candidates": ["qa_currency", "email_sendemail"], "answer": "email_sendemail", "answer_key": "email_sendemail", "question": "Which are there more of in these records: records labeled 'qa_currency' or records labeled 'email_sendemail'? Answer with the label name only.", "role": "core"}
+{"uid": "en_intent_paired:en-core3000-0-q0", "dataset": "en_intent_paired", "id": "en-core3000-0-q0", "haystack_uid": "en_intent_paired:en-core3000-0", "haystack_id": "en-core3000-0", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "email_query", "label_b": "lists_createoradd", "candidates": ["email_query", "lists_createoradd"], "answer": "email_query", "answer_key": "email_query", "question": "Which are there more of in these records: records labeled 'email_query' or records labeled 'lists_createoradd'? Answer with the label name only.", "role": "core", "core_document": true}
+{"uid": "en_intent_paired:en-core3000-0-q1", "dataset": "en_intent_paired", "id": "en-core3000-0-q1", "haystack_uid": "en_intent_paired:en-core3000-0", "haystack_id": "en-core3000-0", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "social_post", "label_b": "weather_query", "candidates": ["social_post", "weather_query"], "answer": "weather_query", "answer_key": "weather_query", "question": "Which are there more of in these records: records labeled 'social_post' or records labeled 'weather_query'? Answer with the label name only.", "role": "core", "core_document": true}
+{"uid": "en_intent_paired:en-core3000-1-q0", "dataset": "en_intent_paired", "id": "en-core3000-1-q0", "haystack_uid": "en_intent_paired:en-core3000-1", "haystack_id": "en-core3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "news_query", "label_b": "calendar_set", "candidates": ["news_query", "calendar_set"], "answer": "news_query", "answer_key": "news_query", "question": "Which are there more of in these records: records labeled 'news_query' or records labeled 'calendar_set'? Answer with the label name only.", "role": "core", "core_document": true}
+{"uid": "en_intent_paired:en-core3000-1-q1", "dataset": "en_intent_paired", "id": "en-core3000-1-q1", "haystack_uid": "en_intent_paired:en-core3000-1", "haystack_id": "en-core3000-1", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_query", "label_b": "calendar_remove", "candidates": ["calendar_query", "calendar_remove"], "answer": "calendar_remove", "answer_key": "calendar_remove", "question": "Which are there more of in these records: records labeled 'calendar_query' or records labeled 'calendar_remove'? Answer with the label name only.", "role": "core", "core_document": true}
+{"uid": "en_intent_paired:en-core3000-2-q0", "dataset": "en_intent_paired", "id": "en-core3000-2-q0", "haystack_uid": "en_intent_paired:en-core3000-2", "haystack_id": "en-core3000-2", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "transport_query", "label_b": "lists_createoradd", "candidates": ["transport_query", "lists_createoradd"], "answer": "transport_query", "answer_key": "transport_query", "question": "Which are there more of in these records: records labeled 'transport_query' or records labeled 'lists_createoradd'? Answer with the label name only.", "role": "core", "core_document": true}
+{"uid": "en_intent_paired:en-core3000-2-q1", "dataset": "en_intent_paired", "id": "en-core3000-2-q1", "haystack_uid": "en_intent_paired:en-core3000-2", "haystack_id": "en-core3000-2", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "play_radio", "label_b": "calendar_set", "candidates": ["play_radio", "calendar_set"], "answer": "calendar_set", "answer_key": "calendar_set", "question": "Which are there more of in these records: records labeled 'play_radio' or records labeled 'calendar_set'? Answer with the label name only.", "role": "core", "core_document": true}
+{"uid": "en_intent_paired:en-core3000-3-q0", "dataset": "en_intent_paired", "id": "en-core3000-3-q0", "haystack_uid": "en_intent_paired:en-core3000-3", "haystack_id": "en-core3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "play_radio", "label_b": "datetime_query", "candidates": ["play_radio", "datetime_query"], "answer": "play_radio", "answer_key": "play_radio", "question": "Which are there more of in these records: records labeled 'play_radio' or records labeled 'datetime_query'? Answer with the label name only.", "role": "core", "core_document": true}
+{"uid": "en_intent_paired:en-core3000-3-q1", "dataset": "en_intent_paired", "id": "en-core3000-3-q1", "haystack_uid": "en_intent_paired:en-core3000-3", "haystack_id": "en-core3000-3", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_query", "label_b": "cooking_recipe", "candidates": ["calendar_query", "cooking_recipe"], "answer": "cooking_recipe", "answer_key": "cooking_recipe", "question": "Which are there more of in these records: records labeled 'calendar_query' or records labeled 'cooking_recipe'? Answer with the label name only.", "role": "core", "core_document": true}
 {"uid": "en_intent_paired:en-core3000-4-q0", "dataset": "en_intent_paired", "id": "en-core3000-4-q0", "haystack_uid": "en_intent_paired:en-core3000-4", "haystack_id": "en-core3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "social_post", "label_b": "transport_query", "candidates": ["social_post", "transport_query"], "answer": "social_post", "answer_key": "social_post", "question": "Which are there more of in these records: records labeled 'social_post' or records labeled 'transport_query'? Answer with the label name only.", "role": "core", "core_document": true}
 {"uid": "en_intent_paired:en-core3000-4-q1", "dataset": "en_intent_paired", "id": "en-core3000-4-q1", "haystack_uid": "en_intent_paired:en-core3000-4", "haystack_id": "en-core3000-4", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "alarm_set", "label_b": "calendar_set", "candidates": ["alarm_set", "calendar_set"], "answer": "calendar_set", "answer_key": "calendar_set", "question": "Which are there more of in these records: records labeled 'alarm_set' or records labeled 'calendar_set'? Answer with the label name only.", "role": "core", "core_document": true}
 {"uid": "en_intent_paired:en-core3000-5-q0", "dataset": "en_intent_paired", "id": "en-core3000-5-q0", "haystack_uid": "en_intent_paired:en-core3000-5", "haystack_id": "en-core3000-5", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_remove", "label_b": "weather_query", "candidates": ["calendar_remove", "weather_query"], "answer": "calendar_remove", "answer_key": "calendar_remove", "question": "Which are there more of in these records: records labeled 'calendar_remove' or records labeled 'weather_query'? Answer with the label name only.", "role": "core", "core_document": true}
 {"uid": "en_intent_paired:en-core3000-5-q1", "dataset": "en_intent_paired", "id": "en-core3000-5-q1", "haystack_uid": "en_intent_paired:en-core3000-5", "haystack_id": "en-core3000-5", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "news_query", "label_b": "cooking_recipe", "candidates": ["news_query", "cooking_recipe"], "answer": "cooking_recipe", "answer_key": "cooking_recipe", "question": "Which are there more of in these records: records labeled 'news_query' or records labeled 'cooking_recipe'? Answer with the label name only.", "role": "core", "core_document": true}
-{"uid": "en_intent_paired:en-core3000-6-q0", "dataset": "en_intent_paired", "id": "en-core3000-6-q0", "haystack_uid": "en_intent_paired:en-core3000-6", "haystack_id": "en-core3000-6", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "general_quirky", "label_b": "qa_definition", "candidates": ["general_quirky", "qa_definition"], "answer": "general_quirky", "answer_key": "general_quirky", "question": "Which are there more of in these records: records labeled 'general_quirky' or records labeled 'qa_definition'? Answer with the label name only.", "role": "core", "core_document": true}
-{"uid": "en_intent_paired:en-core3000-6-q1", "dataset": "en_intent_paired", "id": "en-core3000-6-q1", "haystack_uid": "en_intent_paired:en-core3000-6", "haystack_id": "en-core3000-6", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "play_podcasts", "label_b": "lists_query", "candidates": ["play_podcasts", "lists_query"], "answer": "lists_query", "answer_key": "lists_query", "question": "Which are there more of in these records: records labeled 'play_podcasts' or records labeled 'lists_query'? Answer with the label name only.", "role": "core", "core_document": true}
+{"uid": "en_intent_paired:en-core3000-6-q0", "dataset": "en_intent_paired", "id": "en-core3000-6-q0", "haystack_uid": "en_intent_paired:en-core3000-6", "haystack_id": "en-core3000-6", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "qa_definition", "label_b": "play_podcasts", "candidates": ["qa_definition", "play_podcasts"], "answer": "qa_definition", "answer_key": "qa_definition", "question": "Which are there more of in these records: records labeled 'qa_definition' or records labeled 'play_podcasts'? Answer with the label name only.", "role": "core", "core_document": true}
+{"uid": "en_intent_paired:en-core3000-6-q1", "dataset": "en_intent_paired", "id": "en-core3000-6-q1", "haystack_uid": "en_intent_paired:en-core3000-6", "haystack_id": "en-core3000-6", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_createoradd", "label_b": "news_query", "candidates": ["lists_createoradd", "news_query"], "answer": "news_query", "answer_key": "news_query", "question": "Which are there more of in these records: records labeled 'lists_createoradd' or records labeled 'news_query'? Answer with the label name only.", "role": "core", "core_document": true}
 {"uid": "en_intent_paired:en-core3000-7-q0", "dataset": "en_intent_paired", "id": "en-core3000-7-q0", "haystack_uid": "en_intent_paired:en-core3000-7", "haystack_id": "en-core3000-7", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_remove", "label_b": "email_query", "candidates": ["calendar_remove", "email_query"], "answer": "calendar_remove", "answer_key": "calendar_remove", "question": "Which are there more of in these records: records labeled 'calendar_remove' or records labeled 'email_query'? Answer with the label name only.", "role": "core", "core_document": true}
 {"uid": "en_intent_paired:en-core3000-7-q1", "dataset": "en_intent_paired", "id": "en-core3000-7-q1", "haystack_uid": "en_intent_paired:en-core3000-7", "haystack_id": "en-core3000-7", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_set", "label_b": "alarm_set", "candidates": ["calendar_set", "alarm_set"], "answer": "alarm_set", "answer_key": "alarm_set", "question": "Which are there more of in these records: records labeled 'calendar_set' or records labeled 'alarm_set'? Answer with the label name only.", "role": "core", "core_document": true}
 {"uid": "en_intent_paired:en-core3000-8-q0", "dataset": "en_intent_paired", "id": "en-core3000-8-q0", "haystack_uid": "en_intent_paired:en-core3000-8", "haystack_id": "en-core3000-8", "language": "en", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_set", "label_b": "play_podcasts", "candidates": ["calendar_set", "play_podcasts"], "answer": "calendar_set", "answer_key": "calendar_set", "question": "Which are there more of in these records: records labeled 'calendar_set' or records labeled 'play_podcasts'? Answer with the label name only.", "role": "core", "core_document": true}

interpress_tr/manifest.json

@@ -1,6 +1,6 @@
 {
-  "version": "0.9.0",
-  "date": "2026-09-29",
+  "version": "0.10.0",
+  "date": "2026-10-01",
   "environment": {
     "python": "3.12.9",
     "polars": "1.42.0",
@@ -17,6 +17,39 @@
     "drop_label_leakage": true,
     "leak_label_words": false,
     "min_class_support": 1000,
+    "boilerplate_keywords": [
+      "imtiyaz sahibi",
+      "yazı işleri",
+      "sorumlu müdür",
+      "genel yayın yönetmeni",
+      "yayın sahibi",
+      "reklam müdürü",
+      "görsel yönetmen",
+      "yönetim yeri",
+      "baskı",
+      "issn",
+      "genel koordinatör",
+      "yayın direktörü",
+      "muhabir",
+      "iktibas",
+      "yayın kurulu",
+      "yayın türü",
+      "basım yeri",
+      "tüzel kişi temsilcisi",
+      "satış fiyatı",
+      "yayın koordinatörü",
+      "yayın danışmanı",
+      "reklam grup",
+      "yazıişleri",
+      "yazı işlen",
+      "haber editörü",
+      "görsel yayın"
+    ],
+    "boilerplate_min_keywords": 3,
+    "boilerplate_weak_min": 2,
+    "boilerplate_regex": "y[ıi]l\\s*:?\\s*\\d+.{0,40}?say[ıi]\\s*:?\\s*\\d+",
+    "drop_text_regex": [],
+    "strip_html": true,
     "seed": 42,
     "haystack_target_tokens": [
       100000,
@@ -72,6 +105,7 @@
     "close_min_diff": 2,
     "close_exclude_labels": [],
     "close_ordinal_gap": 0,
+    "close_allowed_pairs_file": "experiments/separable_pairs/interpress_tr.json",
     "core_doc_records": [
       500,
       1000,
@@ -93,11 +127,12 @@
     "out_dir": "interpress_tr_out",
     "dataset": ""
   },
-  "source_hash_first1000": "0b76e9b072b608b3",
-  "rows_after_cleaning": 97835,
+  "source_hash_first1000": "46d7bd7a18407ae6",
+  "rows_after_cleaning": 95321,
   "cleaning": {
-    "rows_dropped_label_leakage": 53192,
-    "label_leakage_rate": 0.350503,
+    "rows_dropped_boilerplate": 4185,
+    "rows_dropped_label_leakage": 51567,
+    "label_leakage_rate": 0.349413,
     "classes_dropped_low_support": [
       "savunma"
     ],
@@ -111,45 +146,45 @@
   "tier_overlap": {
     "100000": {
       "n_haystacks": 10,
-      "mean_jaccard": 0.0021,
-      "max_jaccard": 0.0139,
-      "pool_fraction_per_haystack": 0.0026
+      "mean_jaccard": 0.0024,
+      "max_jaccard": 0.0124,
+      "pool_fraction_per_haystack": 0.0027
     },
     "1000000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.0176,
-      "max_jaccard": 0.0258,
-      "pool_fraction_per_haystack": 0.0275
+      "mean_jaccard": 0.0193,
+      "max_jaccard": 0.024,
+      "pool_fraction_per_haystack": 0.0287
     },
     "250000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.0038,
-      "max_jaccard": 0.0064,
-      "pool_fraction_per_haystack": 0.0066
+      "mean_jaccard": 0.0047,
+      "max_jaccard": 0.0136,
+      "pool_fraction_per_haystack": 0.0067
     },
     "500000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.0068,
-      "max_jaccard": 0.0119,
-      "pool_fraction_per_haystack": 0.0133
+      "mean_jaccard": 0.0079,
+      "max_jaccard": 0.0172,
+      "pool_fraction_per_haystack": 0.0136
     },
     "core1000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0047,
-      "max_jaccard": 0.0116,
-      "pool_fraction_per_haystack": 0.0102
+      "mean_jaccard": 0.0076,
+      "max_jaccard": 0.0168,
+      "pool_fraction_per_haystack": 0.0105
     },
     "core2500": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0147,
-      "max_jaccard": 0.0246,
-      "pool_fraction_per_haystack": 0.0256
+      "mean_jaccard": 0.018,
+      "max_jaccard": 0.0252,
+      "pool_fraction_per_haystack": 0.0262
     },
     "core500": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0025,
-      "max_jaccard": 0.005,
-      "pool_fraction_per_haystack": 0.0051
+      "mean_jaccard": 0.0035,
+      "max_jaccard": 0.007,
+      "pool_fraction_per_haystack": 0.0052
     }
   },
   "max_pool_fraction": 0.35,
@@ -159,14 +194,14 @@
   "proportion_unit": "per_mille",
   "top_k_k": 3,
   "tokenizer": "Qwen/Qwen3-8B",
-  "questions_written": 345,
+  "questions_written": 337,
   "kind_distribution": {
-    "close_comparison": 45,
-    "count": 142,
-    "label_vs_label": 79,
-    "proportion": 79
+    "close_comparison": 37,
+    "count": 144,
+    "label_vs_label": 78,
+    "proportion": 78
   },
-  "rare_count_questions": 59,
+  "rare_count_questions": 58,
   "rare_band": [
     5,
     30
@@ -174,250 +209,250 @@
   "haystacks": [
     {
       "haystack_id": "tr-100000-0",
-      "n_examples": 249,
-      "n_tokens": 99589,
-      "n_chars": 292837,
-      "drift_target": "aktuel",
+      "n_examples": 252,
+      "n_tokens": 99638,
+      "n_chars": 295457,
+      "drift_target": "iletisim",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-1",
-      "n_examples": 266,
-      "n_tokens": 99691,
-      "n_chars": 293006,
-      "drift_target": "bilisim",
+      "n_examples": 280,
+      "n_tokens": 99652,
+      "n_chars": 289040,
+      "drift_target": "yasam",
       "drift_ok": true,
-      "n_questions": 13
+      "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-2",
-      "n_examples": 292,
-      "n_tokens": 99808,
-      "n_chars": 291147,
-      "drift_target": "ticaret",
+      "n_examples": 258,
+      "n_tokens": 99731,
+      "n_chars": 292037,
+      "drift_target": "bilisim",
       "drift_ok": true,
-      "n_questions": 13
+      "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-3",
-      "n_examples": 256,
-      "n_tokens": 99675,
-      "n_chars": 288624,
-      "drift_target": "iletisim",
+      "n_examples": 251,
+      "n_tokens": 99091,
+      "n_chars": 293126,
+      "drift_target": "ekonomi",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-4",
-      "n_examples": 234,
-      "n_tokens": 99844,
-      "n_chars": 287019,
-      "drift_target": "saglik",
+      "n_examples": 241,
+      "n_tokens": 99604,
+      "n_chars": 286773,
+      "drift_target": "kultursanat",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-5",
-      "n_examples": 256,
-      "n_tokens": 99564,
-      "n_chars": 292718,
-      "drift_target": "siyasi",
+      "n_examples": 259,
+      "n_tokens": 99655,
+      "n_chars": 295233,
+      "drift_target": "saglik",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-6",
-      "n_examples": 255,
-      "n_tokens": 99615,
-      "n_chars": 296718,
-      "drift_target": "saglik",
+      "n_examples": 240,
+      "n_tokens": 99830,
+      "n_chars": 302327,
+      "drift_target": "iletisim",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-7",
-      "n_examples": 268,
-      "n_tokens": 99460,
-      "n_chars": 283318,
-      "drift_target": "siyasi",
+      "n_examples": 265,
+      "n_tokens": 99530,
+      "n_chars": 289812,
+      "drift_target": "turizm",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-8",
-      "n_examples": 255,
-      "n_tokens": 99650,
-      "n_chars": 299106,
+      "n_examples": 237,
+      "n_tokens": 99425,
+      "n_chars": 297785,
       "drift_target": "aktuel",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-9",
-      "n_examples": 257,
-      "n_tokens": 99267,
-      "n_chars": 289252,
-      "drift_target": "magazin",
+      "n_examples": 271,
+      "n_tokens": 99443,
+      "n_chars": 292971,
+      "drift_target": "siyasi",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-250000-0",
-      "n_examples": 651,
-      "n_tokens": 249456,
-      "n_chars": 718219,
-      "drift_target": "siyasi",
+      "n_examples": 681,
+      "n_tokens": 249396,
+      "n_chars": 726984,
+      "drift_target": "turizm",
       "drift_ok": true,
-      "n_questions": 13
+      "n_questions": 12
     },
     {
       "haystack_id": "tr-250000-1",
-      "n_examples": 655,
-      "n_tokens": 249448,
-      "n_chars": 731302,
-      "drift_target": "saglik",
+      "n_examples": 625,
+      "n_tokens": 249602,
+      "n_chars": 739705,
+      "drift_target": "turizm",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-250000-2",
-      "n_examples": 614,
-      "n_tokens": 249139,
-      "n_chars": 723849,
-      "drift_target": "iletisim",
+      "n_examples": 639,
+      "n_tokens": 249563,
+      "n_chars": 742279,
+      "drift_target": "teknoloji",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-250000-3",
-      "n_examples": 683,
-      "n_tokens": 249448,
-      "n_chars": 724626,
-      "drift_target": "gida",
+      "n_examples": 659,
+      "n_tokens": 248704,
+      "n_chars": 731016,
+      "drift_target": "kultursanat",
       "drift_ok": true,
-      "n_questions": 14
+      "n_questions": 12
     },
     {
       "haystack_id": "tr-250000-4",
-      "n_examples": 618,
-      "n_tokens": 249092,
-      "n_chars": 732635,
-      "drift_target": "saglik",
+      "n_examples": 588,
+      "n_tokens": 249349,
+      "n_chars": 734366,
+      "drift_target": "spor",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-500000-0",
-      "n_examples": 1225,
-      "n_tokens": 499218,
-      "n_chars": 1471509,
-      "drift_target": "ekonomi",
+      "n_examples": 1233,
+      "n_tokens": 498821,
+      "n_chars": 1480719,
+      "drift_target": "saglik",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-500000-1",
-      "n_examples": 1348,
-      "n_tokens": 498841,
-      "n_chars": 1453103,
-      "drift_target": "gida",
+      "n_examples": 1387,
+      "n_tokens": 499093,
+      "n_chars": 1470323,
+      "drift_target": "saglik",
       "drift_ok": true,
-      "n_questions": 14
+      "n_questions": 12
     },
     {
       "haystack_id": "tr-500000-2",
-      "n_examples": 1276,
-      "n_tokens": 498977,
-      "n_chars": 1464449,
-      "drift_target": "iletisim",
+      "n_examples": 1285,
+      "n_tokens": 499190,
+      "n_chars": 1486749,
+      "drift_target": "turizm",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-500000-3",
-      "n_examples": 1292,
-      "n_tokens": 499112,
-      "n_chars": 1466421,
-      "drift_target": "kultursanat",
+      "n_examples": 1267,
+      "n_tokens": 498913,
+      "n_chars": 1475255,
+      "drift_target": "ekonomi",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-500000-4",
-      "n_examples": 1358,
-      "n_tokens": 498556,
-      "n_chars": 1451250,
-      "drift_target": "gida",
+      "n_examples": 1317,
+      "n_tokens": 498522,
+      "n_chars": 1466946,
+      "drift_target": "magazin",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-1000000-0",
-      "n_examples": 2638,
-      "n_tokens": 998392,
-      "n_chars": 2919551,
-      "drift_target": "magazin",
+      "n_examples": 2703,
+      "n_tokens": 998305,
+      "n_chars": 2961763,
+      "drift_target": "saglik",
       "drift_ok": true,
       "n_questions": 13
     },
     {
       "haystack_id": "tr-1000000-1",
-      "n_examples": 2839,
-      "n_tokens": 998213,
-      "n_chars": 2907205,
+      "n_examples": 2941,
+      "n_tokens": 997667,
+      "n_chars": 2939366,
       "drift_target": "gida",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-1000000-2",
-      "n_examples": 2577,
-      "n_tokens": 998268,
-      "n_chars": 2921877,
-      "drift_target": "iletisim",
+      "n_examples": 2586,
+      "n_tokens": 998194,
+      "n_chars": 2965788,
+      "drift_target": "egitim",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-1000000-3",
-      "n_examples": 2890,
-      "n_tokens": 997955,
-      "n_chars": 2887867,
-      "drift_target": "spor",
+      "n_examples": 2896,
+      "n_tokens": 997561,
+      "n_chars": 2918148,
+      "drift_target": "siyasi",
       "drift_ok": true,
       "n_questions": 12
     },
     {
       "haystack_id": "tr-1000000-4",
-      "n_examples": 2490,
-      "n_tokens": 998160,
-      "n_chars": 2960737,
-      "drift_target": "egitim",
+      "n_examples": 2547,
+      "n_tokens": 997466,
+      "n_chars": 3007131,
+      "drift_target": "iletisim",
       "drift_ok": true,
-      "n_questions": 13
+      "n_questions": 12
     },
     {
       "haystack_id": "tr-core500-0",
       "n_examples": 500,
-      "n_tokens": 194411,
-      "n_chars": 569047,
+      "n_tokens": 182928,
+      "n_chars": 542640,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
-        "aktuel": 75,
-        "bilisim": 100,
-        "egitim": 22,
-        "ekonomi": 1,
-        "iletisim": 70,
-        "kultursanat": 70,
-        "magazin": 21,
-        "saglik": 5,
-        "seyahat": 27,
-        "siyasi": 7,
+        "aktuel": 100,
+        "bilisim": 22,
+        "egitim": 1,
+        "gida": 75,
+        "iletisim": 21,
+        "kultursanat": 5,
+        "magazin": 27,
+        "saglik": 70,
+        "seyahat": 7,
         "spor": 75,
+        "teknoloji": 70,
         "ticaret": 1,
         "turizm": 6,
         "yasam": 20
@@ -426,8 +461,8 @@
     {
       "haystack_id": "tr-core500-1",
       "n_examples": 500,
-      "n_tokens": 191237,
-      "n_chars": 554595,
+      "n_tokens": 175325,
+      "n_chars": 517118,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -437,35 +472,35 @@
         "bilisim": 4,
         "egitim": 1,
         "ekonomi": 8,
-        "gida": 22,
-        "iletisim": 3,
-        "kultursanat": 75,
-        "saglik": 1,
-        "siyasi": 163,
-        "spor": 75,
-        "ticaret": 70,
-        "yasam": 70
+        "gida": 75,
+        "iletisim": 22,
+        "kultursanat": 3,
+        "magazin": 70,
+        "saglik": 75,
+        "siyasi": 1,
+        "spor": 70,
+        "ticaret": 163
       }
     },
     {
       "haystack_id": "tr-core500-2",
       "n_examples": 500,
-      "n_tokens": 199549,
-      "n_chars": 574836,
+      "n_tokens": 184136,
+      "n_chars": 540176,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
-        "bilisim": 70,
-        "ekonomi": 75,
-        "gida": 5,
-        "kultursanat": 70,
-        "magazin": 46,
-        "saglik": 19,
-        "seyahat": 13,
-        "siyasi": 9,
+        "egitim": 5,
+        "gida": 46,
+        "iletisim": 19,
+        "kultursanat": 13,
+        "magazin": 75,
+        "saglik": 70,
+        "seyahat": 9,
         "spor": 75,
+        "teknoloji": 70,
         "ticaret": 105,
         "turizm": 12,
         "yasam": 1
@@ -474,23 +509,23 @@
     {
       "haystack_id": "tr-core500-3",
       "n_examples": 500,
-      "n_tokens": 197768,
-      "n_chars": 577699,
+      "n_tokens": 187242,
+      "n_chars": 550678,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
-        "aktuel": 70,
+        "aktuel": 1,
         "bilisim": 1,
         "egitim": 1,
-        "ekonomi": 1,
-        "gida": 13,
+        "ekonomi": 13,
+        "gida": 75,
         "iletisim": 7,
         "kultursanat": 64,
-        "magazin": 9,
-        "saglik": 75,
-        "siyasi": 70,
+        "magazin": 70,
+        "saglik": 70,
+        "seyahat": 9,
         "spor": 13,
         "teknoloji": 75,
         "turizm": 100,
@@ -500,8 +535,8 @@
     {
       "haystack_id": "tr-core500-4",
       "n_examples": 500,
-      "n_tokens": 194037,
-      "n_chars": 567414,
+      "n_tokens": 176906,
+      "n_chars": 518540,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -510,13 +545,13 @@
         "aktuel": 7,
         "bilisim": 8,
         "egitim": 42,
-        "ekonomi": 75,
-        "iletisim": 70,
-        "kultursanat": 1,
+        "gida": 70,
+        "iletisim": 1,
         "magazin": 75,
         "saglik": 70,
-        "siyasi": 70,
-        "spor": 7,
+        "seyahat": 70,
+        "siyasi": 7,
+        "spor": 75,
         "teknoloji": 11,
         "turizm": 31,
         "yasam": 33
@@ -525,8 +560,8 @@
     {
       "haystack_id": "tr-core500-5",
       "n_examples": 500,
-      "n_tokens": 197723,
-      "n_chars": 579393,
+      "n_tokens": 179708,
+      "n_chars": 531430,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -542,9 +577,9 @@
         "magazin": 3,
         "saglik": 75,
         "seyahat": 1,
-        "spor": 11,
+        "spor": 70,
         "teknoloji": 75,
-        "ticaret": 70,
+        "ticaret": 11,
         "turizm": 1,
         "yasam": 135
       }
@@ -552,8 +587,8 @@
     {
       "haystack_id": "tr-core1000-0",
       "n_examples": 1000,
-      "n_tokens": 395747,
-      "n_chars": 1159064,
+      "n_tokens": 368608,
+      "n_chars": 1081726,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -562,15 +597,15 @@
         "aktuel": 31,
         "bilisim": 3,
         "egitim": 2,
-        "ekonomi": 144,
+        "ekonomi": 1,
         "gida": 150,
         "iletisim": 1,
-        "kultursanat": 1,
-        "magazin": 87,
+        "kultursanat": 87,
+        "magazin": 150,
         "saglik": 145,
         "seyahat": 133,
         "spor": 2,
-        "teknoloji": 150,
+        "teknoloji": 144,
         "ticaret": 150,
         "turizm": 1
       }
@@ -578,8 +613,8 @@
     {
       "haystack_id": "tr-core1000-1",
       "n_examples": 1000,
-      "n_tokens": 356939,
-      "n_chars": 1035679,
+      "n_tokens": 358667,
+      "n_chars": 1061081,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -588,37 +623,37 @@
         "aktuel": 9,
         "egitim": 153,
         "ekonomi": 25,
-        "iletisim": 150,
-        "kultursanat": 5,
+        "iletisim": 5,
+        "kultursanat": 8,
         "magazin": 145,
-        "saglik": 8,
-        "siyasi": 144,
-        "spor": 126,
-        "teknoloji": 59,
-        "ticaret": 26,
-        "turizm": 150
+        "saglik": 150,
+        "siyasi": 126,
+        "spor": 144,
+        "teknoloji": 150,
+        "ticaret": 59,
+        "turizm": 26
       }
     },
     {
       "haystack_id": "tr-core1000-2",
       "n_examples": 1000,
-      "n_tokens": 423706,
-      "n_chars": 1247683,
+      "n_tokens": 353570,
+      "n_chars": 1035728,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
         "aktuel": 45,
-        "bilisim": 145,
-        "egitim": 1,
-        "ekonomi": 150,
-        "iletisim": 14,
-        "kultursanat": 15,
-        "saglik": 118,
-        "siyasi": 144,
-        "spor": 24,
-        "ticaret": 150,
+        "bilisim": 1,
+        "ekonomi": 14,
+        "gida": 145,
+        "iletisim": 15,
+        "magazin": 150,
+        "saglik": 144,
+        "seyahat": 118,
+        "spor": 150,
+        "teknoloji": 24,
         "turizm": 9,
         "yasam": 185
       }
@@ -626,24 +661,24 @@
     {
       "haystack_id": "tr-core1000-3",
       "n_examples": 1000,
-      "n_tokens": 406766,
-      "n_chars": 1188943,
+      "n_tokens": 364004,
+      "n_chars": 1067753,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
-        "aktuel": 150,
-        "bilisim": 13,
-        "ekonomi": 150,
-        "gida": 51,
-        "iletisim": 82,
-        "kultursanat": 2,
-        "saglik": 19,
+        "aktuel": 13,
+        "egitim": 51,
+        "ekonomi": 82,
+        "gida": 150,
+        "iletisim": 2,
+        "magazin": 19,
+        "saglik": 144,
         "seyahat": 7,
-        "siyasi": 143,
-        "teknoloji": 18,
-        "ticaret": 144,
+        "spor": 143,
+        "teknoloji": 150,
+        "ticaret": 18,
         "turizm": 49,
         "yasam": 172
       }
@@ -651,21 +686,21 @@
     {
       "haystack_id": "tr-core1000-4",
       "n_examples": 1000,
-      "n_tokens": 375402,
-      "n_chars": 1097152,
+      "n_tokens": 371914,
+      "n_chars": 1095089,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
-        "aktuel": 144,
-        "bilisim": 12,
-        "egitim": 50,
-        "ekonomi": 108,
-        "gida": 150,
-        "kultursanat": 28,
+        "aktuel": 12,
+        "bilisim": 50,
+        "egitim": 108,
+        "gida": 144,
+        "iletisim": 28,
+        "kultursanat": 12,
         "magazin": 150,
-        "saglik": 12,
+        "saglik": 150,
         "seyahat": 9,
         "siyasi": 14,
         "spor": 33,
@@ -678,22 +713,22 @@
     {
       "haystack_id": "tr-core1000-5",
       "n_examples": 1000,
-      "n_tokens": 392380,
-      "n_chars": 1160852,
+      "n_tokens": 377359,
+      "n_chars": 1120659,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
         "aktuel": 34,
-        "bilisim": 150,
-        "egitim": 75,
-        "ekonomi": 1,
+        "bilisim": 75,
+        "egitim": 1,
+        "gida": 145,
         "kultursanat": 15,
         "magazin": 144,
-        "saglik": 77,
-        "seyahat": 13,
-        "siyasi": 145,
+        "saglik": 150,
+        "seyahat": 77,
+        "siyasi": 13,
         "spor": 9,
         "teknoloji": 150,
         "ticaret": 141,
@@ -704,8 +739,8 @@
     {
       "haystack_id": "tr-core2500-0",
       "n_examples": 2500,
-      "n_tokens": 969472,
-      "n_chars": 2842666,
+      "n_tokens": 893379,
+      "n_chars": 2634622,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -714,10 +749,10 @@
         "aktuel": 7,
         "bilisim": 315,
         "egitim": 12,
-        "ekonomi": 375,
+        "ekonomi": 91,
         "gida": 365,
-        "iletisim": 91,
-        "kultursanat": 38,
+        "iletisim": 38,
+        "magazin": 375,
         "saglik": 33,
         "seyahat": 254,
         "siyasi": 58,
@@ -731,111 +766,111 @@
     {
       "haystack_id": "tr-core2500-1",
       "n_examples": 2500,
-      "n_tokens": 997053,
-      "n_chars": 2933221,
+      "n_tokens": 936161,
+      "n_chars": 2772042,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
-        "bilisim": 375,
-        "ekonomi": 356,
-        "gida": 27,
+        "egitim": 356,
+        "ekonomi": 27,
+        "gida": 367,
         "iletisim": 44,
         "kultursanat": 12,
-        "magazin": 367,
-        "saglik": 366,
+        "magazin": 366,
+        "saglik": 375,
         "seyahat": 49,
         "siyasi": 273,
         "spor": 13,
-        "teknoloji": 234,
-        "turizm": 9,
-        "yasam": 375
+        "teknoloji": 375,
+        "ticaret": 234,
+        "yasam": 9
       }
     },
     {
       "haystack_id": "tr-core2500-2",
       "n_examples": 2500,
-      "n_tokens": 973246,
-      "n_chars": 2810685,
+      "n_tokens": 938948,
+      "n_chars": 2781099,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
-        "aktuel": 375,
-        "bilisim": 60,
-        "egitim": 375,
-        "ekonomi": 50,
-        "gida": 39,
-        "iletisim": 14,
-        "kultursanat": 444,
-        "magazin": 354,
-        "saglik": 365,
-        "seyahat": 367,
-        "siyasi": 25,
-        "teknoloji": 21,
+        "aktuel": 60,
+        "bilisim": 50,
+        "egitim": 39,
+        "ekonomi": 14,
+        "gida": 367,
+        "iletisim": 444,
+        "kultursanat": 354,
+        "magazin": 25,
+        "saglik": 375,
+        "siyasi": 21,
+        "spor": 375,
+        "teknoloji": 365,
         "turizm": 11
       }
     },
     {
       "haystack_id": "tr-core2500-3",
       "n_examples": 2500,
-      "n_tokens": 964889,
-      "n_chars": 2801126,
+      "n_tokens": 896669,
+      "n_chars": 2632638,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
-        "aktuel": 368,
-        "bilisim": 6,
-        "egitim": 36,
-        "ekonomi": 267,
-        "gida": 72,
+        "aktuel": 6,
+        "bilisim": 36,
+        "egitim": 267,
+        "ekonomi": 72,
+        "gida": 368,
         "iletisim": 3,
         "kultursanat": 115,
-        "magazin": 22,
-        "saglik": 49,
-        "seyahat": 375,
-        "siyasi": 52,
+        "magazin": 366,
+        "saglik": 375,
+        "seyahat": 22,
+        "siyasi": 49,
         "spor": 375,
-        "teknoloji": 378,
-        "turizm": 16,
-        "yasam": 366
+        "teknoloji": 52,
+        "ticaret": 378,
+        "yasam": 16
       }
     },
     {
       "haystack_id": "tr-core2500-4",
       "n_examples": 2500,
-      "n_tokens": 1045425,
-      "n_chars": 3082683,
+      "n_tokens": 933095,
+      "n_chars": 2760074,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
-        "aktuel": 365,
-        "bilisim": 367,
-        "egitim": 17,
-        "ekonomi": 17,
-        "gida": 1,
-        "iletisim": 370,
-        "kultursanat": 175,
-        "saglik": 375,
-        "seyahat": 94,
-        "siyasi": 11,
-        "spor": 156,
-        "teknoloji": 145,
-        "ticaret": 375,
+        "aktuel": 17,
+        "bilisim": 17,
+        "egitim": 1,
+        "ekonomi": 370,
+        "gida": 375,
+        "iletisim": 175,
+        "magazin": 94,
+        "saglik": 365,
+        "seyahat": 11,
+        "siyasi": 156,
+        "spor": 375,
+        "teknoloji": 367,
+        "ticaret": 145,
         "turizm": 32
       }
     },
     {
       "haystack_id": "tr-core2500-5",
       "n_examples": 2500,
-      "n_tokens": 949131,
-      "n_chars": 2745961,
+      "n_tokens": 888905,
+      "n_chars": 2610772,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -843,14 +878,14 @@
       "label_counts": {
         "aktuel": 2,
         "ekonomi": 26,
-        "gida": 365,
+        "gida": 375,
         "iletisim": 94,
-        "kultursanat": 375,
-        "magazin": 375,
-        "saglik": 81,
+        "kultursanat": 81,
+        "magazin": 365,
+        "saglik": 375,
         "seyahat": 277,
-        "siyasi": 365,
-        "spor": 1,
+        "siyasi": 1,
+        "spor": 365,
         "teknoloji": 3,
         "ticaret": 349,
         "turizm": 119,

marc_en/haystacks.jsonl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:420f837ae66383b1902d8c57e41c0e56b5569fa418d3570e638f64ad33fe62c1
-size 44496096
+oid sha256:c750aec5bbacffc73d1f8c6c91415d3974bdbb791231142a670a6d29d2c6dc5b
+size 43861286

marc_en/manifest.json

@@ -1,6 +1,6 @@
 {
-  "version": "0.9.0",
-  "date": "2026-09-29",
+  "version": "0.10.0",
+  "date": "2026-10-01",
   "environment": {
     "python": "3.12.9",
     "polars": "1.42.0",
@@ -17,6 +17,15 @@
     "drop_label_leakage": true,
     "leak_label_words": false,
     "min_class_support": 0,
+    "boilerplate_keywords": [],
+    "boilerplate_min_keywords": 3,
+    "boilerplate_weak_min": 2,
+    "boilerplate_regex": "",
+    "drop_text_regex": [
+      "\\b\\d{1,3}([.,]\\d)?\\s*/\\s*(10|5|100)\\b",
+      "\\b(\\d|zero|one|two|three|four|five)[\\s-]+stars?\\b"
+    ],
+    "strip_html": true,
     "seed": 42,
     "haystack_target_tokens": [
       100000,
@@ -69,6 +78,7 @@
       "neutral"
     ],
     "close_ordinal_gap": 0,
+    "close_allowed_pairs_file": "experiments/separable_pairs/marc_en.json",
     "core_doc_records": [
       2500,
       6000,
@@ -102,57 +112,58 @@
     "out_dir": "marc_en_out",
     "dataset": ""
   },
-  "source_hash_first1000": "d0de2c459cf8efa1",
-  "rows_after_cleaning": 118779,
+  "source_hash_first1000": "e08571e911d4ac58",
+  "rows_after_cleaning": 115263,
   "cleaning": {
-    "rows_dropped_label_leakage": 657,
-    "label_leakage_rate": 0.005501,
+    "rows_dropped_text_regex": 3570,
+    "rows_dropped_label_leakage": 603,
+    "label_leakage_rate": 0.005204,
     "classes_dropped_low_support": [],
     "min_class_support": 0
   },
   "ranking_feasibility": {
     "n_classes": 3,
-    "min_class_pool": 39528,
-    "max_records": 118584,
+    "min_class_pool": 38235,
+    "max_records": 114705,
     "binding": true,
-    "ranking_length_ceiling_tokens": 5485458
+    "ranking_length_ceiling_tokens": 5043808
   },
   "tier_overlap": {
     "100000": {
       "n_haystacks": 10,
-      "mean_jaccard": 0.0092,
-      "max_jaccard": 0.017,
-      "pool_fraction_per_haystack": 0.0183
+      "mean_jaccard": 0.0097,
+      "max_jaccard": 0.0186,
+      "pool_fraction_per_haystack": 0.0193
     },
     "250000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.0212,
-      "max_jaccard": 0.0337,
-      "pool_fraction_per_haystack": 0.0467
+      "mean_jaccard": 0.0224,
+      "max_jaccard": 0.0362,
+      "pool_fraction_per_haystack": 0.0491
     },
     "500000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.0447,
-      "max_jaccard": 0.0533,
-      "pool_fraction_per_haystack": 0.0918
+      "mean_jaccard": 0.0474,
+      "max_jaccard": 0.0563,
+      "pool_fraction_per_haystack": 0.097
     },
     "core12000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0582,
-      "max_jaccard": 0.0614,
-      "pool_fraction_per_haystack": 0.101
+      "mean_jaccard": 0.0599,
+      "max_jaccard": 0.0631,
+      "pool_fraction_per_haystack": 0.1041
     },
     "core2500": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0118,
-      "max_jaccard": 0.013,
-      "pool_fraction_per_haystack": 0.021
+      "mean_jaccard": 0.0122,
+      "max_jaccard": 0.0142,
+      "pool_fraction_per_haystack": 0.0217
     },
     "core6000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0275,
-      "max_jaccard": 0.0299,
-      "pool_fraction_per_haystack": 0.0505
+      "mean_jaccard": 0.0285,
+      "max_jaccard": 0.0317,
+      "pool_fraction_per_haystack": 0.0521
     }
   },
   "max_pool_fraction": 0.35,
@@ -162,15 +173,15 @@
   "proportion_unit": "percent",
   "top_k_k": 3,
   "tokenizer": "Qwen/Qwen3-8B",
-  "questions_written": 212,
+  "questions_written": 216,
   "kind_distribution": {
     "close_comparison": 19,
     "count": 59,
     "label_vs_label": 36,
-    "least_common": 13,
-    "most_common": 15,
+    "least_common": 14,
+    "most_common": 16,
     "proportion": 59,
-    "second_most": 11
+    "second_most": 13
   },
   "rare_count_questions": 0,
   "rare_band": [
@@ -180,180 +191,180 @@
   "haystacks": [
     {
       "haystack_id": "en-100000-0",
-      "n_examples": 2161,
-      "n_tokens": 98246,
-      "n_chars": 424938,
-      "drift_target": "neutral",
+      "n_examples": 2272,
+      "n_tokens": 98269,
+      "n_chars": 422425,
+      "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "en-100000-1",
-      "n_examples": 2152,
-      "n_tokens": 98389,
-      "n_chars": 423323,
+      "n_examples": 2248,
+      "n_tokens": 98268,
+      "n_chars": 423211,
       "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "en-100000-2",
-      "n_examples": 2275,
-      "n_tokens": 98232,
-      "n_chars": 422378,
+      "n_examples": 2218,
+      "n_tokens": 98285,
+      "n_chars": 423756,
       "drift_target": "positive",
       "drift_ok": true,
-      "n_questions": 9
+      "n_questions": 11
     },
     {
       "haystack_id": "en-100000-3",
-      "n_examples": 2266,
-      "n_tokens": 98255,
-      "n_chars": 423150,
+      "n_examples": 2234,
+      "n_tokens": 98244,
+      "n_chars": 424672,
       "drift_target": "neutral",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "en-100000-4",
-      "n_examples": 2157,
-      "n_tokens": 98302,
-      "n_chars": 421396,
-      "drift_target": "negative",
+      "n_examples": 2206,
+      "n_tokens": 98318,
+      "n_chars": 421488,
+      "drift_target": "neutral",
       "drift_ok": true,
       "n_questions": 10
     },
     {
       "haystack_id": "en-100000-5",
-      "n_examples": 2112,
-      "n_tokens": 98377,
-      "n_chars": 422651,
-      "drift_target": "neutral",
+      "n_examples": 2201,
+      "n_tokens": 98346,
+      "n_chars": 422849,
+      "drift_target": "negative",
       "drift_ok": true,
       "n_questions": 7
     },
     {
       "haystack_id": "en-100000-6",
-      "n_examples": 2159,
-      "n_tokens": 98266,
-      "n_chars": 422708,
-      "drift_target": "positive",
+      "n_examples": 2251,
+      "n_tokens": 98271,
+      "n_chars": 423078,
+      "drift_target": "negative",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "en-100000-7",
-      "n_examples": 2136,
-      "n_tokens": 98256,
-      "n_chars": 422441,
-      "drift_target": "positive",
+      "n_examples": 2244,
+      "n_tokens": 98290,
+      "n_chars": 422230,
+      "drift_target": "negative",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "en-100000-8",
-      "n_examples": 2157,
-      "n_tokens": 98306,
-      "n_chars": 422918,
+      "n_examples": 2175,
+      "n_tokens": 98267,
+      "n_chars": 423204,
       "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "en-100000-9",
-      "n_examples": 2148,
-      "n_tokens": 98355,
-      "n_chars": 421906,
-      "drift_target": "neutral",
+      "n_examples": 2232,
+      "n_tokens": 98277,
+      "n_chars": 421033,
+      "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "en-250000-0",
-      "n_examples": 5536,
-      "n_tokens": 245770,
-      "n_chars": 1055123,
-      "drift_target": "neutral",
+      "n_examples": 5546,
+      "n_tokens": 245728,
+      "n_chars": 1056594,
+      "drift_target": "negative",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "en-250000-1",
-      "n_examples": 5448,
-      "n_tokens": 245881,
-      "n_chars": 1056964,
-      "drift_target": "neutral",
+      "n_examples": 5595,
+      "n_tokens": 245741,
+      "n_chars": 1056653,
+      "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "en-250000-2",
-      "n_examples": 5505,
-      "n_tokens": 245679,
-      "n_chars": 1054412,
-      "drift_target": "neutral",
+      "n_examples": 5561,
+      "n_tokens": 245810,
+      "n_chars": 1054540,
+      "drift_target": "negative",
       "drift_ok": true,
-      "n_questions": 9
+      "n_questions": 11
     },
     {
       "haystack_id": "en-250000-3",
-      "n_examples": 5415,
-      "n_tokens": 245867,
-      "n_chars": 1056443,
-      "drift_target": "neutral",
+      "n_examples": 5672,
+      "n_tokens": 245667,
+      "n_chars": 1054439,
+      "drift_target": "negative",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "en-250000-4",
-      "n_examples": 5804,
-      "n_tokens": 245503,
-      "n_chars": 1056550,
-      "drift_target": "neutral",
+      "n_examples": 5923,
+      "n_tokens": 245370,
+      "n_chars": 1056551,
+      "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "en-500000-0",
-      "n_examples": 10959,
-      "n_tokens": 491589,
-      "n_chars": 2113173,
-      "drift_target": "neutral",
+      "n_examples": 11105,
+      "n_tokens": 491448,
+      "n_chars": 2114333,
+      "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 7
     },
     {
       "haystack_id": "en-500000-1",
-      "n_examples": 10766,
-      "n_tokens": 491821,
-      "n_chars": 2111135,
-      "drift_target": "positive",
+      "n_examples": 11124,
+      "n_tokens": 491584,
+      "n_chars": 2111075,
+      "drift_target": "neutral",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "en-500000-2",
-      "n_examples": 11080,
-      "n_tokens": 491449,
-      "n_chars": 2112392,
-      "drift_target": "neutral",
+      "n_examples": 11497,
+      "n_tokens": 491249,
+      "n_chars": 2113210,
+      "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "en-500000-3",
-      "n_examples": 10897,
-      "n_tokens": 491617,
-      "n_chars": 2118869,
-      "drift_target": "negative",
+      "n_examples": 11134,
+      "n_tokens": 491446,
+      "n_chars": 2113462,
+      "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 6
     },
     {
       "haystack_id": "en-500000-4",
-      "n_examples": 10823,
-      "n_tokens": 491670,
-      "n_chars": 2116715,
+      "n_examples": 11042,
+      "n_tokens": 491641,
+      "n_chars": 2114643,
       "drift_target": "positive",
       "drift_ok": true,
       "n_questions": 9
@@ -361,8 +372,8 @@
     {
       "haystack_id": "en-core2500-0",
       "n_examples": 2500,
-      "n_tokens": 107650,
-      "n_chars": 463568,
+      "n_tokens": 107820,
+      "n_chars": 464024,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -376,8 +387,8 @@
     {
       "haystack_id": "en-core2500-1",
       "n_examples": 2500,
-      "n_tokens": 109853,
-      "n_chars": 472455,
+      "n_tokens": 109991,
+      "n_chars": 474713,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -391,8 +402,8 @@
     {
       "haystack_id": "en-core2500-2",
       "n_examples": 2500,
-      "n_tokens": 111373,
-      "n_chars": 478679,
+      "n_tokens": 108270,
+      "n_chars": 465949,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -406,8 +417,8 @@
     {
       "haystack_id": "en-core2500-3",
       "n_examples": 2500,
-      "n_tokens": 110145,
-      "n_chars": 472299,
+      "n_tokens": 107186,
+      "n_chars": 461383,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -421,8 +432,8 @@
     {
       "haystack_id": "en-core2500-4",
       "n_examples": 2500,
-      "n_tokens": 110227,
-      "n_chars": 473261,
+      "n_tokens": 101767,
+      "n_chars": 437735,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -436,8 +447,8 @@
     {
       "haystack_id": "en-core2500-5",
       "n_examples": 2500,
-      "n_tokens": 113739,
-      "n_chars": 489602,
+      "n_tokens": 108153,
+      "n_chars": 465901,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -451,8 +462,8 @@
     {
       "haystack_id": "en-core6000-0",
       "n_examples": 6000,
-      "n_tokens": 266081,
-      "n_chars": 1144091,
+      "n_tokens": 254120,
+      "n_chars": 1092666,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -466,8 +477,8 @@
     {
       "haystack_id": "en-core6000-1",
       "n_examples": 6000,
-      "n_tokens": 265698,
-      "n_chars": 1142729,
+      "n_tokens": 257895,
+      "n_chars": 1108708,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -481,8 +492,8 @@
     {
       "haystack_id": "en-core6000-2",
       "n_examples": 6000,
-      "n_tokens": 265003,
-      "n_chars": 1138639,
+      "n_tokens": 255879,
+      "n_chars": 1103665,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -496,8 +507,8 @@
     {
       "haystack_id": "en-core6000-3",
       "n_examples": 6000,
-      "n_tokens": 265980,
-      "n_chars": 1142438,
+      "n_tokens": 258318,
+      "n_chars": 1113627,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -511,8 +522,8 @@
     {
       "haystack_id": "en-core6000-4",
       "n_examples": 6000,
-      "n_tokens": 264203,
-      "n_chars": 1134940,
+      "n_tokens": 254745,
+      "n_chars": 1093701,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -526,8 +537,8 @@
     {
       "haystack_id": "en-core6000-5",
       "n_examples": 6000,
-      "n_tokens": 265451,
-      "n_chars": 1143716,
+      "n_tokens": 258781,
+      "n_chars": 1112980,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -541,8 +552,8 @@
     {
       "haystack_id": "en-core12000-0",
       "n_examples": 12000,
-      "n_tokens": 525193,
-      "n_chars": 2257613,
+      "n_tokens": 521614,
+      "n_chars": 2246510,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -556,8 +567,8 @@
     {
       "haystack_id": "en-core12000-1",
       "n_examples": 12000,
-      "n_tokens": 536428,
-      "n_chars": 2306505,
+      "n_tokens": 508884,
+      "n_chars": 2186299,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -571,8 +582,8 @@
     {
       "haystack_id": "en-core12000-2",
       "n_examples": 12000,
-      "n_tokens": 532504,
-      "n_chars": 2295001,
+      "n_tokens": 522220,
+      "n_chars": 2249325,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -586,8 +597,8 @@
     {
       "haystack_id": "en-core12000-3",
       "n_examples": 12000,
-      "n_tokens": 522346,
-      "n_chars": 2250415,
+      "n_tokens": 515940,
+      "n_chars": 2220547,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -601,8 +612,8 @@
     {
       "haystack_id": "en-core12000-4",
       "n_examples": 12000,
-      "n_tokens": 532914,
-      "n_chars": 2294753,
+      "n_tokens": 518894,
+      "n_chars": 2233429,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -616,8 +627,8 @@
     {
       "haystack_id": "en-core12000-5",
       "n_examples": 12000,
-      "n_tokens": 527931,
-      "n_chars": 2272574,
+      "n_tokens": 513444,
+      "n_chars": 2206191,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,

musteri_tr/haystacks.jsonl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4b15bd5e342ae7b8941bd70c047de0ef842528e122b2c1c99cfefd6ee00363f0
-size 30400044
+oid sha256:e2743877e64762fbf4506ab439d74b3ee2d836a1d85492b750363bdcc296bcac
+size 30200541

musteri_tr/manifest.json

@@ -1,6 +1,6 @@
 {
-  "version": "0.9.0",
-  "date": "2026-09-29",
+  "version": "0.10.0",
+  "date": "2026-10-01",
   "environment": {
     "python": "3.12.9",
     "polars": "1.42.0",
@@ -17,6 +17,17 @@
     "drop_label_leakage": true,
     "leak_label_words": false,
     "min_class_support": 0,
+    "boilerplate_keywords": [],
+    "boilerplate_min_keywords": 3,
+    "boilerplate_weak_min": 2,
+    "boilerplate_regex": "",
+    "drop_text_regex": [
+      "\\b\\d{1,3}([.,]\\d)?\\s*/\\s*(10|5|100)\\b",
+      "\\b\\d{1,2}\\s*(yıldız|puan)",
+      "\\bpuan[ıi]m\\b",
+      "\\b(on|dokuz|sekiz|yedi|altı|beş|dört|üç|iki|bir|sıfır)\\s+(yıldız|puan)"
+    ],
+    "strip_html": true,
     "seed": 42,
     "haystack_target_tokens": [
       100000,
@@ -69,6 +80,7 @@
       "nötr"
     ],
     "close_ordinal_gap": 0,
+    "close_allowed_pairs_file": "experiments/separable_pairs/musteri_tr.json",
     "core_doc_records": [
       2500,
       6000,
@@ -102,57 +114,58 @@
     "out_dir": "musteri_tr_out",
     "dataset": ""
   },
-  "source_hash_first1000": "754ffa0b668646e1",
-  "rows_after_cleaning": 36924,
+  "source_hash_first1000": "1a9d1d9cad623881",
+  "rows_after_cleaning": 36289,
   "cleaning": {
-    "rows_dropped_label_leakage": 80,
-    "label_leakage_rate": 0.002162,
+    "rows_dropped_text_regex": 639,
+    "rows_dropped_label_leakage": 78,
+    "label_leakage_rate": 0.002145,
     "classes_dropped_low_support": [],
     "min_class_support": 0
   },
   "ranking_feasibility": {
     "n_classes": 3,
-    "min_class_pool": 12135,
-    "max_records": 36405,
+    "min_class_pool": 11995,
+    "max_records": 35985,
     "binding": true,
-    "ranking_length_ceiling_tokens": 1369774
+    "ranking_length_ceiling_tokens": 1317194
   },
   "tier_overlap": {
     "100000": {
       "n_haystacks": 10,
-      "mean_jaccard": 0.0387,
-      "max_jaccard": 0.0622,
-      "pool_fraction_per_haystack": 0.0713
+      "mean_jaccard": 0.0399,
+      "max_jaccard": 0.0685,
+      "pool_fraction_per_haystack": 0.0733
     },
     "250000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.0934,
-      "max_jaccard": 0.119,
-      "pool_fraction_per_haystack": 0.178
+      "mean_jaccard": 0.0971,
+      "max_jaccard": 0.1227,
+      "pool_fraction_per_haystack": 0.1855
     },
     "500000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.2091,
-      "max_jaccard": 0.26,
-      "pool_fraction_per_haystack": 0.361
+      "mean_jaccard": 0.2175,
+      "max_jaccard": 0.2707,
+      "pool_fraction_per_haystack": 0.3742
     },
     "core12000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.213,
-      "max_jaccard": 0.2186,
-      "pool_fraction_per_haystack": 0.325
+      "mean_jaccard": 0.2171,
+      "max_jaccard": 0.2199,
+      "pool_fraction_per_haystack": 0.3307
     },
     "core2500": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0374,
-      "max_jaccard": 0.0432,
-      "pool_fraction_per_haystack": 0.0677
+      "mean_jaccard": 0.0382,
+      "max_jaccard": 0.0443,
+      "pool_fraction_per_haystack": 0.0689
     },
     "core6000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0956,
-      "max_jaccard": 0.1015,
-      "pool_fraction_per_haystack": 0.1625
+      "mean_jaccard": 0.0973,
+      "max_jaccard": 0.1026,
+      "pool_fraction_per_haystack": 0.1653
     }
   },
   "max_pool_fraction": 0.35,
@@ -160,22 +173,22 @@
   "tiers_built_over_pool_cap": [
     {
       "tier": 500000,
-      "pool_fraction": 0.3779
+      "pool_fraction": 0.3952
     }
   ],
   "label_space": 3,
   "proportion_unit": "percent",
   "top_k_k": 3,
   "tokenizer": "Qwen/Qwen3-8B",
-  "questions_written": 217,
+  "questions_written": 213,
   "kind_distribution": {
     "close_comparison": 18,
     "count": 59,
     "label_vs_label": 38,
-    "least_common": 16,
+    "least_common": 14,
     "most_common": 15,
     "proportion": 58,
-    "second_most": 13
+    "second_most": 11
   },
   "rare_count_questions": 0,
   "rare_band": [
@@ -185,180 +198,180 @@
   "haystacks": [
     {
       "haystack_id": "tr-100000-0",
-      "n_examples": 2766,
-      "n_tokens": 99137,
-      "n_chars": 289366,
-      "drift_target": "olumsuz",
+      "n_examples": 2862,
+      "n_tokens": 99129,
+      "n_chars": 290022,
+      "drift_target": "nötr",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "tr-100000-1",
-      "n_examples": 2637,
-      "n_tokens": 99229,
-      "n_chars": 290146,
+      "n_examples": 2573,
+      "n_tokens": 99249,
+      "n_chars": 288940,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "tr-100000-2",
-      "n_examples": 2553,
-      "n_tokens": 99281,
-      "n_chars": 288451,
-      "drift_target": "nötr",
+      "n_examples": 2604,
+      "n_tokens": 99249,
+      "n_chars": 287280,
+      "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "tr-100000-3",
-      "n_examples": 2696,
-      "n_tokens": 99192,
-      "n_chars": 289314,
-      "drift_target": "olumsuz",
+      "n_examples": 2660,
+      "n_tokens": 99115,
+      "n_chars": 289027,
+      "drift_target": "nötr",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-100000-4",
-      "n_examples": 2639,
-      "n_tokens": 99205,
-      "n_chars": 288661,
+      "n_examples": 2669,
+      "n_tokens": 99199,
+      "n_chars": 288831,
       "drift_target": "olumlu",
       "drift_ok": true,
-      "n_questions": 11
+      "n_questions": 9
     },
     {
       "haystack_id": "tr-100000-5",
-      "n_examples": 2609,
-      "n_tokens": 99260,
-      "n_chars": 289532,
+      "n_examples": 2653,
+      "n_tokens": 99214,
+      "n_chars": 288773,
       "drift_target": "nötr",
       "drift_ok": true,
       "n_questions": 7
     },
     {
       "haystack_id": "tr-100000-6",
-      "n_examples": 2646,
-      "n_tokens": 99174,
-      "n_chars": 288073,
+      "n_examples": 2635,
+      "n_tokens": 99194,
+      "n_chars": 288343,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-100000-7",
-      "n_examples": 2642,
-      "n_tokens": 99228,
-      "n_chars": 290000,
+      "n_examples": 2703,
+      "n_tokens": 99208,
+      "n_chars": 289937,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-100000-8",
-      "n_examples": 2500,
-      "n_tokens": 99251,
-      "n_chars": 287011,
+      "n_examples": 2555,
+      "n_tokens": 99210,
+      "n_chars": 287064,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "tr-100000-9",
-      "n_examples": 2622,
-      "n_tokens": 99202,
-      "n_chars": 290207,
+      "n_examples": 2697,
+      "n_tokens": 99261,
+      "n_chars": 289966,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "tr-250000-0",
-      "n_examples": 6899,
-      "n_tokens": 247808,
-      "n_chars": 721464,
+      "n_examples": 6980,
+      "n_tokens": 247880,
+      "n_chars": 722861,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 10
     },
     {
       "haystack_id": "tr-250000-1",
-      "n_examples": 6455,
-      "n_tokens": 248139,
-      "n_chars": 721276,
-      "drift_target": "nötr",
+      "n_examples": 6563,
+      "n_tokens": 248130,
+      "n_chars": 722026,
+      "drift_target": "olumlu",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "tr-250000-2",
-      "n_examples": 6439,
-      "n_tokens": 248143,
-      "n_chars": 720111,
+      "n_examples": 6560,
+      "n_tokens": 248138,
+      "n_chars": 721314,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "tr-250000-3",
-      "n_examples": 6469,
-      "n_tokens": 248040,
-      "n_chars": 720100,
-      "drift_target": "olumsuz",
+      "n_examples": 6781,
+      "n_tokens": 248006,
+      "n_chars": 719407,
+      "drift_target": "olumlu",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-250000-4",
-      "n_examples": 6598,
-      "n_tokens": 248041,
-      "n_chars": 725141,
-      "drift_target": "olumsuz",
+      "n_examples": 6776,
+      "n_tokens": 248106,
+      "n_chars": 725581,
+      "drift_target": "nötr",
       "drift_ok": true,
-      "n_questions": 11
+      "n_questions": 9
     },
     {
       "haystack_id": "tr-500000-0",
-      "n_examples": 13142,
-      "n_tokens": 496089,
-      "n_chars": 1442979,
+      "n_examples": 13276,
+      "n_tokens": 496041,
+      "n_chars": 1441568,
       "drift_target": "olumlu",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "tr-500000-1",
-      "n_examples": 13120,
-      "n_tokens": 496238,
-      "n_chars": 1450298,
-      "drift_target": "nötr",
+      "n_examples": 13517,
+      "n_tokens": 496144,
+      "n_chars": 1450939,
+      "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 11
     },
     {
       "haystack_id": "tr-500000-2",
-      "n_examples": 13322,
-      "n_tokens": 495969,
-      "n_chars": 1447935,
+      "n_examples": 13625,
+      "n_tokens": 496026,
+      "n_chars": 1445573,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-500000-3",
-      "n_examples": 13618,
-      "n_tokens": 495996,
-      "n_chars": 1448876,
-      "drift_target": "nötr",
+      "n_examples": 13825,
+      "n_tokens": 495828,
+      "n_chars": 1446311,
+      "drift_target": "olumlu",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-500000-4",
-      "n_examples": 13439,
-      "n_tokens": 495998,
-      "n_chars": 1444628,
+      "n_examples": 13649,
+      "n_tokens": 495977,
+      "n_chars": 1445890,
       "drift_target": "olumlu",
       "drift_ok": true,
       "n_questions": 7
@@ -366,8 +379,8 @@
     {
       "haystack_id": "tr-core2500-0",
       "n_examples": 2500,
-      "n_tokens": 91242,
-      "n_chars": 265269,
+      "n_tokens": 88854,
+      "n_chars": 257249,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -381,8 +394,8 @@
     {
       "haystack_id": "tr-core2500-1",
       "n_examples": 2500,
-      "n_tokens": 94120,
-      "n_chars": 273657,
+      "n_tokens": 91043,
+      "n_chars": 264499,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -396,8 +409,8 @@
     {
       "haystack_id": "tr-core2500-2",
       "n_examples": 2500,
-      "n_tokens": 94295,
-      "n_chars": 275616,
+      "n_tokens": 94304,
+      "n_chars": 274843,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -411,8 +424,8 @@
     {
       "haystack_id": "tr-core2500-3",
       "n_examples": 2500,
-      "n_tokens": 93050,
-      "n_chars": 269611,
+      "n_tokens": 93503,
+      "n_chars": 272479,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -426,8 +439,8 @@
     {
       "haystack_id": "tr-core2500-4",
       "n_examples": 2500,
-      "n_tokens": 93846,
-      "n_chars": 273703,
+      "n_tokens": 91550,
+      "n_chars": 265624,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -441,8 +454,8 @@
     {
       "haystack_id": "tr-core2500-5",
       "n_examples": 2500,
-      "n_tokens": 92423,
-      "n_chars": 270116,
+      "n_tokens": 90739,
+      "n_chars": 264609,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -456,8 +469,8 @@
     {
       "haystack_id": "tr-core6000-0",
       "n_examples": 6000,
-      "n_tokens": 224015,
-      "n_chars": 651875,
+      "n_tokens": 221633,
+      "n_chars": 644271,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -471,8 +484,8 @@
     {
       "haystack_id": "tr-core6000-1",
       "n_examples": 6000,
-      "n_tokens": 220337,
-      "n_chars": 640318,
+      "n_tokens": 217081,
+      "n_chars": 629914,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -486,8 +499,8 @@
     {
       "haystack_id": "tr-core6000-2",
       "n_examples": 6000,
-      "n_tokens": 222931,
-      "n_chars": 649124,
+      "n_tokens": 222819,
+      "n_chars": 649350,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -501,8 +514,8 @@
     {
       "haystack_id": "tr-core6000-3",
       "n_examples": 6000,
-      "n_tokens": 224955,
-      "n_chars": 655435,
+      "n_tokens": 215573,
+      "n_chars": 627427,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -516,8 +529,8 @@
     {
       "haystack_id": "tr-core6000-4",
       "n_examples": 6000,
-      "n_tokens": 225198,
-      "n_chars": 654783,
+      "n_tokens": 219565,
+      "n_chars": 638298,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -531,8 +544,8 @@
     {
       "haystack_id": "tr-core6000-5",
       "n_examples": 6000,
-      "n_tokens": 223398,
-      "n_chars": 650251,
+      "n_tokens": 218925,
+      "n_chars": 637834,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -546,8 +559,8 @@
     {
       "haystack_id": "tr-core12000-0",
       "n_examples": 12000,
-      "n_tokens": 443416,
-      "n_chars": 1289561,
+      "n_tokens": 440117,
+      "n_chars": 1281614,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -561,8 +574,8 @@
     {
       "haystack_id": "tr-core12000-1",
       "n_examples": 12000,
-      "n_tokens": 444183,
-      "n_chars": 1294136,
+      "n_tokens": 441176,
+      "n_chars": 1283277,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -576,8 +589,8 @@
     {
       "haystack_id": "tr-core12000-2",
       "n_examples": 12000,
-      "n_tokens": 447231,
-      "n_chars": 1302864,
+      "n_tokens": 440563,
+      "n_chars": 1281405,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -591,8 +604,8 @@
     {
       "haystack_id": "tr-core12000-3",
       "n_examples": 12000,
-      "n_tokens": 444261,
-      "n_chars": 1292386,
+      "n_tokens": 440757,
+      "n_chars": 1284216,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -606,8 +619,8 @@
     {
       "haystack_id": "tr-core12000-4",
       "n_examples": 12000,
-      "n_tokens": 443705,
-      "n_chars": 1291086,
+      "n_tokens": 444255,
+      "n_chars": 1291408,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -621,8 +634,8 @@
     {
       "haystack_id": "tr-core12000-5",
       "n_examples": 12000,
-      "n_tokens": 448404,
-      "n_chars": 1305967,
+      "n_tokens": 438471,
+      "n_chars": 1276240,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,

sikayet_tr/manifest.json

@@ -1,6 +1,6 @@
 {
-  "version": "0.9.0",
-  "date": "2026-09-29",
+  "version": "0.10.0",
+  "date": "2026-10-01",
   "environment": {
     "python": "3.12.9",
     "polars": "1.42.0",
@@ -17,6 +17,12 @@
     "drop_label_leakage": true,
     "leak_label_words": true,
     "min_class_support": 1000,
+    "boilerplate_keywords": [],
+    "boilerplate_min_keywords": 3,
+    "boilerplate_weak_min": 2,
+    "boilerplate_regex": "",
+    "drop_text_regex": [],
+    "strip_html": false,
     "seed": 42,
     "haystack_target_tokens": [
       100000,
@@ -69,6 +75,7 @@
     "close_min_diff": 2,
     "close_exclude_labels": [],
     "close_ordinal_gap": 0,
+    "close_allowed_pairs_file": "experiments/separable_pairs/sikayet_tr.json",
     "core_doc_records": [
       1000,
       2500,
@@ -131,26 +138,26 @@
     },
     "core1000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0026,
-      "max_jaccard": 0.0055,
+      "mean_jaccard": 0.0027,
+      "max_jaccard": 0.0065,
       "pool_fraction_per_haystack": 0.0036
     },
     "core10000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0226,
-      "max_jaccard": 0.0447,
+      "mean_jaccard": 0.0255,
+      "max_jaccard": 0.0424,
       "pool_fraction_per_haystack": 0.0357
     },
     "core2500": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0044,
-      "max_jaccard": 0.006,
+      "mean_jaccard": 0.0053,
+      "max_jaccard": 0.0085,
       "pool_fraction_per_haystack": 0.0089
     },
     "core5000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0116,
-      "max_jaccard": 0.0175,
+      "mean_jaccard": 0.0096,
+      "max_jaccard": 0.0201,
       "pool_fraction_per_haystack": 0.0178
     }
   },
@@ -161,9 +168,9 @@
   "proportion_unit": "per_mille",
   "top_k_k": 3,
   "tokenizer": "Qwen/Qwen3-8B",
-  "questions_written": 381,
+  "questions_written": 366,
   "kind_distribution": {
-    "close_comparison": 81,
+    "close_comparison": 66,
     "count": 150,
     "label_vs_label": 25,
     "least_common": 25,
@@ -193,7 +200,7 @@
       "n_chars": 290890,
       "drift_target": "emlak ve inşaat",
       "drift_ok": true,
-      "n_questions": 13
+      "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-2",
@@ -211,7 +218,7 @@
       "n_chars": 289218,
       "drift_target": "sağlık",
       "drift_ok": true,
-      "n_questions": 13
+      "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-4",
@@ -229,7 +236,7 @@
       "n_chars": 292002,
       "drift_target": "kişisel bakım ve kozmetik",
       "drift_ok": true,
-      "n_questions": 13
+      "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-6",
@@ -238,7 +245,7 @@
       "n_chars": 288100,
       "drift_target": "enerji",
       "drift_ok": true,
-      "n_questions": 14
+      "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-7",
@@ -247,7 +254,7 @@
       "n_chars": 292520,
       "drift_target": "elektronik",
       "drift_ok": true,
-      "n_questions": 13
+      "n_questions": 12
     },
     {
       "haystack_id": "tr-100000-8",
@@ -283,7 +290,7 @@
       "n_chars": 721704,
       "drift_target": "kamu hizmetleri",
       "drift_ok": true,
-      "n_questions": 15
+      "n_questions": 13
     },
     {
       "haystack_id": "tr-250000-2",
@@ -301,7 +308,7 @@
       "n_chars": 729751,
       "drift_target": "küçük ev aletleri",
       "drift_ok": true,
-      "n_questions": 14
+      "n_questions": 13
     },
     {
       "haystack_id": "tr-250000-4",
@@ -310,7 +317,7 @@
       "n_chars": 723520,
       "drift_target": "hizmet sektörü",
       "drift_ok": true,
-      "n_questions": 15
+      "n_questions": 14
     },
     {
       "haystack_id": "tr-500000-0",
@@ -355,7 +362,7 @@
       "n_chars": 1456056,
       "drift_target": "kamu hizmetleri",
       "drift_ok": true,
-      "n_questions": 15
+      "n_questions": 13
     },
     {
       "haystack_id": "tr-1000000-0",
@@ -364,7 +371,7 @@
       "n_chars": 2904155,
       "drift_target": "emlak ve inşaat",
       "drift_ok": true,
-      "n_questions": 14
+      "n_questions": 13
     },
     {
       "haystack_id": "tr-1000000-1",
@@ -373,7 +380,7 @@
       "n_chars": 2904974,
       "drift_target": "enerji",
       "drift_ok": true,
-      "n_questions": 14
+      "n_questions": 13
     },
     {
       "haystack_id": "tr-1000000-2",
@@ -382,7 +389,7 @@
       "n_chars": 2925639,
       "drift_target": "turizm",
       "drift_ok": true,
-      "n_questions": 13
+      "n_questions": 12
     },
     {
       "haystack_id": "tr-1000000-3",
@@ -477,8 +484,8 @@
     {
       "haystack_id": "tr-core1000-2",
       "n_examples": 1000,
-      "n_tokens": 100784,
-      "n_chars": 295114,
+      "n_tokens": 100890,
+      "n_chars": 294565,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -487,10 +494,10 @@
         "alışveriş": 7,
         "bilgisayar": 9,
         "elektronik": 50,
-        "emlak ve inşaat": 145,
-        "enerji": 9,
-        "etkinlik ve organizasyon": 10,
-        "eğitim": 25,
+        "emlak ve inşaat": 9,
+        "enerji": 10,
+        "etkinlik ve organizasyon": 25,
+        "eğitim": 150,
         "finans": 38,
         "giyim": 145,
         "hizmet sektörü": 150,
@@ -504,15 +511,15 @@
         "mücevher saat gözlük": 1,
         "sağlık": 40,
         "spor": 2,
-        "temizlik": 150,
+        "temizlik": 145,
         "turizm": 16
       }
     },
     {
       "haystack_id": "tr-core1000-3",
       "n_examples": 1000,
-      "n_tokens": 101851,
-      "n_chars": 296776,
+      "n_tokens": 101879,
+      "n_chars": 296808,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -534,11 +541,11 @@
         "küçük ev aletleri": 3,
         "medya": 60,
         "mekan ve eğlence": 145,
-        "mobilya ve ev tekstili": 150,
-        "mutfak araç gereç": 4,
-        "otomotiv": 7,
-        "sağlık": 11,
-        "sigortacılık": 26,
+        "mobilya ve ev tekstili": 4,
+        "mücevher saat gözlük": 7,
+        "otomotiv": 11,
+        "sağlık": 26,
+        "sigortacılık": 150,
         "ulaşım": 23
       }
     },
@@ -687,8 +694,8 @@
     {
       "haystack_id": "tr-core2500-2",
       "n_examples": 2500,
-      "n_tokens": 255636,
-      "n_chars": 742810,
+      "n_tokens": 254093,
+      "n_chars": 740460,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -701,21 +708,21 @@
         "enerji": 7,
         "eğitim": 3,
         "giyim": 122,
-        "gıda": 375,
+        "gıda": 366,
         "hizmet sektörü": 96,
         "internet": 223,
         "içecek": 22,
-        "kamu hizmetleri": 52,
-        "kişisel bakım ve kozmetik": 70,
-        "küçük ev aletleri": 18,
-        "mekan ve eğlence": 4,
-        "mobilya ve ev tekstili": 367,
+        "kamu hizmetleri": 375,
+        "kişisel bakım ve kozmetik": 52,
+        "küçük ev aletleri": 70,
+        "medya": 18,
+        "mobilya ve ev tekstili": 4,
         "mutfak araç gereç": 15,
         "mücevher saat gözlük": 15,
         "otomotiv": 4,
         "sağlık": 375,
         "sigortacılık": 15,
-        "spor": 366,
+        "spor": 367,
         "temizlik": 9,
         "turizm": 55
       }
@@ -723,8 +730,8 @@
     {
       "haystack_id": "tr-core2500-3",
       "n_examples": 2500,
-      "n_tokens": 251565,
-      "n_chars": 731520,
+      "n_tokens": 251467,
+      "n_chars": 732770,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -733,17 +740,17 @@
         "alışveriş": 44,
         "beyaz eşya": 2,
         "bilgisayar": 71,
-        "elektronik": 13,
-        "emlak ve inşaat": 30,
-        "enerji": 32,
-        "etkinlik ve organizasyon": 241,
-        "eğitim": 365,
-        "giyim": 10,
+        "elektronik": 365,
+        "emlak ve inşaat": 13,
+        "enerji": 30,
+        "etkinlik ve organizasyon": 32,
+        "eğitim": 375,
+        "finans": 241,
         "gıda": 365,
-        "hizmet sektörü": 5,
-        "internet": 223,
+        "hizmet sektörü": 10,
+        "internet": 5,
         "içecek": 375,
-        "kişisel bakım ve kozmetik": 375,
+        "kamu hizmetleri": 223,
         "küçük ev aletleri": 8,
         "medya": 90,
         "mekan ve eğlence": 30,
@@ -760,8 +767,8 @@
     {
       "haystack_id": "tr-core2500-4",
       "n_examples": 2500,
-      "n_tokens": 253577,
-      "n_chars": 740057,
+      "n_tokens": 253874,
+      "n_chars": 739909,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -769,12 +776,12 @@
       "label_counts": {
         "alışveriş": 1,
         "beyaz eşya": 375,
-        "bilgisayar": 367,
-        "elektronik": 17,
-        "emlak ve inşaat": 2,
-        "enerji": 44,
-        "eğitim": 7,
-        "finans": 18,
+        "bilgisayar": 17,
+        "elektronik": 2,
+        "emlak ve inşaat": 44,
+        "etkinlik ve organizasyon": 7,
+        "eğitim": 18,
+        "finans": 375,
         "giyim": 8,
         "gıda": 47,
         "hizmet sektörü": 112,
@@ -784,7 +791,7 @@
         "mekan ve eğlence": 229,
         "mobilya ve ev tekstili": 1,
         "mutfak araç gereç": 181,
-        "mücevher saat gözlük": 375,
+        "mücevher saat gözlük": 367,
         "otomotiv": 365,
         "sağlık": 108,
         "sigortacılık": 46,
@@ -796,8 +803,8 @@
     {
       "haystack_id": "tr-core2500-5",
       "n_examples": 2500,
-      "n_tokens": 254895,
-      "n_chars": 740364,
+      "n_tokens": 256172,
+      "n_chars": 743794,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -813,19 +820,19 @@
         "giyim": 33,
         "gıda": 26,
         "hizmet sektörü": 57,
-        "internet": 367,
-        "içecek": 41,
-        "kamu hizmetleri": 24,
-        "kişisel bakım ve kozmetik": 44,
-        "küçük ev aletleri": 6,
-        "medya": 4,
-        "mekan ve eğlence": 128,
-        "mobilya ve ev tekstili": 135,
-        "mutfak araç gereç": 365,
+        "internet": 41,
+        "içecek": 24,
+        "kamu hizmetleri": 44,
+        "kişisel bakım ve kozmetik": 6,
+        "küçük ev aletleri": 4,
+        "medya": 128,
+        "mekan ve eğlence": 135,
+        "mobilya ve ev tekstili": 16,
+        "mutfak araç gereç": 375,
         "mücevher saat gözlük": 375,
-        "otomotiv": 16,
-        "sağlık": 44,
-        "sigortacılık": 375,
+        "otomotiv": 44,
+        "sağlık": 365,
+        "sigortacılık": 367,
         "spor": 21,
         "temizlik": 7,
         "turizm": 9,
@@ -835,26 +842,26 @@
     {
       "haystack_id": "tr-core5000-0",
       "n_examples": 5000,
-      "n_tokens": 504289,
-      "n_chars": 1469567,
+      "n_tokens": 505310,
+      "n_chars": 1465764,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
         "alışveriş": 43,
-        "bilgisayar": 739,
-        "elektronik": 9,
-        "emlak ve inşaat": 32,
-        "enerji": 1,
-        "etkinlik ve organizasyon": 738,
-        "eğitim": 7,
-        "finans": 47,
-        "giyim": 18,
+        "bilgisayar": 9,
+        "elektronik": 32,
+        "emlak ve inşaat": 1,
+        "enerji": 7,
+        "etkinlik ve organizasyon": 739,
+        "eğitim": 47,
+        "finans": 18,
+        "giyim": 738,
         "gıda": 356,
         "hizmet sektörü": 750,
-        "internet": 750,
-        "içecek": 2,
+        "internet": 2,
+        "içecek": 750,
         "kamu hizmetleri": 22,
         "kişisel bakım ve kozmetik": 5,
         "küçük ev aletleri": 5,
@@ -874,16 +881,16 @@
     {
       "haystack_id": "tr-core5000-1",
       "n_examples": 5000,
-      "n_tokens": 507556,
-      "n_chars": 1480160,
+      "n_tokens": 507050,
+      "n_chars": 1477384,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
         "alışveriş": 327,
-        "beyaz eşya": 268,
-        "bilgisayar": 750,
+        "beyaz eşya": 737,
+        "bilgisayar": 268,
         "elektronik": 39,
         "emlak ve inşaat": 19,
         "enerji": 1,
@@ -893,30 +900,30 @@
         "giyim": 99,
         "gıda": 6,
         "hizmet sektörü": 14,
-        "internet": 738,
-        "içecek": 20,
-        "kamu hizmetleri": 68,
-        "kişisel bakım ve kozmetik": 166,
-        "küçük ev aletleri": 276,
-        "medya": 13,
-        "mekan ve eğlence": 87,
-        "mobilya ve ev tekstili": 750,
-        "mutfak araç gereç": 1,
-        "mücevher saat gözlük": 8,
-        "otomotiv": 96,
-        "sağlık": 737,
-        "sigortacılık": 17,
-        "spor": 35,
-        "temizlik": 1,
-        "turizm": 105,
+        "internet": 20,
+        "içecek": 68,
+        "kamu hizmetleri": 166,
+        "kişisel bakım ve kozmetik": 276,
+        "küçük ev aletleri": 13,
+        "medya": 87,
+        "mekan ve eğlence": 1,
+        "mobilya ve ev tekstili": 8,
+        "mutfak araç gereç": 96,
+        "mücevher saat gözlük": 17,
+        "otomotiv": 35,
+        "sağlık": 738,
+        "sigortacılık": 1,
+        "spor": 750,
+        "temizlik": 105,
+        "turizm": 750,
         "ulaşım": 95
       }
     },
     {
       "haystack_id": "tr-core5000-2",
       "n_examples": 5000,
-      "n_tokens": 505038,
-      "n_chars": 1470625,
+      "n_tokens": 505857,
+      "n_chars": 1471538,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -924,12 +931,12 @@
       "label_counts": {
         "alışveriş": 268,
         "beyaz eşya": 6,
-        "bilgisayar": 738,
-        "elektronik": 19,
-        "emlak ve inşaat": 738,
-        "enerji": 146,
-        "etkinlik ve organizasyon": 750,
-        "eğitim": 117,
+        "bilgisayar": 19,
+        "elektronik": 146,
+        "emlak ve inşaat": 117,
+        "etkinlik ve organizasyon": 738,
+        "finans": 750,
+        "giyim": 738,
         "gıda": 80,
         "internet": 4,
         "içecek": 404,
@@ -989,8 +996,8 @@
     {
       "haystack_id": "tr-core5000-4",
       "n_examples": 5000,
-      "n_tokens": 508847,
-      "n_chars": 1476959,
+      "n_tokens": 507312,
+      "n_chars": 1477023,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -1006,15 +1013,15 @@
         "finans": 12,
         "gıda": 5,
         "hizmet sektörü": 70,
-        "internet": 750,
-        "içecek": 70,
-        "kamu hizmetleri": 4,
-        "kişisel bakım ve kozmetik": 76,
+        "internet": 70,
+        "içecek": 4,
+        "kamu hizmetleri": 76,
+        "kişisel bakım ve kozmetik": 100,
         "küçük ev aletleri": 750,
-        "medya": 100,
-        "mekan ve eğlence": 271,
-        "mobilya ve ev tekstili": 19,
-        "mutfak araç gereç": 2,
+        "medya": 271,
+        "mekan ve eğlence": 19,
+        "mobilya ve ev tekstili": 2,
+        "mutfak araç gereç": 750,
         "mücevher saat gözlük": 276,
         "otomotiv": 739,
         "sağlık": 23,
@@ -1028,8 +1035,8 @@
     {
       "haystack_id": "tr-core5000-5",
       "n_examples": 5000,
-      "n_tokens": 509110,
-      "n_chars": 1477475,
+      "n_tokens": 511765,
+      "n_chars": 1481856,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -1038,17 +1045,17 @@
         "alışveriş": 30,
         "bilgisayar": 2,
         "elektronik": 15,
-        "emlak ve inşaat": 750,
-        "enerji": 12,
-        "etkinlik ve organizasyon": 81,
-        "eğitim": 6,
-        "finans": 49,
-        "giyim": 24,
-        "gıda": 735,
-        "hizmet sektörü": 71,
-        "internet": 8,
-        "kişisel bakım ve kozmetik": 3,
-        "küçük ev aletleri": 1,
+        "emlak ve inşaat": 12,
+        "enerji": 81,
+        "etkinlik ve organizasyon": 6,
+        "eğitim": 49,
+        "finans": 24,
+        "giyim": 735,
+        "gıda": 71,
+        "hizmet sektörü": 8,
+        "kamu hizmetleri": 3,
+        "kişisel bakım ve kozmetik": 1,
+        "küçük ev aletleri": 737,
         "medya": 750,
         "mekan ve eğlence": 43,
         "mobilya ve ev tekstili": 269,
@@ -1058,14 +1065,14 @@
         "sağlık": 737,
         "sigortacılık": 103,
         "spor": 4,
-        "ulaşım": 737
+        "ulaşım": 750
       }
     },
     {
       "haystack_id": "tr-core10000-0",
       "n_examples": 10000,
-      "n_tokens": 1017683,
-      "n_chars": 2954507,
+      "n_tokens": 1014796,
+      "n_chars": 2953210,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -1083,19 +1090,19 @@
         "giyim": 9,
         "gıda": 163,
         "hizmet sektörü": 208,
-        "internet": 1485,
-        "içecek": 411,
-        "kamu hizmetleri": 328,
-        "kişisel bakım ve kozmetik": 144,
-        "küçük ev aletleri": 10,
-        "medya": 419,
+        "internet": 411,
+        "içecek": 328,
+        "kamu hizmetleri": 144,
+        "kişisel bakım ve kozmetik": 10,
+        "küçük ev aletleri": 419,
+        "medya": 1500,
         "mekan ve eğlence": 1500,
         "mobilya ve ev tekstili": 57,
         "mutfak araç gereç": 600,
         "mücevher saat gözlük": 224,
         "otomotiv": 1484,
         "sağlık": 14,
-        "sigortacılık": 1500,
+        "sigortacılık": 1485,
         "temizlik": 8,
         "turizm": 251,
         "ulaşım": 104
@@ -1104,8 +1111,8 @@
     {
       "haystack_id": "tr-core10000-1",
       "n_examples": 10000,
-      "n_tokens": 1009886,
-      "n_chars": 2946575,
+      "n_tokens": 1015937,
+      "n_chars": 2950920,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -1124,26 +1131,26 @@
         "gıda": 2,
         "hizmet sektörü": 4,
         "içecek": 66,
-        "kişisel bakım ve kozmetik": 1500,
+        "kişisel bakım ve kozmetik": 703,
         "küçük ev aletleri": 1480,
-        "medya": 703,
-        "mekan ve eğlence": 42,
-        "mobilya ve ev tekstili": 17,
+        "medya": 42,
+        "mekan ve eğlence": 17,
+        "mobilya ve ev tekstili": 5,
         "mutfak araç gereç": 1480,
-        "mücevher saat gözlük": 5,
+        "mücevher saat gözlük": 340,
         "otomotiv": 1500,
-        "sağlık": 340,
-        "sigortacılık": 59,
-        "spor": 251,
-        "temizlik": 290,
-        "ulaşım": 292
+        "sağlık": 59,
+        "sigortacılık": 251,
+        "spor": 290,
+        "turizm": 292,
+        "ulaşım": 1500
       }
     },
     {
       "haystack_id": "tr-core10000-2",
       "n_examples": 10000,
-      "n_tokens": 1011001,
-      "n_chars": 2940244,
+      "n_tokens": 1011681,
+      "n_chars": 2941277,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -1156,11 +1163,11 @@
         "emlak ve inşaat": 2,
         "enerji": 1383,
         "eğitim": 26,
-        "finans": 75,
-        "giyim": 9,
-        "gıda": 12,
-        "internet": 37,
-        "kişisel bakım ve kozmetik": 1500,
+        "finans": 1500,
+        "giyim": 75,
+        "gıda": 9,
+        "hizmet sektörü": 12,
+        "içecek": 37,
         "küçük ev aletleri": 812,
         "medya": 1500,
         "mekan ve eğlence": 17,
@@ -1178,8 +1185,8 @@
     {
       "haystack_id": "tr-core10000-3",
       "n_examples": 10000,
-      "n_tokens": 1015778,
-      "n_chars": 2961627,
+      "n_tokens": 1016268,
+      "n_chars": 2960925,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -1202,13 +1209,13 @@
         "küçük ev aletleri": 167,
         "medya": 351,
         "mekan ve eğlence": 314,
-        "mobilya ve ev tekstili": 1481,
+        "mobilya ve ev tekstili": 66,
         "mutfak araç gereç": 1500,
-        "mücevher saat gözlük": 66,
-        "otomotiv": 8,
+        "mücevher saat gözlük": 8,
+        "otomotiv": 1500,
         "sağlık": 1483,
         "sigortacılık": 312,
-        "spor": 1500,
+        "spor": 1481,
         "temizlik": 69,
         "turizm": 13,
         "ulaşım": 131
@@ -1256,25 +1263,25 @@
     {
       "haystack_id": "tr-core10000-5",
       "n_examples": 10000,
-      "n_tokens": 1014826,
-      "n_chars": 2946741,
+      "n_tokens": 1013040,
+      "n_chars": 2946936,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
       "core_document": true,
       "label_counts": {
-        "alışveriş": 1500,
-        "beyaz eşya": 7,
-        "bilgisayar": 699,
-        "elektronik": 1500,
-        "enerji": 871,
-        "etkinlik ve organizasyon": 40,
-        "eğitim": 41,
+        "alışveriş": 7,
+        "beyaz eşya": 699,
+        "elektronik": 1481,
+        "emlak ve inşaat": 871,
+        "enerji": 40,
+        "etkinlik ve organizasyon": 41,
+        "eğitim": 1500,
         "giyim": 8,
         "gıda": 33,
         "internet": 90,
         "içecek": 828,
-        "kamu hizmetleri": 1481,
+        "kamu hizmetleri": 1483,
         "küçük ev aletleri": 157,
         "medya": 10,
         "mekan ve eğlence": 292,
@@ -1286,7 +1293,7 @@
         "sigortacılık": 6,
         "spor": 2,
         "temizlik": 249,
-        "turizm": 1483,
+        "turizm": 1500,
         "ulaşım": 1
       }
     }

sikayet_tr/questions.jsonl

@@ -23,7 +23,6 @@
 {"uid": "sikayet_tr:tr-100000-1-q9", "dataset": "sikayet_tr", "id": "tr-100000-1-q9", "haystack_uid": "sikayet_tr:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "turizm", "answer": "27", "question": "Bu kayıtlarda kaç tane 'turizm' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-100000-1-q10", "dataset": "sikayet_tr", "id": "tr-100000-1-q10", "haystack_uid": "sikayet_tr:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "medya", "answer": "25", "question": "Bu kayıtlarda kaç tane 'medya' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-100000-1-q11", "dataset": "sikayet_tr", "id": "tr-100000-1-q11", "haystack_uid": "sikayet_tr:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "temizlik", "answer": "12", "question": "Bu kayıtlarda kaç tane 'temizlik' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "sikayet_tr:tr-100000-1-q12", "dataset": "sikayet_tr", "id": "tr-100000-1-q12", "haystack_uid": "sikayet_tr:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "mücevher saat gözlük", "label_b": "alışveriş", "candidates": ["mücevher saat gözlük", "alışveriş"], "answer": "mücevher saat gözlük", "answer_key": "mücevher saat gözlük", "question": "Bu kayıtlarda hangisi daha çok: 'mücevher saat gözlük' etiketli kayıtlar mı, 'alışveriş' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-100000-2-q0", "dataset": "sikayet_tr", "id": "tr-100000-2-q0", "haystack_uid": "sikayet_tr:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "beyaz eşya", "answer": "57", "question": "Bu kayıtlarda kaç tane 'beyaz eşya' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-100000-2-q1", "dataset": "sikayet_tr", "id": "tr-100000-2-q1", "haystack_uid": "sikayet_tr:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "emlak ve inşaat", "answer": "42", "question": "Bu kayıtlarda kaç tane 'emlak ve inşaat' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-100000-2-q2", "dataset": "sikayet_tr", "id": "tr-100000-2-q2", "haystack_uid": "sikayet_tr:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "giyim", "answer": "11", "question": "Bu kayıtlarda kaç tane 'giyim' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -49,7 +48,6 @@
 {"uid": "sikayet_tr:tr-100000-3-q9", "dataset": "sikayet_tr", "id": "tr-100000-3-q9", "haystack_uid": "sikayet_tr:tr-100000-3", "haystack_id": "tr-100000-3", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "ulaşım", "answer": "25", "question": "Bu kayıtlarda kaç tane 'ulaşım' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-100000-3-q10", "dataset": "sikayet_tr", "id": "tr-100000-3-q10", "haystack_uid": "sikayet_tr:tr-100000-3", "haystack_id": "tr-100000-3", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "hizmet sektörü", "answer": "30", "question": "Bu kayıtlarda kaç tane 'hizmet sektörü' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-100000-3-q11", "dataset": "sikayet_tr", "id": "tr-100000-3-q11", "haystack_uid": "sikayet_tr:tr-100000-3", "haystack_id": "tr-100000-3", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "medya", "answer": "6", "question": "Bu kayıtlarda kaç tane 'medya' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "sikayet_tr:tr-100000-3-q12", "dataset": "sikayet_tr", "id": "tr-100000-3-q12", "haystack_uid": "sikayet_tr:tr-100000-3", "haystack_id": "tr-100000-3", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "emlak ve inşaat", "label_b": "gıda", "candidates": ["emlak ve inşaat", "gıda"], "answer": "gıda", "answer_key": "gıda", "question": "Bu kayıtlarda hangisi daha çok: 'emlak ve inşaat' etiketli kayıtlar mı, 'gıda' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-100000-4-q0", "dataset": "sikayet_tr", "id": "tr-100000-4-q0", "haystack_uid": "sikayet_tr:tr-100000-4", "haystack_id": "tr-100000-4", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "küçük ev aletleri", "answer": "19", "question": "Bu kayıtlarda kaç tane 'küçük ev aletleri' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-100000-4-q1", "dataset": "sikayet_tr", "id": "tr-100000-4-q1", "haystack_uid": "sikayet_tr:tr-100000-4", "haystack_id": "tr-100000-4", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "sağlık", "answer": "156", "question": "Bu kayıtlarda kaç tane 'sağlık' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-100000-4-q2", "dataset": "sikayet_tr", "id": "tr-100000-4-q2", "haystack_uid": "sikayet_tr:tr-100000-4", "haystack_id": "tr-100000-4", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "bilgisayar", "answer": "10", "question": "Bu kayıtlarda kaç tane 'bilgisayar' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -75,7 +73,6 @@
 {"uid": "sikayet_tr:tr-100000-5-q9", "dataset": "sikayet_tr", "id": "tr-100000-5-q9", "haystack_uid": "sikayet_tr:tr-100000-5", "haystack_id": "tr-100000-5", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "enerji", "answer": "5", "question": "Bu kayıtlarda kaç tane 'enerji' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-100000-5-q10", "dataset": "sikayet_tr", "id": "tr-100000-5-q10", "haystack_uid": "sikayet_tr:tr-100000-5", "haystack_id": "tr-100000-5", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "elektronik", "answer": "7", "question": "Bu kayıtlarda kaç tane 'elektronik' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-100000-5-q11", "dataset": "sikayet_tr", "id": "tr-100000-5-q11", "haystack_uid": "sikayet_tr:tr-100000-5", "haystack_id": "tr-100000-5", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "beyaz eşya", "answer": "29", "question": "Bu kayıtlarda kaç tane 'beyaz eşya' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "sikayet_tr:tr-100000-5-q12", "dataset": "sikayet_tr", "id": "tr-100000-5-q12", "haystack_uid": "sikayet_tr:tr-100000-5", "haystack_id": "tr-100000-5", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "kişisel bakım ve kozmetik", "label_b": "otomotiv", "candidates": ["kişisel bakım ve kozmetik", "otomotiv"], "answer": "kişisel bakım ve kozmetik", "answer_key": "kişisel bakım ve kozmetik", "question": "Bu kayıtlarda hangisi daha çok: 'kişisel bakım ve kozmetik' etiketli kayıtlar mı, 'otomotiv' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-100000-6-q0", "dataset": "sikayet_tr", "id": "tr-100000-6-q0", "haystack_uid": "sikayet_tr:tr-100000-6", "haystack_id": "tr-100000-6", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "mobilya ve ev tekstili", "answer": "45", "question": "Bu kayıtlarda kaç tane 'mobilya ve ev tekstili' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-100000-6-q1", "dataset": "sikayet_tr", "id": "tr-100000-6-q1", "haystack_uid": "sikayet_tr:tr-100000-6", "haystack_id": "tr-100000-6", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "beyaz eşya", "answer": "50", "question": "Bu kayıtlarda kaç tane 'beyaz eşya' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-100000-6-q2", "dataset": "sikayet_tr", "id": "tr-100000-6-q2", "haystack_uid": "sikayet_tr:tr-100000-6", "haystack_id": "tr-100000-6", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "hizmet sektörü", "answer": "17", "question": "Bu kayıtlarda kaç tane 'hizmet sektörü' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -88,8 +85,6 @@
 {"uid": "sikayet_tr:tr-100000-6-q9", "dataset": "sikayet_tr", "id": "tr-100000-6-q9", "haystack_uid": "sikayet_tr:tr-100000-6", "haystack_id": "tr-100000-6", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "bilgisayar", "answer": "7", "question": "Bu kayıtlarda kaç tane 'bilgisayar' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-100000-6-q10", "dataset": "sikayet_tr", "id": "tr-100000-6-q10", "haystack_uid": "sikayet_tr:tr-100000-6", "haystack_id": "tr-100000-6", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "küçük ev aletleri", "answer": "11", "question": "Bu kayıtlarda kaç tane 'küçük ev aletleri' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-100000-6-q11", "dataset": "sikayet_tr", "id": "tr-100000-6-q11", "haystack_uid": "sikayet_tr:tr-100000-6", "haystack_id": "tr-100000-6", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "internet", "answer": "8", "question": "Bu kayıtlarda kaç tane 'internet' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "sikayet_tr:tr-100000-6-q12", "dataset": "sikayet_tr", "id": "tr-100000-6-q12", "haystack_uid": "sikayet_tr:tr-100000-6", "haystack_id": "tr-100000-6", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "beyaz eşya", "label_b": "kişisel bakım ve kozmetik", "candidates": ["beyaz eşya", "kişisel bakım ve kozmetik"], "answer": "kişisel bakım ve kozmetik", "answer_key": "kişisel bakım ve kozmetik", "question": "Bu kayıtlarda hangisi daha çok: 'beyaz eşya' etiketli kayıtlar mı, 'kişisel bakım ve kozmetik' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "sikayet_tr:tr-100000-6-q13", "dataset": "sikayet_tr", "id": "tr-100000-6-q13", "haystack_uid": "sikayet_tr:tr-100000-6", "haystack_id": "tr-100000-6", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "mobilya ve ev tekstili", "label_b": "alışveriş", "candidates": ["mobilya ve ev tekstili", "alışveriş"], "answer": "alışveriş", "answer_key": "alışveriş", "question": "Bu kayıtlarda hangisi daha çok: 'mobilya ve ev tekstili' etiketli kayıtlar mı, 'alışveriş' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-100000-7-q0", "dataset": "sikayet_tr", "id": "tr-100000-7-q0", "haystack_uid": "sikayet_tr:tr-100000-7", "haystack_id": "tr-100000-7", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "hizmet sektörü", "answer": "190", "question": "Bu kayıtlarda kaç tane 'hizmet sektörü' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-100000-7-q1", "dataset": "sikayet_tr", "id": "tr-100000-7-q1", "haystack_uid": "sikayet_tr:tr-100000-7", "haystack_id": "tr-100000-7", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "sağlık", "answer": "11", "question": "Bu kayıtlarda kaç tane 'sağlık' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-100000-7-q2", "dataset": "sikayet_tr", "id": "tr-100000-7-q2", "haystack_uid": "sikayet_tr:tr-100000-7", "haystack_id": "tr-100000-7", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "otomotiv", "answer": "43", "question": "Bu kayıtlarda kaç tane 'otomotiv' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -102,7 +97,6 @@
 {"uid": "sikayet_tr:tr-100000-7-q9", "dataset": "sikayet_tr", "id": "tr-100000-7-q9", "haystack_uid": "sikayet_tr:tr-100000-7", "haystack_id": "tr-100000-7", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "spor", "answer": "14", "question": "Bu kayıtlarda kaç tane 'spor' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-100000-7-q10", "dataset": "sikayet_tr", "id": "tr-100000-7-q10", "haystack_uid": "sikayet_tr:tr-100000-7", "haystack_id": "tr-100000-7", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "emlak ve inşaat", "answer": "28", "question": "Bu kayıtlarda kaç tane 'emlak ve inşaat' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-100000-7-q11", "dataset": "sikayet_tr", "id": "tr-100000-7-q11", "haystack_uid": "sikayet_tr:tr-100000-7", "haystack_id": "tr-100000-7", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "küçük ev aletleri", "answer": "12", "question": "Bu kayıtlarda kaç tane 'küçük ev aletleri' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "sikayet_tr:tr-100000-7-q12", "dataset": "sikayet_tr", "id": "tr-100000-7-q12", "haystack_uid": "sikayet_tr:tr-100000-7", "haystack_id": "tr-100000-7", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "otomotiv", "label_b": "bilgisayar", "candidates": ["otomotiv", "bilgisayar"], "answer": "bilgisayar", "answer_key": "bilgisayar", "question": "Bu kayıtlarda hangisi daha çok: 'otomotiv' etiketli kayıtlar mı, 'bilgisayar' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-100000-8-q0", "dataset": "sikayet_tr", "id": "tr-100000-8-q0", "haystack_uid": "sikayet_tr:tr-100000-8", "haystack_id": "tr-100000-8", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "mutfak araç gereç", "answer": "18", "question": "Bu kayıtlarda kaç tane 'mutfak araç gereç' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-100000-8-q1", "dataset": "sikayet_tr", "id": "tr-100000-8-q1", "haystack_uid": "sikayet_tr:tr-100000-8", "haystack_id": "tr-100000-8", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "giyim", "answer": "31", "question": "Bu kayıtlarda kaç tane 'giyim' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-100000-8-q2", "dataset": "sikayet_tr", "id": "tr-100000-8-q2", "haystack_uid": "sikayet_tr:tr-100000-8", "haystack_id": "tr-100000-8", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "internet", "answer": "38", "question": "Bu kayıtlarda kaç tane 'internet' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -127,7 +121,7 @@
 {"uid": "sikayet_tr:tr-100000-9-q9", "dataset": "sikayet_tr", "id": "tr-100000-9-q9", "haystack_uid": "sikayet_tr:tr-100000-9", "haystack_id": "tr-100000-9", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "spor", "answer": "26", "question": "Bu kayıtlarda kaç tane 'spor' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-100000-9-q10", "dataset": "sikayet_tr", "id": "tr-100000-9-q10", "haystack_uid": "sikayet_tr:tr-100000-9", "haystack_id": "tr-100000-9", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "küçük ev aletleri", "answer": "18", "question": "Bu kayıtlarda kaç tane 'küçük ev aletleri' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-100000-9-q11", "dataset": "sikayet_tr", "id": "tr-100000-9-q11", "haystack_uid": "sikayet_tr:tr-100000-9", "haystack_id": "tr-100000-9", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "ulaşım", "answer": "11", "question": "Bu kayıtlarda kaç tane 'ulaşım' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "sikayet_tr:tr-100000-9-q12", "dataset": "sikayet_tr", "id": "tr-100000-9-q12", "haystack_uid": "sikayet_tr:tr-100000-9", "haystack_id": "tr-100000-9", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "mobilya ve ev tekstili", "label_b": "etkinlik ve organizasyon", "candidates": ["mobilya ve ev tekstili", "etkinlik ve organizasyon"], "answer": "etkinlik ve organizasyon", "answer_key": "etkinlik ve organizasyon", "question": "Bu kayıtlarda hangisi daha çok: 'mobilya ve ev tekstili' etiketli kayıtlar mı, 'etkinlik ve organizasyon' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "sikayet_tr:tr-100000-9-q12", "dataset": "sikayet_tr", "id": "tr-100000-9-q12", "haystack_uid": "sikayet_tr:tr-100000-9", "haystack_id": "tr-100000-9", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "gıda", "label_b": "etkinlik ve organizasyon", "candidates": ["gıda", "etkinlik ve organizasyon"], "answer": "etkinlik ve organizasyon", "answer_key": "etkinlik ve organizasyon", "question": "Bu kayıtlarda hangisi daha çok: 'gıda' etiketli kayıtlar mı, 'etkinlik ve organizasyon' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-250000-0-q0", "dataset": "sikayet_tr", "id": "tr-250000-0-q0", "haystack_uid": "sikayet_tr:tr-250000-0", "haystack_id": "tr-250000-0", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "label": "sağlık", "answer": "833", "question": "Bu kayıtlarda kaç tane 'sağlık' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-250000-0-q1", "dataset": "sikayet_tr", "id": "tr-250000-0-q1", "haystack_uid": "sikayet_tr:tr-250000-0", "haystack_id": "tr-250000-0", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "label": "gıda", "answer": "117", "question": "Bu kayıtlarda kaç tane 'gıda' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-250000-0-q2", "dataset": "sikayet_tr", "id": "tr-250000-0-q2", "haystack_uid": "sikayet_tr:tr-250000-0", "haystack_id": "tr-250000-0", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "label": "turizm", "answer": "32", "question": "Bu kayıtlarda kaç tane 'turizm' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -152,9 +146,7 @@
 {"uid": "sikayet_tr:tr-250000-1-q9", "dataset": "sikayet_tr", "id": "tr-250000-1-q9", "haystack_uid": "sikayet_tr:tr-250000-1", "haystack_id": "tr-250000-1", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "rare": true, "label": "hizmet sektörü", "answer": "26", "question": "Bu kayıtlarda kaç tane 'hizmet sektörü' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-250000-1-q10", "dataset": "sikayet_tr", "id": "tr-250000-1-q10", "haystack_uid": "sikayet_tr:tr-250000-1", "haystack_id": "tr-250000-1", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "rare": true, "label": "gıda", "answer": "11", "question": "Bu kayıtlarda kaç tane 'gıda' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-250000-1-q11", "dataset": "sikayet_tr", "id": "tr-250000-1-q11", "haystack_uid": "sikayet_tr:tr-250000-1", "haystack_id": "tr-250000-1", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "rare": true, "label": "spor", "answer": "15", "question": "Bu kayıtlarda kaç tane 'spor' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "sikayet_tr:tr-250000-1-q12", "dataset": "sikayet_tr", "id": "tr-250000-1-q12", "haystack_uid": "sikayet_tr:tr-250000-1", "haystack_id": "tr-250000-1", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "kamu hizmetleri", "label_b": "internet", "candidates": ["kamu hizmetleri", "internet"], "answer": "internet", "answer_key": "internet", "question": "Bu kayıtlarda hangisi daha çok: 'kamu hizmetleri' etiketli kayıtlar mı, 'internet' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "sikayet_tr:tr-250000-1-q13", "dataset": "sikayet_tr", "id": "tr-250000-1-q13", "haystack_uid": "sikayet_tr:tr-250000-1", "haystack_id": "tr-250000-1", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "otomotiv", "label_b": "kişisel bakım ve kozmetik", "candidates": ["otomotiv", "kişisel bakım ve kozmetik"], "answer": "otomotiv", "answer_key": "otomotiv", "question": "Bu kayıtlarda hangisi daha çok: 'otomotiv' etiketli kayıtlar mı, 'kişisel bakım ve kozmetik' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "sikayet_tr:tr-250000-1-q14", "dataset": "sikayet_tr", "id": "tr-250000-1-q14", "haystack_uid": "sikayet_tr:tr-250000-1", "haystack_id": "tr-250000-1", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "medya", "label_b": "beyaz eşya", "candidates": ["medya", "beyaz eşya"], "answer": "medya", "answer_key": "medya", "question": "Bu kayıtlarda hangisi daha çok: 'medya' etiketli kayıtlar mı, 'beyaz eşya' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "sikayet_tr:tr-250000-1-q12", "dataset": "sikayet_tr", "id": "tr-250000-1-q12", "haystack_uid": "sikayet_tr:tr-250000-1", "haystack_id": "tr-250000-1", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "medya", "label_b": "beyaz eşya", "candidates": ["medya", "beyaz eşya"], "answer": "medya", "answer_key": "medya", "question": "Bu kayıtlarda hangisi daha çok: 'medya' etiketli kayıtlar mı, 'beyaz eşya' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-250000-2-q0", "dataset": "sikayet_tr", "id": "tr-250000-2-q0", "haystack_uid": "sikayet_tr:tr-250000-2", "haystack_id": "tr-250000-2", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "label": "medya", "answer": "263", "question": "Bu kayıtlarda kaç tane 'medya' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-250000-2-q1", "dataset": "sikayet_tr", "id": "tr-250000-2-q1", "haystack_uid": "sikayet_tr:tr-250000-2", "haystack_id": "tr-250000-2", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "label": "kamu hizmetleri", "answer": "20", "question": "Bu kayıtlarda kaç tane 'kamu hizmetleri' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-250000-2-q2", "dataset": "sikayet_tr", "id": "tr-250000-2-q2", "haystack_uid": "sikayet_tr:tr-250000-2", "haystack_id": "tr-250000-2", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "label": "finans", "answer": "23", "question": "Bu kayıtlarda kaç tane 'finans' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -181,8 +173,7 @@
 {"uid": "sikayet_tr:tr-250000-3-q9", "dataset": "sikayet_tr", "id": "tr-250000-3-q9", "haystack_uid": "sikayet_tr:tr-250000-3", "haystack_id": "tr-250000-3", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "rare": true, "label": "mekan ve eğlence", "answer": "27", "question": "Bu kayıtlarda kaç tane 'mekan ve eğlence' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-250000-3-q10", "dataset": "sikayet_tr", "id": "tr-250000-3-q10", "haystack_uid": "sikayet_tr:tr-250000-3", "haystack_id": "tr-250000-3", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "rare": true, "label": "giyim", "answer": "5", "question": "Bu kayıtlarda kaç tane 'giyim' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-250000-3-q11", "dataset": "sikayet_tr", "id": "tr-250000-3-q11", "haystack_uid": "sikayet_tr:tr-250000-3", "haystack_id": "tr-250000-3", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "rare": true, "label": "turizm", "answer": "19", "question": "Bu kayıtlarda kaç tane 'turizm' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "sikayet_tr:tr-250000-3-q12", "dataset": "sikayet_tr", "id": "tr-250000-3-q12", "haystack_uid": "sikayet_tr:tr-250000-3", "haystack_id": "tr-250000-3", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "bilgisayar", "label_b": "küçük ev aletleri", "candidates": ["bilgisayar", "küçük ev aletleri"], "answer": "küçük ev aletleri", "answer_key": "küçük ev aletleri", "question": "Bu kayıtlarda hangisi daha çok: 'bilgisayar' etiketli kayıtlar mı, 'küçük ev aletleri' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "sikayet_tr:tr-250000-3-q13", "dataset": "sikayet_tr", "id": "tr-250000-3-q13", "haystack_uid": "sikayet_tr:tr-250000-3", "haystack_id": "tr-250000-3", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "hizmet sektörü", "label_b": "finans", "candidates": ["hizmet sektörü", "finans"], "answer": "hizmet sektörü", "answer_key": "hizmet sektörü", "question": "Bu kayıtlarda hangisi daha çok: 'hizmet sektörü' etiketli kayıtlar mı, 'finans' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "sikayet_tr:tr-250000-3-q12", "dataset": "sikayet_tr", "id": "tr-250000-3-q12", "haystack_uid": "sikayet_tr:tr-250000-3", "haystack_id": "tr-250000-3", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "elektronik", "label_b": "hizmet sektörü", "candidates": ["elektronik", "hizmet sektörü"], "answer": "hizmet sektörü", "answer_key": "hizmet sektörü", "question": "Bu kayıtlarda hangisi daha çok: 'elektronik' etiketli kayıtlar mı, 'hizmet sektörü' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-250000-4-q0", "dataset": "sikayet_tr", "id": "tr-250000-4-q0", "haystack_uid": "sikayet_tr:tr-250000-4", "haystack_id": "tr-250000-4", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "label": "etkinlik ve organizasyon", "answer": "334", "question": "Bu kayıtlarda kaç tane 'etkinlik ve organizasyon' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-250000-4-q1", "dataset": "sikayet_tr", "id": "tr-250000-4-q1", "haystack_uid": "sikayet_tr:tr-250000-4", "haystack_id": "tr-250000-4", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "label": "temizlik", "answer": "12", "question": "Bu kayıtlarda kaç tane 'temizlik' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-250000-4-q2", "dataset": "sikayet_tr", "id": "tr-250000-4-q2", "haystack_uid": "sikayet_tr:tr-250000-4", "haystack_id": "tr-250000-4", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "label": "turizm", "answer": "26", "question": "Bu kayıtlarda kaç tane 'turizm' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -195,9 +186,8 @@
 {"uid": "sikayet_tr:tr-250000-4-q9", "dataset": "sikayet_tr", "id": "tr-250000-4-q9", "haystack_uid": "sikayet_tr:tr-250000-4", "haystack_id": "tr-250000-4", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "rare": true, "label": "spor", "answer": "12", "question": "Bu kayıtlarda kaç tane 'spor' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-250000-4-q10", "dataset": "sikayet_tr", "id": "tr-250000-4-q10", "haystack_uid": "sikayet_tr:tr-250000-4", "haystack_id": "tr-250000-4", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "rare": true, "label": "mutfak araç gereç", "answer": "6", "question": "Bu kayıtlarda kaç tane 'mutfak araç gereç' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-250000-4-q11", "dataset": "sikayet_tr", "id": "tr-250000-4-q11", "haystack_uid": "sikayet_tr:tr-250000-4", "haystack_id": "tr-250000-4", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "count", "rare": true, "label": "mücevher saat gözlük", "answer": "7", "question": "Bu kayıtlarda kaç tane 'mücevher saat gözlük' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "sikayet_tr:tr-250000-4-q12", "dataset": "sikayet_tr", "id": "tr-250000-4-q12", "haystack_uid": "sikayet_tr:tr-250000-4", "haystack_id": "tr-250000-4", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "mekan ve eğlence", "label_b": "hizmet sektörü", "candidates": ["mekan ve eğlence", "hizmet sektörü"], "answer": "mekan ve eğlence", "answer_key": "mekan ve eğlence", "question": "Bu kayıtlarda hangisi daha çok: 'mekan ve eğlence' etiketli kayıtlar mı, 'hizmet sektörü' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "sikayet_tr:tr-250000-4-q12", "dataset": "sikayet_tr", "id": "tr-250000-4-q12", "haystack_uid": "sikayet_tr:tr-250000-4", "haystack_id": "tr-250000-4", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "hizmet sektörü", "label_b": "mekan ve eğlence", "candidates": ["hizmet sektörü", "mekan ve eğlence"], "answer": "mekan ve eğlence", "answer_key": "mekan ve eğlence", "question": "Bu kayıtlarda hangisi daha çok: 'hizmet sektörü' etiketli kayıtlar mı, 'mekan ve eğlence' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-250000-4-q13", "dataset": "sikayet_tr", "id": "tr-250000-4-q13", "haystack_uid": "sikayet_tr:tr-250000-4", "haystack_id": "tr-250000-4", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "finans", "label_b": "etkinlik ve organizasyon", "candidates": ["finans", "etkinlik ve organizasyon"], "answer": "etkinlik ve organizasyon", "answer_key": "etkinlik ve organizasyon", "question": "Bu kayıtlarda hangisi daha çok: 'finans' etiketli kayıtlar mı, 'etkinlik ve organizasyon' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "sikayet_tr:tr-250000-4-q14", "dataset": "sikayet_tr", "id": "tr-250000-4-q14", "haystack_uid": "sikayet_tr:tr-250000-4", "haystack_id": "tr-250000-4", "language": "tr", "target_tokens": 250000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "alışveriş", "label_b": "içecek", "candidates": ["alışveriş", "içecek"], "answer": "alışveriş", "answer_key": "alışveriş", "question": "Bu kayıtlarda hangisi daha çok: 'alışveriş' etiketli kayıtlar mı, 'içecek' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-500000-0-q0", "dataset": "sikayet_tr", "id": "tr-500000-0-q0", "haystack_uid": "sikayet_tr:tr-500000-0", "haystack_id": "tr-500000-0", "language": "tr", "target_tokens": 500000, "target_records": null, "kind": "count", "label": "sağlık", "answer": "293", "question": "Bu kayıtlarda kaç tane 'sağlık' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-500000-0-q1", "dataset": "sikayet_tr", "id": "tr-500000-0-q1", "haystack_uid": "sikayet_tr:tr-500000-0", "haystack_id": "tr-500000-0", "language": "tr", "target_tokens": 500000, "target_records": null, "kind": "count", "label": "kamu hizmetleri", "answer": "171", "question": "Bu kayıtlarda kaç tane 'kamu hizmetleri' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-500000-0-q2", "dataset": "sikayet_tr", "id": "tr-500000-0-q2", "haystack_uid": "sikayet_tr:tr-500000-0", "haystack_id": "tr-500000-0", "language": "tr", "target_tokens": 500000, "target_records": null, "kind": "count", "label": "finans", "answer": "415", "question": "Bu kayıtlarda kaç tane 'finans' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -260,9 +250,7 @@
 {"uid": "sikayet_tr:tr-500000-4-q9", "dataset": "sikayet_tr", "id": "tr-500000-4-q9", "haystack_uid": "sikayet_tr:tr-500000-4", "haystack_id": "tr-500000-4", "language": "tr", "target_tokens": 500000, "target_records": null, "kind": "count", "rare": true, "label": "mobilya ve ev tekstili", "answer": "11", "question": "Bu kayıtlarda kaç tane 'mobilya ve ev tekstili' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-500000-4-q10", "dataset": "sikayet_tr", "id": "tr-500000-4-q10", "haystack_uid": "sikayet_tr:tr-500000-4", "haystack_id": "tr-500000-4", "language": "tr", "target_tokens": 500000, "target_records": null, "kind": "count", "rare": true, "label": "alışveriş", "answer": "9", "question": "Bu kayıtlarda kaç tane 'alışveriş' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-500000-4-q11", "dataset": "sikayet_tr", "id": "tr-500000-4-q11", "haystack_uid": "sikayet_tr:tr-500000-4", "haystack_id": "tr-500000-4", "language": "tr", "target_tokens": 500000, "target_records": null, "kind": "count", "rare": true, "label": "hizmet sektörü", "answer": "19", "question": "Bu kayıtlarda kaç tane 'hizmet sektörü' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "sikayet_tr:tr-500000-4-q12", "dataset": "sikayet_tr", "id": "tr-500000-4-q12", "haystack_uid": "sikayet_tr:tr-500000-4", "haystack_id": "tr-500000-4", "language": "tr", "target_tokens": 500000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "turizm", "label_b": "emlak ve inşaat", "candidates": ["turizm", "emlak ve inşaat"], "answer": "emlak ve inşaat", "answer_key": "emlak ve inşaat", "question": "Bu kayıtlarda hangisi daha çok: 'turizm' etiketli kayıtlar mı, 'emlak ve inşaat' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "sikayet_tr:tr-500000-4-q13", "dataset": "sikayet_tr", "id": "tr-500000-4-q13", "haystack_uid": "sikayet_tr:tr-500000-4", "haystack_id": "tr-500000-4", "language": "tr", "target_tokens": 500000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "temizlik", "label_b": "eğitim", "candidates": ["temizlik", "eğitim"], "answer": "eğitim", "answer_key": "eğitim", "question": "Bu kayıtlarda hangisi daha çok: 'temizlik' etiketli kayıtlar mı, 'eğitim' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "sikayet_tr:tr-500000-4-q14", "dataset": "sikayet_tr", "id": "tr-500000-4-q14", "haystack_uid": "sikayet_tr:tr-500000-4", "haystack_id": "tr-500000-4", "language": "tr", "target_tokens": 500000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "bilgisayar", "label_b": "ulaşım", "candidates": ["bilgisayar", "ulaşım"], "answer": "ulaşım", "answer_key": "ulaşım", "question": "Bu kayıtlarda hangisi daha çok: 'bilgisayar' etiketli kayıtlar mı, 'ulaşım' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "sikayet_tr:tr-500000-4-q12", "dataset": "sikayet_tr", "id": "tr-500000-4-q12", "haystack_uid": "sikayet_tr:tr-500000-4", "haystack_id": "tr-500000-4", "language": "tr", "target_tokens": 500000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "temizlik", "label_b": "eğitim", "candidates": ["temizlik", "eğitim"], "answer": "eğitim", "answer_key": "eğitim", "question": "Bu kayıtlarda hangisi daha çok: 'temizlik' etiketli kayıtlar mı, 'eğitim' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-1000000-0-q0", "dataset": "sikayet_tr", "id": "tr-1000000-0-q0", "haystack_uid": "sikayet_tr:tr-1000000-0", "haystack_id": "tr-1000000-0", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "label": "içecek", "answer": "35", "question": "Bu kayıtlarda kaç tane 'içecek' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-1000000-0-q1", "dataset": "sikayet_tr", "id": "tr-1000000-0-q1", "haystack_uid": "sikayet_tr:tr-1000000-0", "haystack_id": "tr-1000000-0", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "label": "küçük ev aletleri", "answer": "753", "question": "Bu kayıtlarda kaç tane 'küçük ev aletleri' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-1000000-0-q2", "dataset": "sikayet_tr", "id": "tr-1000000-0-q2", "haystack_uid": "sikayet_tr:tr-1000000-0", "haystack_id": "tr-1000000-0", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "label": "spor", "answer": "104", "question": "Bu kayıtlarda kaç tane 'spor' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -276,7 +264,6 @@
 {"uid": "sikayet_tr:tr-1000000-0-q10", "dataset": "sikayet_tr", "id": "tr-1000000-0-q10", "haystack_uid": "sikayet_tr:tr-1000000-0", "haystack_id": "tr-1000000-0", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "rare": true, "label": "mekan ve eğlence", "answer": "5", "question": "Bu kayıtlarda kaç tane 'mekan ve eğlence' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-1000000-0-q11", "dataset": "sikayet_tr", "id": "tr-1000000-0-q11", "haystack_uid": "sikayet_tr:tr-1000000-0", "haystack_id": "tr-1000000-0", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "rare": true, "label": "elektronik", "answer": "12", "question": "Bu kayıtlarda kaç tane 'elektronik' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-1000000-0-q12", "dataset": "sikayet_tr", "id": "tr-1000000-0-q12", "haystack_uid": "sikayet_tr:tr-1000000-0", "haystack_id": "tr-1000000-0", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "küçük ev aletleri", "label_b": "finans", "candidates": ["küçük ev aletleri", "finans"], "answer": "finans", "answer_key": "finans", "question": "Bu kayıtlarda hangisi daha çok: 'küçük ev aletleri' etiketli kayıtlar mı, 'finans' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "sikayet_tr:tr-1000000-0-q13", "dataset": "sikayet_tr", "id": "tr-1000000-0-q13", "haystack_uid": "sikayet_tr:tr-1000000-0", "haystack_id": "tr-1000000-0", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "bilgisayar", "label_b": "emlak ve inşaat", "candidates": ["bilgisayar", "emlak ve inşaat"], "answer": "emlak ve inşaat", "answer_key": "emlak ve inşaat", "question": "Bu kayıtlarda hangisi daha çok: 'bilgisayar' etiketli kayıtlar mı, 'emlak ve inşaat' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-1000000-1-q0", "dataset": "sikayet_tr", "id": "tr-1000000-1-q0", "haystack_uid": "sikayet_tr:tr-1000000-1", "haystack_id": "tr-1000000-1", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "label": "giyim", "answer": "225", "question": "Bu kayıtlarda kaç tane 'giyim' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-1000000-1-q1", "dataset": "sikayet_tr", "id": "tr-1000000-1-q1", "haystack_uid": "sikayet_tr:tr-1000000-1", "haystack_id": "tr-1000000-1", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "label": "sağlık", "answer": "555", "question": "Bu kayıtlarda kaç tane 'sağlık' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-1000000-1-q2", "dataset": "sikayet_tr", "id": "tr-1000000-1-q2", "haystack_uid": "sikayet_tr:tr-1000000-1", "haystack_id": "tr-1000000-1", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "label": "enerji", "answer": "682", "question": "Bu kayıtlarda kaç tane 'enerji' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -289,8 +276,7 @@
 {"uid": "sikayet_tr:tr-1000000-1-q9", "dataset": "sikayet_tr", "id": "tr-1000000-1-q9", "haystack_uid": "sikayet_tr:tr-1000000-1", "haystack_id": "tr-1000000-1", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "rare": true, "label": "sigortacılık", "answer": "5", "question": "Bu kayıtlarda kaç tane 'sigortacılık' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-1000000-1-q10", "dataset": "sikayet_tr", "id": "tr-1000000-1-q10", "haystack_uid": "sikayet_tr:tr-1000000-1", "haystack_id": "tr-1000000-1", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "rare": true, "label": "alışveriş", "answer": "9", "question": "Bu kayıtlarda kaç tane 'alışveriş' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-1000000-1-q11", "dataset": "sikayet_tr", "id": "tr-1000000-1-q11", "haystack_uid": "sikayet_tr:tr-1000000-1", "haystack_id": "tr-1000000-1", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "label": "elektronik", "answer": "59", "question": "Bu kayıtlarda kaç tane 'elektronik' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
-{"uid": "sikayet_tr:tr-1000000-1-q12", "dataset": "sikayet_tr", "id": "tr-1000000-1-q12", "haystack_uid": "sikayet_tr:tr-1000000-1", "haystack_id": "tr-1000000-1", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "mobilya ve ev tekstili", "label_b": "turizm", "candidates": ["mobilya ve ev tekstili", "turizm"], "answer": "mobilya ve ev tekstili", "answer_key": "mobilya ve ev tekstili", "question": "Bu kayıtlarda hangisi daha çok: 'mobilya ve ev tekstili' etiketli kayıtlar mı, 'turizm' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "sikayet_tr:tr-1000000-1-q13", "dataset": "sikayet_tr", "id": "tr-1000000-1-q13", "haystack_uid": "sikayet_tr:tr-1000000-1", "haystack_id": "tr-1000000-1", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "etkinlik ve organizasyon", "label_b": "medya", "candidates": ["etkinlik ve organizasyon", "medya"], "answer": "etkinlik ve organizasyon", "answer_key": "etkinlik ve organizasyon", "question": "Bu kayıtlarda hangisi daha çok: 'etkinlik ve organizasyon' etiketli kayıtlar mı, 'medya' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "sikayet_tr:tr-1000000-1-q12", "dataset": "sikayet_tr", "id": "tr-1000000-1-q12", "haystack_uid": "sikayet_tr:tr-1000000-1", "haystack_id": "tr-1000000-1", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "etkinlik ve organizasyon", "label_b": "medya", "candidates": ["etkinlik ve organizasyon", "medya"], "answer": "etkinlik ve organizasyon", "answer_key": "etkinlik ve organizasyon", "question": "Bu kayıtlarda hangisi daha çok: 'etkinlik ve organizasyon' etiketli kayıtlar mı, 'medya' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-1000000-2-q0", "dataset": "sikayet_tr", "id": "tr-1000000-2-q0", "haystack_uid": "sikayet_tr:tr-1000000-2", "haystack_id": "tr-1000000-2", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "label": "bilgisayar", "answer": "2949", "question": "Bu kayıtlarda kaç tane 'bilgisayar' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-1000000-2-q1", "dataset": "sikayet_tr", "id": "tr-1000000-2-q1", "haystack_uid": "sikayet_tr:tr-1000000-2", "haystack_id": "tr-1000000-2", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "label": "giyim", "answer": "213", "question": "Bu kayıtlarda kaç tane 'giyim' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-1000000-2-q2", "dataset": "sikayet_tr", "id": "tr-1000000-2-q2", "haystack_uid": "sikayet_tr:tr-1000000-2", "haystack_id": "tr-1000000-2", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "label": "sağlık", "answer": "32", "question": "Bu kayıtlarda kaç tane 'sağlık' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -303,7 +289,6 @@
 {"uid": "sikayet_tr:tr-1000000-2-q9", "dataset": "sikayet_tr", "id": "tr-1000000-2-q9", "haystack_uid": "sikayet_tr:tr-1000000-2", "haystack_id": "tr-1000000-2", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "rare": true, "label": "internet", "answer": "8", "question": "Bu kayıtlarda kaç tane 'internet' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-1000000-2-q10", "dataset": "sikayet_tr", "id": "tr-1000000-2-q10", "haystack_uid": "sikayet_tr:tr-1000000-2", "haystack_id": "tr-1000000-2", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "rare": true, "label": "içecek", "answer": "10", "question": "Bu kayıtlarda kaç tane 'içecek' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "sikayet_tr:tr-1000000-2-q11", "dataset": "sikayet_tr", "id": "tr-1000000-2-q11", "haystack_uid": "sikayet_tr:tr-1000000-2", "haystack_id": "tr-1000000-2", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "rare": true, "label": "mutfak araç gereç", "answer": "8", "question": "Bu kayıtlarda kaç tane 'mutfak araç gereç' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "sikayet_tr:tr-1000000-2-q12", "dataset": "sikayet_tr", "id": "tr-1000000-2-q12", "haystack_uid": "sikayet_tr:tr-1000000-2", "haystack_id": "tr-1000000-2", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "giyim", "label_b": "alışveriş", "candidates": ["giyim", "alışveriş"], "answer": "giyim", "answer_key": "giyim", "question": "Bu kayıtlarda hangisi daha çok: 'giyim' etiketli kayıtlar mı, 'alışveriş' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "sikayet_tr:tr-1000000-3-q0", "dataset": "sikayet_tr", "id": "tr-1000000-3-q0", "haystack_uid": "sikayet_tr:tr-1000000-3", "haystack_id": "tr-1000000-3", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "label": "elektronik", "answer": "327", "question": "Bu kayıtlarda kaç tane 'elektronik' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-1000000-3-q1", "dataset": "sikayet_tr", "id": "tr-1000000-3-q1", "haystack_uid": "sikayet_tr:tr-1000000-3", "haystack_id": "tr-1000000-3", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "label": "sigortacılık", "answer": "141", "question": "Bu kayıtlarda kaç tane 'sigortacılık' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "sikayet_tr:tr-1000000-3-q2", "dataset": "sikayet_tr", "id": "tr-1000000-3-q2", "haystack_uid": "sikayet_tr:tr-1000000-3", "haystack_id": "tr-1000000-3", "language": "tr", "target_tokens": 1000000, "target_records": null, "kind": "count", "label": "eğitim", "answer": "1330", "question": "Bu kayıtlarda kaç tane 'eğitim' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -335,10 +320,10 @@
 {"uid": "sikayet_tr:tr-core1000-0-q1", "dataset": "sikayet_tr", "id": "tr-core1000-0-q1", "haystack_uid": "sikayet_tr:tr-core1000-0", "haystack_id": "tr-core1000-0", "language": "tr", "target_tokens": null, "target_records": 1000, "kind": "close_comparison", "label": null, "label_a": "temizlik", "label_b": "hizmet sektörü", "candidates": ["temizlik", "hizmet sektörü"], "answer": "hizmet sektörü", "answer_key": "hizmet sektörü", "question": "Bu kayıtlarda hangisi daha çok: 'temizlik' etiketli kayıtlar mı, 'hizmet sektörü' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core1000-1-q0", "dataset": "sikayet_tr", "id": "tr-core1000-1-q0", "haystack_uid": "sikayet_tr:tr-core1000-1", "haystack_id": "tr-core1000-1", "language": "tr", "target_tokens": null, "target_records": 1000, "kind": "close_comparison", "label": null, "label_a": "temizlik", "label_b": "mekan ve eğlence", "candidates": ["temizlik", "mekan ve eğlence"], "answer": "temizlik", "answer_key": "temizlik", "question": "Bu kayıtlarda hangisi daha çok: 'temizlik' etiketli kayıtlar mı, 'mekan ve eğlence' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core1000-1-q1", "dataset": "sikayet_tr", "id": "tr-core1000-1-q1", "haystack_uid": "sikayet_tr:tr-core1000-1", "haystack_id": "tr-core1000-1", "language": "tr", "target_tokens": null, "target_records": 1000, "kind": "close_comparison", "label": null, "label_a": "kamu hizmetleri", "label_b": "etkinlik ve organizasyon", "candidates": ["kamu hizmetleri", "etkinlik ve organizasyon"], "answer": "etkinlik ve organizasyon", "answer_key": "etkinlik ve organizasyon", "question": "Bu kayıtlarda hangisi daha çok: 'kamu hizmetleri' etiketli kayıtlar mı, 'etkinlik ve organizasyon' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core1000-2-q0", "dataset": "sikayet_tr", "id": "tr-core1000-2-q0", "haystack_uid": "sikayet_tr:tr-core1000-2", "haystack_id": "tr-core1000-2", "language": "tr", "target_tokens": null, "target_records": 1000, "kind": "close_comparison", "label": null, "label_a": "hizmet sektörü", "label_b": "emlak ve inşaat", "candidates": ["hizmet sektörü", "emlak ve inşaat"], "answer": "hizmet sektörü", "answer_key": "hizmet sektörü", "question": "Bu kayıtlarda hangisi daha çok: 'hizmet sektörü' etiketli kayıtlar mı, 'emlak ve inşaat' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core1000-2-q1", "dataset": "sikayet_tr", "id": "tr-core1000-2-q1", "haystack_uid": "sikayet_tr:tr-core1000-2", "haystack_id": "tr-core1000-2", "language": "tr", "target_tokens": null, "target_records": 1000, "kind": "close_comparison", "label": null, "label_a": "giyim", "label_b": "temizlik", "candidates": ["giyim", "temizlik"], "answer": "temizlik", "answer_key": "temizlik", "question": "Bu kayıtlarda hangisi daha çok: 'giyim' etiketli kayıtlar mı, 'temizlik' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core1000-2-q0", "dataset": "sikayet_tr", "id": "tr-core1000-2-q0", "haystack_uid": "sikayet_tr:tr-core1000-2", "haystack_id": "tr-core1000-2", "language": "tr", "target_tokens": null, "target_records": 1000, "kind": "close_comparison", "label": null, "label_a": "hizmet sektörü", "label_b": "temizlik", "candidates": ["hizmet sektörü", "temizlik"], "answer": "hizmet sektörü", "answer_key": "hizmet sektörü", "question": "Bu kayıtlarda hangisi daha çok: 'hizmet sektörü' etiketli kayıtlar mı, 'temizlik' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core1000-2-q1", "dataset": "sikayet_tr", "id": "tr-core1000-2-q1", "haystack_uid": "sikayet_tr:tr-core1000-2", "haystack_id": "tr-core1000-2", "language": "tr", "target_tokens": null, "target_records": 1000, "kind": "close_comparison", "label": null, "label_a": "giyim", "label_b": "eğitim", "candidates": ["giyim", "eğitim"], "answer": "eğitim", "answer_key": "eğitim", "question": "Bu kayıtlarda hangisi daha çok: 'giyim' etiketli kayıtlar mı, 'eğitim' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core1000-3-q0", "dataset": "sikayet_tr", "id": "tr-core1000-3-q0", "haystack_uid": "sikayet_tr:tr-core1000-3", "haystack_id": "tr-core1000-3", "language": "tr", "target_tokens": null, "target_records": 1000, "kind": "close_comparison", "label": null, "label_a": "beyaz eşya", "label_b": "kamu hizmetleri", "candidates": ["beyaz eşya", "kamu hizmetleri"], "answer": "beyaz eşya", "answer_key": "beyaz eşya", "question": "Bu kayıtlarda hangisi daha çok: 'beyaz eşya' etiketli kayıtlar mı, 'kamu hizmetleri' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core1000-3-q1", "dataset": "sikayet_tr", "id": "tr-core1000-3-q1", "haystack_uid": "sikayet_tr:tr-core1000-3", "haystack_id": "tr-core1000-3", "language": "tr", "target_tokens": null, "target_records": 1000, "kind": "close_comparison", "label": null, "label_a": "mekan ve eğlence", "label_b": "mobilya ve ev tekstili", "candidates": ["mekan ve eğlence", "mobilya ve ev tekstili"], "answer": "mobilya ve ev tekstili", "answer_key": "mobilya ve ev tekstili", "question": "Bu kayıtlarda hangisi daha çok: 'mekan ve eğlence' etiketli kayıtlar mı, 'mobilya ve ev tekstili' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core1000-3-q1", "dataset": "sikayet_tr", "id": "tr-core1000-3-q1", "haystack_uid": "sikayet_tr:tr-core1000-3", "haystack_id": "tr-core1000-3", "language": "tr", "target_tokens": null, "target_records": 1000, "kind": "close_comparison", "label": null, "label_a": "mekan ve eğlence", "label_b": "sigortacılık", "candidates": ["mekan ve eğlence", "sigortacılık"], "answer": "sigortacılık", "answer_key": "sigortacılık", "question": "Bu kayıtlarda hangisi daha çok: 'mekan ve eğlence' etiketli kayıtlar mı, 'sigortacılık' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core1000-4-q0", "dataset": "sikayet_tr", "id": "tr-core1000-4-q0", "haystack_uid": "sikayet_tr:tr-core1000-4", "haystack_id": "tr-core1000-4", "language": "tr", "target_tokens": null, "target_records": 1000, "kind": "close_comparison", "label": null, "label_a": "medya", "label_b": "sigortacılık", "candidates": ["medya", "sigortacılık"], "answer": "medya", "answer_key": "medya", "question": "Bu kayıtlarda hangisi daha çok: 'medya' etiketli kayıtlar mı, 'sigortacılık' etiketli kayıtlar m��? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core1000-4-q1", "dataset": "sikayet_tr", "id": "tr-core1000-4-q1", "haystack_uid": "sikayet_tr:tr-core1000-4", "haystack_id": "tr-core1000-4", "language": "tr", "target_tokens": null, "target_records": 1000, "kind": "close_comparison", "label": null, "label_a": "beyaz eşya", "label_b": "finans", "candidates": ["beyaz eşya", "finans"], "answer": "finans", "answer_key": "finans", "question": "Bu kayıtlarda hangisi daha çok: 'beyaz eşya' etiketli kayıtlar mı, 'finans' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core1000-5-q0", "dataset": "sikayet_tr", "id": "tr-core1000-5-q0", "haystack_uid": "sikayet_tr:tr-core1000-5", "haystack_id": "tr-core1000-5", "language": "tr", "target_tokens": null, "target_records": 1000, "kind": "close_comparison", "label": null, "label_a": "giyim", "label_b": "temizlik", "candidates": ["giyim", "temizlik"], "answer": "giyim", "answer_key": "giyim", "question": "Bu kayıtlarda hangisi daha çok: 'giyim' etiketli kayıtlar mı, 'temizlik' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
@@ -347,35 +332,35 @@
 {"uid": "sikayet_tr:tr-core2500-0-q1", "dataset": "sikayet_tr", "id": "tr-core2500-0-q1", "haystack_uid": "sikayet_tr:tr-core2500-0", "haystack_id": "tr-core2500-0", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "spor", "label_b": "finans", "candidates": ["spor", "finans"], "answer": "finans", "answer_key": "finans", "question": "Bu kayıtlarda hangisi daha çok: 'spor' etiketli kayıtlar mı, 'finans' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core2500-1-q0", "dataset": "sikayet_tr", "id": "tr-core2500-1-q0", "haystack_uid": "sikayet_tr:tr-core2500-1", "haystack_id": "tr-core2500-1", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "mücevher saat gözlük", "label_b": "medya", "candidates": ["mücevher saat gözlük", "medya"], "answer": "mücevher saat gözlük", "answer_key": "mücevher saat gözlük", "question": "Bu kayıtlarda hangisi daha çok: 'mücevher saat gözlük' etiketli kayıtlar mı, 'medya' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core2500-1-q1", "dataset": "sikayet_tr", "id": "tr-core2500-1-q1", "haystack_uid": "sikayet_tr:tr-core2500-1", "haystack_id": "tr-core2500-1", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "kamu hizmetleri", "label_b": "elektronik", "candidates": ["kamu hizmetleri", "elektronik"], "answer": "elektronik", "answer_key": "elektronik", "question": "Bu kayıtlarda hangisi daha çok: 'kamu hizmetleri' etiketli kayıtlar mı, 'elektronik' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core2500-2-q0", "dataset": "sikayet_tr", "id": "tr-core2500-2-q0", "haystack_uid": "sikayet_tr:tr-core2500-2", "haystack_id": "tr-core2500-2", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "sağlık", "label_b": "mobilya ve ev tekstili", "candidates": ["sağlık", "mobilya ve ev tekstili"], "answer": "sağlık", "answer_key": "sağlık", "question": "Bu kayıtlarda hangisi daha çok: 'sağlık' etiketli kayıtlar mı, 'mobilya ve ev tekstili' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core2500-2-q1", "dataset": "sikayet_tr", "id": "tr-core2500-2-q1", "haystack_uid": "sikayet_tr:tr-core2500-2", "haystack_id": "tr-core2500-2", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "spor", "label_b": "gıda", "candidates": ["spor", "gıda"], "answer": "gıda", "answer_key": "gıda", "question": "Bu kayıtlarda hangisi daha çok: 'spor' etiketli kayıtlar mı, 'gıda' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core2500-3-q0", "dataset": "sikayet_tr", "id": "tr-core2500-3-q0", "haystack_uid": "sikayet_tr:tr-core2500-3", "haystack_id": "tr-core2500-3", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "kişisel bakım ve kozmetik", "label_b": "gıda", "candidates": ["kişisel bakım ve kozmetik", "gıda"], "answer": "kişisel bakım ve kozmetik", "answer_key": "kişisel bakım ve kozmetik", "question": "Bu kayıtlarda hangisi daha çok: 'kişisel bakım ve kozmetik' etiketli kayıtlar mı, 'gıda' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core2500-3-q1", "dataset": "sikayet_tr", "id": "tr-core2500-3-q1", "haystack_uid": "sikayet_tr:tr-core2500-3", "haystack_id": "tr-core2500-3", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "eğitim", "label_b": "içecek", "candidates": ["eğitim", "içecek"], "answer": "içecek", "answer_key": "içecek", "question": "Bu kayıtlarda hangisi daha çok: 'eğitim' etiketli kayıtlar mı, 'içecek' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core2500-4-q0", "dataset": "sikayet_tr", "id": "tr-core2500-4-q0", "haystack_uid": "sikayet_tr:tr-core2500-4", "haystack_id": "tr-core2500-4", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "beyaz eşya", "label_b": "bilgisayar", "candidates": ["beyaz eşya", "bilgisayar"], "answer": "beyaz eşya", "answer_key": "beyaz eşya", "question": "Bu kayıtlarda hangisi daha çok: 'beyaz eşya' etiketli kayıtlar mı, 'bilgisayar' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core2500-4-q1", "dataset": "sikayet_tr", "id": "tr-core2500-4-q1", "haystack_uid": "sikayet_tr:tr-core2500-4", "haystack_id": "tr-core2500-4", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "otomotiv", "label_b": "mücevher saat gözlük", "candidates": ["otomotiv", "mücevher saat gözlük"], "answer": "mücevher saat gözlük", "answer_key": "mücevher saat gözlük", "question": "Bu kayıtlarda hangisi daha çok: 'otomotiv' etiketli kayıtlar mı, 'mücevher saat gözlük' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core2500-5-q0", "dataset": "sikayet_tr", "id": "tr-core2500-5-q0", "haystack_uid": "sikayet_tr:tr-core2500-5", "haystack_id": "tr-core2500-5", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "mücevher saat gözlük", "label_b": "internet", "candidates": ["mücevher saat gözlük", "internet"], "answer": "mücevher saat gözlük", "answer_key": "mücevher saat gözlük", "question": "Bu kayıtlarda hangisi daha çok: 'mücevher saat gözlük' etiketli kayıtlar mı, 'internet' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core2500-5-q1", "dataset": "sikayet_tr", "id": "tr-core2500-5-q1", "haystack_uid": "sikayet_tr:tr-core2500-5", "haystack_id": "tr-core2500-5", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "mutfak araç gereç", "label_b": "sigortacılık", "candidates": ["mutfak araç gereç", "sigortacılık"], "answer": "sigortacılık", "answer_key": "sigortacılık", "question": "Bu kayıtlarda hangisi daha çok: 'mutfak araç gereç' etiketli kayıtlar mı, 'sigortacılık' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core5000-0-q0", "dataset": "sikayet_tr", "id": "tr-core5000-0-q0", "haystack_uid": "sikayet_tr:tr-core5000-0", "haystack_id": "tr-core5000-0", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "internet", "label_b": "bilgisayar", "candidates": ["internet", "bilgisayar"], "answer": "internet", "answer_key": "internet", "question": "Bu kayıtlarda hangisi daha çok: 'internet' etiketli kayıtlar mı, 'bilgisayar' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core5000-0-q1", "dataset": "sikayet_tr", "id": "tr-core5000-0-q1", "haystack_uid": "sikayet_tr:tr-core5000-0", "haystack_id": "tr-core5000-0", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "etkinlik ve organizasyon", "label_b": "hizmet sektörü", "candidates": ["etkinlik ve organizasyon", "hizmet sektörü"], "answer": "hizmet sektörü", "answer_key": "hizmet sektörü", "question": "Bu kayıtlarda hangisi daha çok: 'etkinlik ve organizasyon' etiketli kayıtlar mı, 'hizmet sektörü' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core5000-1-q0", "dataset": "sikayet_tr", "id": "tr-core5000-1-q0", "haystack_uid": "sikayet_tr:tr-core5000-1", "haystack_id": "tr-core5000-1", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "bilgisayar", "label_b": "internet", "candidates": ["bilgisayar", "internet"], "answer": "bilgisayar", "answer_key": "bilgisayar", "question": "Bu kayıtlarda hangisi daha çok: 'bilgisayar' etiketli kayıtlar mı, 'internet' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core5000-1-q1", "dataset": "sikayet_tr", "id": "tr-core5000-1-q1", "haystack_uid": "sikayet_tr:tr-core5000-1", "haystack_id": "tr-core5000-1", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "sağlık", "label_b": "mobilya ve ev tekstili", "candidates": ["sağlık", "mobilya ve ev tekstili"], "answer": "mobilya ve ev tekstili", "answer_key": "mobilya ve ev tekstili", "question": "Bu kayıtlarda hangisi daha çok: 'sağlık' etiketli kayıtlar mı, 'mobilya ve ev tekstili' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core5000-2-q0", "dataset": "sikayet_tr", "id": "tr-core5000-2-q0", "haystack_uid": "sikayet_tr:tr-core5000-2", "haystack_id": "tr-core5000-2", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "etkinlik ve organizasyon", "label_b": "bilgisayar", "candidates": ["etkinlik ve organizasyon", "bilgisayar"], "answer": "etkinlik ve organizasyon", "answer_key": "etkinlik ve organizasyon", "question": "Bu kayıtlarda hangisi daha çok: 'etkinlik ve organizasyon' etiketli kayıtlar mı, 'bilgisayar' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core5000-2-q1", "dataset": "sikayet_tr", "id": "tr-core5000-2-q1", "haystack_uid": "sikayet_tr:tr-core5000-2", "haystack_id": "tr-core5000-2", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "emlak ve inşaat", "label_b": "mutfak araç gereç", "candidates": ["emlak ve inşaat", "mutfak araç gereç"], "answer": "mutfak araç gereç", "answer_key": "mutfak araç gereç", "question": "Bu kayıtlarda hangisi daha çok: 'emlak ve inşaat' etiketli kayıtlar mı, 'mutfak araç gereç' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core2500-2-q0", "dataset": "sikayet_tr", "id": "tr-core2500-2-q0", "haystack_uid": "sikayet_tr:tr-core2500-2", "haystack_id": "tr-core2500-2", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "sağlık", "label_b": "spor", "candidates": ["sağlık", "spor"], "answer": "sağlık", "answer_key": "sağlık", "question": "Bu kayıtlarda hangisi daha çok: 'sağlık' etiketli kayıtlar mı, 'spor' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core2500-2-q1", "dataset": "sikayet_tr", "id": "tr-core2500-2-q1", "haystack_uid": "sikayet_tr:tr-core2500-2", "haystack_id": "tr-core2500-2", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "gıda", "label_b": "kamu hizmetleri", "candidates": ["gıda", "kamu hizmetleri"], "answer": "kamu hizmetleri", "answer_key": "kamu hizmetleri", "question": "Bu kayıtlarda hangisi daha çok: 'gıda' etiketli kayıtlar mı, 'kamu hizmetleri' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core2500-3-q0", "dataset": "sikayet_tr", "id": "tr-core2500-3-q0", "haystack_uid": "sikayet_tr:tr-core2500-3", "haystack_id": "tr-core2500-3", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "içecek", "label_b": "gıda", "candidates": ["içecek", "gıda"], "answer": "içecek", "answer_key": "içecek", "question": "Bu kayıtlarda hangisi daha çok: 'içecek' etiketli kayıtlar mı, 'gıda' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core2500-3-q1", "dataset": "sikayet_tr", "id": "tr-core2500-3-q1", "haystack_uid": "sikayet_tr:tr-core2500-3", "haystack_id": "tr-core2500-3", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "elektronik", "label_b": "eğitim", "candidates": ["elektronik", "eğitim"], "answer": "eğitim", "answer_key": "eğitim", "question": "Bu kayıtlarda hangisi daha çok: 'elektronik' etiketli kayıtlar mı, 'eğitim' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core2500-4-q0", "dataset": "sikayet_tr", "id": "tr-core2500-4-q0", "haystack_uid": "sikayet_tr:tr-core2500-4", "haystack_id": "tr-core2500-4", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "beyaz eşya", "label_b": "mücevher saat gözlük", "candidates": ["beyaz eşya", "mücevher saat gözlük"], "answer": "beyaz eşya", "answer_key": "beyaz eşya", "question": "Bu kayıtlarda hangisi daha çok: 'beyaz eşya' etiketli kayıtlar mı, 'mücevher saat gözlük' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core2500-4-q1", "dataset": "sikayet_tr", "id": "tr-core2500-4-q1", "haystack_uid": "sikayet_tr:tr-core2500-4", "haystack_id": "tr-core2500-4", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "otomotiv", "label_b": "finans", "candidates": ["otomotiv", "finans"], "answer": "finans", "answer_key": "finans", "question": "Bu kayıtlarda hangisi daha çok: 'otomotiv' etiketli kayıtlar mı, 'finans' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core2500-5-q0", "dataset": "sikayet_tr", "id": "tr-core2500-5-q0", "haystack_uid": "sikayet_tr:tr-core2500-5", "haystack_id": "tr-core2500-5", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "mücevher saat gözlük", "label_b": "sigortacılık", "candidates": ["mücevher saat gözlük", "sigortacılık"], "answer": "mücevher saat gözlük", "answer_key": "mücevher saat gözlük", "question": "Bu kayıtlarda hangisi daha çok: 'mücevher saat gözlük' etiketli kayıtlar mı, 'sigortacılık' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core2500-5-q1", "dataset": "sikayet_tr", "id": "tr-core2500-5-q1", "haystack_uid": "sikayet_tr:tr-core2500-5", "haystack_id": "tr-core2500-5", "language": "tr", "target_tokens": null, "target_records": 2500, "kind": "close_comparison", "label": null, "label_a": "sağlık", "label_b": "mutfak araç gereç", "candidates": ["sağlık", "mutfak araç gereç"], "answer": "mutfak araç gereç", "answer_key": "mutfak araç gereç", "question": "Bu kayıtlarda hangisi daha çok: 'sağlık' etiketli kayıtlar mı, 'mutfak araç gereç' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core5000-0-q0", "dataset": "sikayet_tr", "id": "tr-core5000-0-q0", "haystack_uid": "sikayet_tr:tr-core5000-0", "haystack_id": "tr-core5000-0", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "hizmet sektörü", "label_b": "etkinlik ve organizasyon", "candidates": ["hizmet sektörü", "etkinlik ve organizasyon"], "answer": "hizmet sektörü", "answer_key": "hizmet sektörü", "question": "Bu kayıtlarda hangisi daha çok: 'hizmet sektörü' etiketli kayıtlar mı, 'etkinlik ve organizasyon' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core5000-0-q1", "dataset": "sikayet_tr", "id": "tr-core5000-0-q1", "haystack_uid": "sikayet_tr:tr-core5000-0", "haystack_id": "tr-core5000-0", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "giyim", "label_b": "içecek", "candidates": ["giyim", "içecek"], "answer": "içecek", "answer_key": "içecek", "question": "Bu kayıtlarda hangisi daha çok: 'giyim' etiketli kayıtlar mı, 'içecek' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core5000-1-q0", "dataset": "sikayet_tr", "id": "tr-core5000-1-q0", "haystack_uid": "sikayet_tr:tr-core5000-1", "haystack_id": "tr-core5000-1", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "spor", "label_b": "sağlık", "candidates": ["spor", "sağlık"], "answer": "spor", "answer_key": "spor", "question": "Bu kayıtlarda hangisi daha çok: 'spor' etiketli kayıtlar mı, 'sağlık' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core5000-1-q1", "dataset": "sikayet_tr", "id": "tr-core5000-1-q1", "haystack_uid": "sikayet_tr:tr-core5000-1", "haystack_id": "tr-core5000-1", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "beyaz eşya", "label_b": "turizm", "candidates": ["beyaz eşya", "turizm"], "answer": "turizm", "answer_key": "turizm", "question": "Bu kayıtlarda hangisi daha çok: 'beyaz eşya' etiketli kayıtlar mı, 'turizm' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core5000-2-q0", "dataset": "sikayet_tr", "id": "tr-core5000-2-q0", "haystack_uid": "sikayet_tr:tr-core5000-2", "haystack_id": "tr-core5000-2", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "mutfak araç gereç", "label_b": "etkinlik ve organizasyon", "candidates": ["mutfak araç gereç", "etkinlik ve organizasyon"], "answer": "mutfak araç gereç", "answer_key": "mutfak araç gereç", "question": "Bu kayıtlarda hangisi daha çok: 'mutfak araç gereç' etiketli kayıtlar mı, 'etkinlik ve organizasyon' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core5000-2-q1", "dataset": "sikayet_tr", "id": "tr-core5000-2-q1", "haystack_uid": "sikayet_tr:tr-core5000-2", "haystack_id": "tr-core5000-2", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "giyim", "label_b": "finans", "candidates": ["giyim", "finans"], "answer": "finans", "answer_key": "finans", "question": "Bu kayıtlarda hangisi daha çok: 'giyim' etiketli kayıtlar mı, 'finans' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core5000-3-q0", "dataset": "sikayet_tr", "id": "tr-core5000-3-q0", "haystack_uid": "sikayet_tr:tr-core5000-3", "haystack_id": "tr-core5000-3", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "etkinlik ve organizasyon", "label_b": "turizm", "candidates": ["etkinlik ve organizasyon", "turizm"], "answer": "etkinlik ve organizasyon", "answer_key": "etkinlik ve organizasyon", "question": "Bu kayıtlarda hangisi daha çok: 'etkinlik ve organizasyon' etiketli kayıtlar mı, 'turizm' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core5000-3-q1", "dataset": "sikayet_tr", "id": "tr-core5000-3-q1", "haystack_uid": "sikayet_tr:tr-core5000-3", "haystack_id": "tr-core5000-3", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "mekan ve eğlence", "label_b": "spor", "candidates": ["mekan ve eğlence", "spor"], "answer": "spor", "answer_key": "spor", "question": "Bu kayıtlarda hangisi daha çok: 'mekan ve eğlence' etiketli kayıtlar mı, 'spor' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core5000-4-q0", "dataset": "sikayet_tr", "id": "tr-core5000-4-q0", "haystack_uid": "sikayet_tr:tr-core5000-4", "haystack_id": "tr-core5000-4", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "internet", "label_b": "otomotiv", "candidates": ["internet", "otomotiv"], "answer": "internet", "answer_key": "internet", "question": "Bu kayıtlarda hangisi daha çok: 'internet' etiketli kayıtlar mı, 'otomotiv' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core5000-4-q1", "dataset": "sikayet_tr", "id": "tr-core5000-4-q1", "haystack_uid": "sikayet_tr:tr-core5000-4", "haystack_id": "tr-core5000-4", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "sigortacılık", "label_b": "küçük ev aletleri", "candidates": ["sigortacılık", "küçük ev aletleri"], "answer": "küçük ev aletleri", "answer_key": "küçük ev aletleri", "question": "Bu kayıtlarda hangisi daha çok: 'sigortacılık' etiketli kayıtlar mı, 'küçük ev aletleri' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core5000-5-q0", "dataset": "sikayet_tr", "id": "tr-core5000-5-q0", "haystack_uid": "sikayet_tr:tr-core5000-5", "haystack_id": "tr-core5000-5", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "emlak ve inşaat", "label_b": "sağlık", "candidates": ["emlak ve inşaat", "sağlık"], "answer": "emlak ve inşaat", "answer_key": "emlak ve inşaat", "question": "Bu kayıtlarda hangisi daha çok: 'emlak ve inşaat' etiketli kayıtlar mı, 'sağlık' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core5000-5-q1", "dataset": "sikayet_tr", "id": "tr-core5000-5-q1", "haystack_uid": "sikayet_tr:tr-core5000-5", "haystack_id": "tr-core5000-5", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "ulaşım", "label_b": "medya", "candidates": ["ulaşım", "medya"], "answer": "medya", "answer_key": "medya", "question": "Bu kayıtlarda hangisi daha çok: 'ulaşım' etiketli kayıtlar mı, 'medya' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core10000-0-q0", "dataset": "sikayet_tr", "id": "tr-core10000-0-q0", "haystack_uid": "sikayet_tr:tr-core10000-0", "haystack_id": "tr-core10000-0", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "mekan ve eğlence", "label_b": "internet", "candidates": ["mekan ve eğlence", "internet"], "answer": "mekan ve eğlence", "answer_key": "mekan ve eğlence", "question": "Bu kayıtlarda hangisi daha çok: 'mekan ve eğlence' etiketli kayıtlar mı, 'internet' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core10000-0-q1", "dataset": "sikayet_tr", "id": "tr-core10000-0-q1", "haystack_uid": "sikayet_tr:tr-core10000-0", "haystack_id": "tr-core10000-0", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "otomotiv", "label_b": "sigortacılık", "candidates": ["otomotiv", "sigortacılık"], "answer": "sigortacılık", "answer_key": "sigortacılık", "question": "Bu kayıtlarda hangisi daha çok: 'otomotiv' etiketli kayıtlar mı, 'sigortacılık' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core5000-4-q0", "dataset": "sikayet_tr", "id": "tr-core5000-4-q0", "haystack_uid": "sikayet_tr:tr-core5000-4", "haystack_id": "tr-core5000-4", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "küçük ev aletleri", "label_b": "otomotiv", "candidates": ["küçük ev aletleri", "otomotiv"], "answer": "küçük ev aletleri", "answer_key": "küçük ev aletleri", "question": "Bu kayıtlarda hangisi daha çok: 'küçük ev aletleri' etiketli kayıtlar mı, 'otomotiv' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core5000-4-q1", "dataset": "sikayet_tr", "id": "tr-core5000-4-q1", "haystack_uid": "sikayet_tr:tr-core5000-4", "haystack_id": "tr-core5000-4", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "sigortacılık", "label_b": "mutfak araç gereç", "candidates": ["sigortacılık", "mutfak araç gereç"], "answer": "mutfak araç gereç", "answer_key": "mutfak araç gereç", "question": "Bu kayıtlarda hangisi daha çok: 'sigortacılık' etiketli kayıtlar mı, 'mutfak araç gereç' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core5000-5-q0", "dataset": "sikayet_tr", "id": "tr-core5000-5-q0", "haystack_uid": "sikayet_tr:tr-core5000-5", "haystack_id": "tr-core5000-5", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "medya", "label_b": "sağlık", "candidates": ["medya", "sağlık"], "answer": "medya", "answer_key": "medya", "question": "Bu kayıtlarda hangisi daha çok: 'medya' etiketli kayıtlar mı, 'sağlık' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core5000-5-q1", "dataset": "sikayet_tr", "id": "tr-core5000-5-q1", "haystack_uid": "sikayet_tr:tr-core5000-5", "haystack_id": "tr-core5000-5", "language": "tr", "target_tokens": null, "target_records": 5000, "kind": "close_comparison", "label": null, "label_a": "küçük ev aletleri", "label_b": "ulaşım", "candidates": ["küçük ev aletleri", "ulaşım"], "answer": "ulaşım", "answer_key": "ulaşım", "question": "Bu kayıtlarda hangisi daha çok: 'küçük ev aletleri' etiketli kayıtlar mı, 'ulaşım' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core10000-0-q0", "dataset": "sikayet_tr", "id": "tr-core10000-0-q0", "haystack_uid": "sikayet_tr:tr-core10000-0", "haystack_id": "tr-core10000-0", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "mekan ve eğlence", "label_b": "sigortacılık", "candidates": ["mekan ve eğlence", "sigortacılık"], "answer": "mekan ve eğlence", "answer_key": "mekan ve eğlence", "question": "Bu kayıtlarda hangisi daha çok: 'mekan ve eğlence' etiketli kayıtlar mı, 'sigortacılık' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core10000-0-q1", "dataset": "sikayet_tr", "id": "tr-core10000-0-q1", "haystack_uid": "sikayet_tr:tr-core10000-0", "haystack_id": "tr-core10000-0", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "otomotiv", "label_b": "medya", "candidates": ["otomotiv", "medya"], "answer": "medya", "answer_key": "medya", "question": "Bu kayıtlarda hangisi daha çok: 'otomotiv' etiketli kayıtlar mı, 'medya' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core10000-1-q0", "dataset": "sikayet_tr", "id": "tr-core10000-1-q0", "haystack_uid": "sikayet_tr:tr-core10000-1", "haystack_id": "tr-core10000-1", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "otomotiv", "label_b": "mutfak araç gereç", "candidates": ["otomotiv", "mutfak araç gereç"], "answer": "otomotiv", "answer_key": "otomotiv", "question": "Bu kayıtlarda hangisi daha çok: 'otomotiv' etiketli kayıtlar mı, 'mutfak araç gereç' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core10000-1-q1", "dataset": "sikayet_tr", "id": "tr-core10000-1-q1", "haystack_uid": "sikayet_tr:tr-core10000-1", "haystack_id": "tr-core10000-1", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "küçük ev aletleri", "label_b": "kişisel bakım ve kozmetik", "candidates": ["küçük ev aletleri", "kişisel bakım ve kozmetik"], "answer": "kişisel bakım ve kozmetik", "answer_key": "kişisel bakım ve kozmetik", "question": "Bu kayıtlarda hangisi daha çok: 'küçük ev aletleri' etiketli kayıtlar mı, 'kişisel bakım ve kozmetik' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core10000-1-q1", "dataset": "sikayet_tr", "id": "tr-core10000-1-q1", "haystack_uid": "sikayet_tr:tr-core10000-1", "haystack_id": "tr-core10000-1", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "küçük ev aletleri", "label_b": "ulaşım", "candidates": ["küçük ev aletleri", "ulaşım"], "answer": "ulaşım", "answer_key": "ulaşım", "question": "Bu kayıtlarda hangisi daha çok: 'küçük ev aletleri' etiketli kayıtlar mı, 'ulaşım' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core10000-2-q0", "dataset": "sikayet_tr", "id": "tr-core10000-2-q0", "haystack_uid": "sikayet_tr:tr-core10000-2", "haystack_id": "tr-core10000-2", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "medya", "label_b": "spor", "candidates": ["medya", "spor"], "answer": "medya", "answer_key": "medya", "question": "Bu kayıtlarda hangisi daha çok: 'medya' etiketli kayıtlar mı, 'spor' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core10000-2-q1", "dataset": "sikayet_tr", "id": "tr-core10000-2-q1", "haystack_uid": "sikayet_tr:tr-core10000-2", "haystack_id": "tr-core10000-2", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "otomotiv", "label_b": "kişisel bakım ve kozmetik", "candidates": ["otomotiv", "kişisel bakım ve kozmetik"], "answer": "kişisel bakım ve kozmetik", "answer_key": "kişisel bakım ve kozmetik", "question": "Bu kayıtlarda hangisi daha çok: 'otomotiv' etiketli kayıtlar mı, 'kişisel bakım ve kozmetik' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core10000-2-q1", "dataset": "sikayet_tr", "id": "tr-core10000-2-q1", "haystack_uid": "sikayet_tr:tr-core10000-2", "haystack_id": "tr-core10000-2", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "otomotiv", "label_b": "finans", "candidates": ["otomotiv", "finans"], "answer": "finans", "answer_key": "finans", "question": "Bu kayıtlarda hangisi daha çok: 'otomotiv' etiketli kayıtlar mı, 'finans' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core10000-3-q0", "dataset": "sikayet_tr", "id": "tr-core10000-3-q0", "haystack_uid": "sikayet_tr:tr-core10000-3", "haystack_id": "tr-core10000-3", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "mutfak araç gereç", "label_b": "sağlık", "candidates": ["mutfak araç gereç", "sağlık"], "answer": "mutfak araç gereç", "answer_key": "mutfak araç gereç", "question": "Bu kayıtlarda hangisi daha çok: 'mutfak araç gereç' etiketli kayıtlar mı, 'sağlık' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core10000-3-q1", "dataset": "sikayet_tr", "id": "tr-core10000-3-q1", "haystack_uid": "sikayet_tr:tr-core10000-3", "haystack_id": "tr-core10000-3", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "mobilya ve ev tekstili", "label_b": "spor", "candidates": ["mobilya ve ev tekstili", "spor"], "answer": "spor", "answer_key": "spor", "question": "Bu kayıtlarda hangisi daha çok: 'mobilya ve ev tekstili' etiketli kayıtlar mı, 'spor' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core10000-3-q1", "dataset": "sikayet_tr", "id": "tr-core10000-3-q1", "haystack_uid": "sikayet_tr:tr-core10000-3", "haystack_id": "tr-core10000-3", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "spor", "label_b": "otomotiv", "candidates": ["spor", "otomotiv"], "answer": "otomotiv", "answer_key": "otomotiv", "question": "Bu kayıtlarda hangisi daha çok: 'spor' etiketli kayıtlar mı, 'otomotiv' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core10000-4-q0", "dataset": "sikayet_tr", "id": "tr-core10000-4-q0", "haystack_uid": "sikayet_tr:tr-core10000-4", "haystack_id": "tr-core10000-4", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "mutfak araç gereç", "label_b": "otomotiv", "candidates": ["mutfak araç gereç", "otomotiv"], "answer": "mutfak araç gereç", "answer_key": "mutfak araç gereç", "question": "Bu kayıtlarda hangisi daha çok: 'mutfak araç gereç' etiketli kayıtlar mı, 'otomotiv' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "sikayet_tr:tr-core10000-4-q1", "dataset": "sikayet_tr", "id": "tr-core10000-4-q1", "haystack_uid": "sikayet_tr:tr-core10000-4", "haystack_id": "tr-core10000-4", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "elektronik", "label_b": "beyaz eşya", "candidates": ["elektronik", "beyaz eşya"], "answer": "beyaz eşya", "answer_key": "beyaz eşya", "question": "Bu kayıtlarda hangisi daha çok: 'elektronik' etiketli kayıtlar mı, 'beyaz eşya' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core10000-5-q0", "dataset": "sikayet_tr", "id": "tr-core10000-5-q0", "haystack_uid": "sikayet_tr:tr-core10000-5", "haystack_id": "tr-core10000-5", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "alışveriş", "label_b": "turizm", "candidates": ["alışveriş", "turizm"], "answer": "alışveriş", "answer_key": "alışveriş", "question": "Bu kayıtlarda hangisi daha çok: 'alışveriş' etiketli kayıtlar mı, 'turizm' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "sikayet_tr:tr-core10000-5-q1", "dataset": "sikayet_tr", "id": "tr-core10000-5-q1", "haystack_uid": "sikayet_tr:tr-core10000-5", "haystack_id": "tr-core10000-5", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "kamu hizmetleri", "label_b": "elektronik", "candidates": ["kamu hizmetleri", "elektronik"], "answer": "elektronik", "answer_key": "elektronik", "question": "Bu kayıtlarda hangisi daha çok: 'kamu hizmetleri' etiketli kayıtlar mı, 'elektronik' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core10000-5-q0", "dataset": "sikayet_tr", "id": "tr-core10000-5-q0", "haystack_uid": "sikayet_tr:tr-core10000-5", "haystack_id": "tr-core10000-5", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "turizm", "label_b": "kamu hizmetleri", "candidates": ["turizm", "kamu hizmetleri"], "answer": "turizm", "answer_key": "turizm", "question": "Bu kayıtlarda hangisi daha çok: 'turizm' etiketli kayıtlar mı, 'kamu hizmetleri' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "sikayet_tr:tr-core10000-5-q1", "dataset": "sikayet_tr", "id": "tr-core10000-5-q1", "haystack_uid": "sikayet_tr:tr-core10000-5", "haystack_id": "tr-core10000-5", "language": "tr", "target_tokens": null, "target_records": 10000, "kind": "close_comparison", "label": null, "label_a": "elektronik", "label_b": "eğitim", "candidates": ["elektronik", "eğitim"], "answer": "eğitim", "answer_key": "eğitim", "question": "Bu kayıtlarda hangisi daha çok: 'elektronik' etiketli kayıtlar mı, 'eğitim' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}

sinema_tr/haystacks.jsonl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bdbfbed56a3d808844ee13db5d1088f57ebc23736d9070c2f047b44a3239cb0e
-size 52889185
+oid sha256:10fe2533e399c5a3af6777e373c887d2634d6bf341fe5686db28390db0dc6096
+size 44541810

sinema_tr/manifest.json

@@ -1,6 +1,6 @@
 {
-  "version": "0.9.0",
-  "date": "2026-09-29",
+  "version": "0.10.0",
+  "date": "2026-10-01",
   "environment": {
     "python": "3.12.9",
     "polars": "1.42.0",
@@ -17,6 +17,17 @@
     "drop_label_leakage": true,
     "leak_label_words": false,
     "min_class_support": 1000,
+    "boilerplate_keywords": [],
+    "boilerplate_min_keywords": 3,
+    "boilerplate_weak_min": 2,
+    "boilerplate_regex": "",
+    "drop_text_regex": [
+      "\\b\\d{1,3}([.,]\\d)?\\s*/\\s*(10|5|100)\\b",
+      "\\b\\d{1,2}\\s*(yıldız|puan)",
+      "\\bpuan[ıi]m\\b",
+      "\\b(on|dokuz|sekiz|yedi|altı|beş|dört|üç|iki|bir|sıfır)\\s+(yıldız|puan)"
+    ],
+    "strip_html": true,
     "seed": 42,
     "haystack_target_tokens": [
       100000,
@@ -34,7 +45,18 @@
     "licence": "cc-by-sa-4.0",
     "label_provenance": "author_stars",
     "text_provenance": "human_written",
-    "label_translation": {},
+    "label_translation": {
+      "1 yıldız": "olumsuz",
+      "2 yıldız": "olumsuz",
+      "3 yıldız": "olumsuz",
+      "4 yıldız": "olumsuz",
+      "5 yıldız": "nötr",
+      "6 yıldız": "nötr",
+      "7 yıldız": "olumlu",
+      "8 yıldız": "olumlu",
+      "9 yıldız": "olumlu",
+      "10 yıldız": "olumlu"
+    },
     "lvl_same_tol": 0.02,
     "lvl_min_margin": 0.1,
     "render_entity": false,
@@ -65,19 +87,26 @@
     "close_gap_lo": 0.35,
     "close_gap_hi": 0.6,
     "close_min_diff": 2,
-    "close_exclude_labels": [],
-    "close_ordinal_gap": 3,
+    "close_exclude_labels": [
+      "nötr"
+    ],
+    "close_ordinal_gap": 0,
+    "close_allowed_pairs_file": "experiments/separable_pairs/sinema_tr.json",
     "core_doc_records": [
-      1000,
       2500,
       4000,
-      8000
+      6000
     ],
     "core_docs_per_tier": 6,
-    "core_share": 0.15,
-    "core_pairs": [],
-    "core_pairs_per_doc": 2,
-    "core_seed_key": "",
+    "core_share": 0.35,
+    "core_pairs": [
+      [
+        "olumlu",
+        "olumsuz"
+      ]
+    ],
+    "core_pairs_per_doc": 1,
+    "core_seed_key": "sinema",
     "core_min_diff": 5,
     "entity_band_tol": 0.05,
     "label_band_tol": 0.15,
@@ -89,80 +118,79 @@
     "out_dir": "sinema_tr_out",
     "dataset": ""
   },
-  "source_hash_first1000": "7d7cc80e027320e9",
-  "rows_after_cleaning": 51171,
+  "source_hash_first1000": "f26f003975341f36",
+  "rows_after_cleaning": 44572,
   "cleaning": {
-    "rows_dropped_label_leakage": 141,
-    "label_leakage_rate": 0.002748,
+    "rows_dropped_text_regex": 6129,
+    "label_translation_applied": true,
+    "rows_dropped_label_leakage": 611,
+    "label_leakage_rate": 0.013523,
     "classes_dropped_low_support": [],
     "min_class_support": 1000
   },
   "ranking_feasibility": {
-    "n_classes": 10,
-    "min_class_pool": 1255,
-    "max_records": 12550,
+    "n_classes": 3,
+    "min_class_pool": 7556,
+    "max_records": 22668,
     "binding": true,
-    "ranking_length_ceiling_tokens": 1518524
+    "ranking_length_ceiling_tokens": 2868590
   },
   "tier_overlap": {
     "100000": {
       "n_haystacks": 10,
-      "mean_jaccard": 0.012,
-      "max_jaccard": 0.0475,
-      "pool_fraction_per_haystack": 0.0154
+      "mean_jaccard": 0.0128,
+      "max_jaccard": 0.0368,
+      "pool_fraction_per_haystack": 0.0186
     },
     "250000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.0227,
-      "max_jaccard": 0.0388,
-      "pool_fraction_per_haystack": 0.039
+      "mean_jaccard": 0.0275,
+      "max_jaccard": 0.1198,
+      "pool_fraction_per_haystack": 0.0478
     },
     "500000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.0647,
-      "max_jaccard": 0.1054,
-      "pool_fraction_per_haystack": 0.0777
-    },
-    "core1000": {
-      "n_haystacks": 6,
-      "mean_jaccard": 0.013,
-      "max_jaccard": 0.0199,
-      "pool_fraction_per_haystack": 0.0195
+      "mean_jaccard": 0.1129,
+      "max_jaccard": 0.3086,
+      "pool_fraction_per_haystack": 0.0955
     },
     "core2500": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0456,
-      "max_jaccard": 0.0725,
-      "pool_fraction_per_haystack": 0.0489
+      "mean_jaccard": 0.0424,
+      "max_jaccard": 0.046,
+      "pool_fraction_per_haystack": 0.0561
     },
     "core4000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0567,
-      "max_jaccard": 0.1151,
-      "pool_fraction_per_haystack": 0.0782
+      "mean_jaccard": 0.0666,
+      "max_jaccard": 0.0711,
+      "pool_fraction_per_haystack": 0.0897
     },
-    "core8000": {
+    "core6000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.1532,
-      "max_jaccard": 0.2245,
-      "pool_fraction_per_haystack": 0.1563
+      "mean_jaccard": 0.1047,
+      "max_jaccard": 0.1089,
+      "pool_fraction_per_haystack": 0.1346
     }
   },
   "max_pool_fraction": 0.35,
   "tiers_dropped_over_pool_cap": [],
   "tiers_built_over_pool_cap": [],
-  "label_space": 10,
+  "label_space": 3,
   "proportion_unit": "percent",
   "top_k_k": 3,
   "tokenizer": "Qwen/Qwen3-8B",
-  "questions_written": 290,
+  "questions_written": 212,
   "kind_distribution": {
-    "close_comparison": 50,
-    "count": 155,
-    "label_vs_label": 26,
-    "proportion": 59
+    "close_comparison": 18,
+    "count": 55,
+    "label_vs_label": 35,
+    "least_common": 16,
+    "most_common": 18,
+    "proportion": 53,
+    "second_most": 17
   },
-  "rare_count_questions": 27,
+  "rare_count_questions": 0,
   "rare_band": [
     5,
     30
@@ -170,692 +198,452 @@
   "haystacks": [
     {
       "haystack_id": "tr-100000-0",
-      "n_examples": 734,
-      "n_tokens": 99179,
-      "n_chars": 298281,
-      "drift_target": "1 yıldız",
+      "n_examples": 786,
+      "n_tokens": 99327,
+      "n_chars": 300988,
+      "drift_target": "olumlu",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 7
     },
     {
       "haystack_id": "tr-100000-1",
-      "n_examples": 808,
-      "n_tokens": 99378,
-      "n_chars": 297951,
-      "drift_target": "4 yıldız",
+      "n_examples": 793,
+      "n_tokens": 99256,
+      "n_chars": 299745,
+      "drift_target": "olumsuz",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 11
     },
     {
       "haystack_id": "tr-100000-2",
-      "n_examples": 809,
-      "n_tokens": 99352,
-      "n_chars": 297214,
-      "drift_target": "10 yıldız",
+      "n_examples": 801,
+      "n_tokens": 99298,
+      "n_chars": 299487,
+      "drift_target": "nötr",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 11
     },
     {
       "haystack_id": "tr-100000-3",
-      "n_examples": 725,
-      "n_tokens": 99212,
-      "n_chars": 295574,
-      "drift_target": "5 yıldız",
-      "drift_ok": true,
-      "n_questions": 12
+      "n_examples": 945,
+      "n_tokens": 99297,
+      "n_chars": 292937,
+      "drift_target": null,
+      "drift_ok": false,
+      "n_questions": 7
     },
     {
       "haystack_id": "tr-100000-4",
-      "n_examples": 748,
-      "n_tokens": 99340,
-      "n_chars": 297528,
-      "drift_target": "4 yıldız",
+      "n_examples": 819,
+      "n_tokens": 99322,
+      "n_chars": 298368,
+      "drift_target": "olumlu",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 11
     },
     {
       "haystack_id": "tr-100000-5",
-      "n_examples": 841,
-      "n_tokens": 99371,
-      "n_chars": 296905,
-      "drift_target": "4 yıldız",
+      "n_examples": 831,
+      "n_tokens": 99320,
+      "n_chars": 298062,
+      "drift_target": "olumsuz",
       "drift_ok": true,
-      "n_questions": 13
+      "n_questions": 11
     },
     {
       "haystack_id": "tr-100000-6",
-      "n_examples": 807,
-      "n_tokens": 99283,
-      "n_chars": 296520,
-      "drift_target": "5 yıldız",
+      "n_examples": 830,
+      "n_tokens": 99176,
+      "n_chars": 299961,
+      "drift_target": "olumsuz",
       "drift_ok": true,
-      "n_questions": 13
+      "n_questions": 9
     },
     {
       "haystack_id": "tr-100000-7",
-      "n_examples": 798,
+      "n_examples": 853,
       "n_tokens": 99322,
-      "n_chars": 293571,
-      "drift_target": "1 yıldız",
+      "n_chars": 297658,
+      "drift_target": "nötr",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 9
     },
     {
       "haystack_id": "tr-100000-8",
-      "n_examples": 825,
-      "n_tokens": 99321,
-      "n_chars": 295118,
-      "drift_target": "10 yıldız",
+      "n_examples": 815,
+      "n_tokens": 99264,
+      "n_chars": 299719,
+      "drift_target": "olumsuz",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 11
     },
     {
       "haystack_id": "tr-100000-9",
-      "n_examples": 787,
-      "n_tokens": 99379,
-      "n_chars": 297534,
-      "drift_target": "1 yıldız",
+      "n_examples": 824,
+      "n_tokens": 99312,
+      "n_chars": 296850,
+      "drift_target": "olumsuz",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 11
     },
     {
       "haystack_id": "tr-250000-0",
-      "n_examples": 2007,
-      "n_tokens": 248466,
-      "n_chars": 741652,
-      "drift_target": "10 yıldız",
-      "drift_ok": true,
-      "n_questions": 12
+      "n_examples": 2070,
+      "n_tokens": 248164,
+      "n_chars": 749114,
+      "drift_target": null,
+      "drift_ok": false,
+      "n_questions": 2
     },
     {
       "haystack_id": "tr-250000-1",
-      "n_examples": 1960,
-      "n_tokens": 248422,
-      "n_chars": 745956,
-      "drift_target": "5 yıldız",
+      "n_examples": 2014,
+      "n_tokens": 248347,
+      "n_chars": 750896,
+      "drift_target": "olumlu",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 7
     },
     {
       "haystack_id": "tr-250000-2",
-      "n_examples": 2078,
-      "n_tokens": 248489,
-      "n_chars": 742975,
-      "drift_target": "7 yıldız",
+      "n_examples": 2233,
+      "n_tokens": 247834,
+      "n_chars": 737649,
+      "drift_target": "olumlu",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 11
     },
     {
       "haystack_id": "tr-250000-3",
-      "n_examples": 1832,
-      "n_tokens": 248547,
-      "n_chars": 744754,
-      "drift_target": "2 yıldız",
+      "n_examples": 2271,
+      "n_tokens": 248281,
+      "n_chars": 736792,
+      "drift_target": "olumlu",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 11
     },
     {
       "haystack_id": "tr-250000-4",
-      "n_examples": 2093,
-      "n_tokens": 248540,
-      "n_chars": 744312,
-      "drift_target": "5 yıldız",
+      "n_examples": 2068,
+      "n_tokens": 248331,
+      "n_chars": 746493,
+      "drift_target": "olumlu",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 11
     },
     {
       "haystack_id": "tr-500000-0",
-      "n_examples": 4057,
-      "n_tokens": 497002,
-      "n_chars": 1490458,
-      "drift_target": "10 yıldız",
+      "n_examples": 4233,
+      "n_tokens": 496661,
+      "n_chars": 1492597,
+      "drift_target": "olumlu",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 10
     },
     {
       "haystack_id": "tr-500000-1",
-      "n_examples": 3902,
-      "n_tokens": 497147,
-      "n_chars": 1487836,
-      "drift_target": "2 yıldız",
+      "n_examples": 4182,
+      "n_tokens": 496596,
+      "n_chars": 1487284,
+      "drift_target": "nötr",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 11
     },
     {
       "haystack_id": "tr-500000-2",
-      "n_examples": 4056,
-      "n_tokens": 496912,
-      "n_chars": 1482626,
-      "drift_target": "3 yıldız",
+      "n_examples": 4012,
+      "n_tokens": 496698,
+      "n_chars": 1495504,
+      "drift_target": "nötr",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 11
     },
     {
       "haystack_id": "tr-500000-3",
-      "n_examples": 3968,
-      "n_tokens": 497150,
-      "n_chars": 1489147,
-      "drift_target": "2 yıldız",
+      "n_examples": 4454,
+      "n_tokens": 496559,
+      "n_chars": 1473699,
+      "drift_target": "nötr",
       "drift_ok": true,
-      "n_questions": 12
+      "n_questions": 11
     },
     {
       "haystack_id": "tr-500000-4",
-      "n_examples": 3905,
-      "n_tokens": 497135,
-      "n_chars": 1486574,
-      "drift_target": "3 yıldız",
+      "n_examples": 4409,
+      "n_tokens": 496597,
+      "n_chars": 1475891,
+      "drift_target": "olumlu",
       "drift_ok": true,
-      "n_questions": 12
-    },
-    {
-      "haystack_id": "tr-core1000-0",
-      "n_examples": 1000,
-      "n_tokens": 138663,
-      "n_chars": 417536,
-      "drift_target": null,
-      "drift_ok": false,
-      "n_questions": 2,
-      "core_document": true,
-      "label_counts": {
-        "1 yıldız": 2,
-        "10 yıldız": 150,
-        "3 yıldız": 144,
-        "4 yıldız": 145,
-        "5 yıldız": 112,
-        "6 yıldız": 54,
-        "7 yıldız": 238,
-        "8 yıldız": 5,
-        "9 yıldız": 150
-      }
-    },
-    {
-      "haystack_id": "tr-core1000-1",
-      "n_examples": 1000,
-      "n_tokens": 124477,
-      "n_chars": 370651,
-      "drift_target": null,
-      "drift_ok": false,
-      "n_questions": 2,
-      "core_document": true,
-      "label_counts": {
-        "1 yıldız": 150,
-        "10 yıldız": 335,
-        "2 yıldız": 24,
-        "5 yıldız": 150,
-        "6 yıldız": 143,
-        "8 yıldız": 145,
-        "9 yıldız": 53
-      }
-    },
-    {
-      "haystack_id": "tr-core1000-2",
-      "n_examples": 1000,
-      "n_tokens": 117214,
-      "n_chars": 349787,
-      "drift_target": null,
-      "drift_ok": false,
-      "n_questions": 2,
-      "core_document": true,
-      "label_counts": {
-        "1 yıldız": 144,
-        "10 yıldız": 274,
-        "2 yıldız": 144,
-        "3 yıldız": 3,
-        "5 yıldız": 40,
-        "6 yıldız": 150,
-        "7 yıldız": 150,
-        "8 yıldız": 9,
-        "9 yıldız": 86
-      }
-    },
-    {
-      "haystack_id": "tr-core1000-3",
-      "n_examples": 1000,
-      "n_tokens": 122592,
-      "n_chars": 365773,
-      "drift_target": null,
-      "drift_ok": false,
-      "n_questions": 2,
-      "core_document": true,
-      "label_counts": {
-        "1 yıldız": 144,
-        "10 yıldız": 19,
-        "2 yıldız": 131,
-        "4 yıldız": 150,
-        "5 yıldız": 143,
-        "6 yıldız": 150,
-        "7 yıldız": 120,
-        "9 yıldız": 143
-      }
-    },
-    {
-      "haystack_id": "tr-core1000-4",
-      "n_examples": 1000,
-      "n_tokens": 124647,
-      "n_chars": 373090,
-      "drift_target": null,
-      "drift_ok": false,
-      "n_questions": 2,
-      "core_document": true,
-      "label_counts": {
-        "1 yıldız": 23,
-        "10 yıldız": 150,
-        "2 yıldız": 145,
-        "3 yıldız": 144,
-        "4 yıldız": 1,
-        "5 yıldız": 7,
-        "6 yıldız": 173,
-        "7 yıldız": 196,
-        "8 yıldız": 11,
-        "9 yıldız": 150
-      }
-    },
-    {
-      "haystack_id": "tr-core1000-5",
-      "n_examples": 1000,
-      "n_tokens": 123525,
-      "n_chars": 370217,
-      "drift_target": null,
-      "drift_ok": false,
-      "n_questions": 2,
-      "core_document": true,
-      "label_counts": {
-        "1 yıldız": 150,
-        "10 yıldız": 144,
-        "2 yıldız": 10,
-        "3 yıldız": 81,
-        "4 yıldız": 150,
-        "5 yıldız": 99,
-        "6 yıldız": 218,
-        "7 yıldız": 3,
-        "8 yıldız": 1,
-        "9 yıldız": 144
-      }
+      "n_questions": 11
     },
     {
       "haystack_id": "tr-core2500-0",
       "n_examples": 2500,
-      "n_tokens": 305943,
-      "n_chars": 909344,
+      "n_tokens": 295199,
+      "n_chars": 885528,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 366,
-        "10 yıldız": 375,
-        "2 yıldız": 1,
-        "3 yıldız": 367,
-        "4 yıldız": 892,
-        "5 yıldız": 82,
-        "6 yıldız": 30,
-        "7 yıldız": 12,
-        "9 yıldız": 375
+        "nötr": 766,
+        "olumlu": 875,
+        "olumsuz": 859
       }
     },
     {
       "haystack_id": "tr-core2500-1",
       "n_examples": 2500,
-      "n_tokens": 314733,
-      "n_chars": 942145,
+      "n_tokens": 293075,
+      "n_chars": 880436,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 414,
-        "2 yıldız": 375,
-        "3 yıldız": 5,
-        "4 yıldız": 375,
-        "6 yıldız": 365,
-        "7 yıldız": 280,
-        "8 yıldız": 367,
-        "9 yıldız": 319
+        "nötr": 764,
+        "olumlu": 875,
+        "olumsuz": 861
       }
     },
     {
       "haystack_id": "tr-core2500-2",
       "n_examples": 2500,
-      "n_tokens": 307882,
-      "n_chars": 919853,
+      "n_tokens": 294377,
+      "n_chars": 885443,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 271,
-        "10 yıldız": 375,
-        "3 yıldız": 366,
-        "4 yıldız": 376,
-        "5 yıldız": 366,
-        "6 yıldız": 4,
-        "7 yıldız": 329,
-        "8 yıldız": 38,
-        "9 yıldız": 375
+        "nötr": 763,
+        "olumlu": 862,
+        "olumsuz": 875
       }
     },
     {
       "haystack_id": "tr-core2500-3",
       "n_examples": 2500,
-      "n_tokens": 311243,
-      "n_chars": 929974,
+      "n_tokens": 295705,
+      "n_chars": 888088,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 375,
-        "10 yıldız": 366,
-        "2 yıldız": 37,
-        "4 yıldız": 375,
-        "5 yıldız": 520,
-        "6 yıldız": 35,
-        "7 yıldız": 367,
-        "8 yıldız": 422,
-        "9 yıldız": 3
+        "nötr": 764,
+        "olumlu": 861,
+        "olumsuz": 875
       }
     },
     {
       "haystack_id": "tr-core2500-4",
       "n_examples": 2500,
-      "n_tokens": 299744,
-      "n_chars": 893610,
+      "n_tokens": 301557,
+      "n_chars": 904424,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 366,
-        "10 yıldız": 375,
-        "2 yıldız": 255,
-        "3 yıldız": 70,
-        "4 yıldız": 418,
-        "5 yıldız": 367,
-        "6 yıldız": 5,
-        "7 yıldız": 158,
-        "8 yıldız": 375,
-        "9 yıldız": 111
+        "nötr": 762,
+        "olumlu": 875,
+        "olumsuz": 863
       }
     },
     {
       "haystack_id": "tr-core2500-5",
       "n_examples": 2500,
-      "n_tokens": 304205,
-      "n_chars": 908746,
+      "n_tokens": 287360,
+      "n_chars": 862085,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 260,
-        "2 yıldız": 536,
-        "3 yıldız": 375,
-        "4 yıldız": 375,
-        "5 yıldız": 88,
-        "6 yıldız": 367,
-        "7 yıldız": 95,
-        "8 yıldız": 38,
-        "9 yıldız": 366
+        "nötr": 763,
+        "olumlu": 875,
+        "olumsuz": 862
       }
     },
     {
       "haystack_id": "tr-core4000-0",
       "n_examples": 4000,
-      "n_tokens": 493962,
-      "n_chars": 1478201,
+      "n_tokens": 470371,
+      "n_chars": 1409925,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 600,
-        "10 yıldız": 40,
-        "2 yıldız": 587,
-        "3 yıldız": 18,
-        "4 yıldız": 587,
-        "5 yıldız": 42,
-        "7 yıldız": 103,
-        "8 yıldız": 1423,
-        "9 yıldız": 600
+        "nötr": 1216,
+        "olumlu": 1384,
+        "olumsuz": 1400
       }
     },
     {
       "haystack_id": "tr-core4000-1",
       "n_examples": 4000,
-      "n_tokens": 512732,
-      "n_chars": 1540714,
+      "n_tokens": 475418,
+      "n_chars": 1430816,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 600,
-        "2 yıldız": 17,
-        "3 yıldız": 293,
-        "4 yıldız": 600,
-        "5 yıldız": 589,
-        "6 yıldız": 570,
-        "7 yıldız": 702,
-        "8 yıldız": 41,
-        "9 yıldız": 588
+        "nötr": 1218,
+        "olumlu": 1382,
+        "olumsuz": 1400
       }
     },
     {
       "haystack_id": "tr-core4000-2",
       "n_examples": 4000,
-      "n_tokens": 496802,
-      "n_chars": 1486862,
+      "n_tokens": 476473,
+      "n_chars": 1431369,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 590,
-        "10 yıldız": 2,
-        "2 yıldız": 192,
-        "3 yıldız": 1255,
-        "4 yıldız": 40,
-        "5 yıldız": 590,
-        "6 yıldız": 131,
-        "7 yıldız": 600,
-        "8 yıldız": 600
+        "nötr": 1218,
+        "olumlu": 1400,
+        "olumsuz": 1382
       }
     },
     {
       "haystack_id": "tr-core4000-3",
       "n_examples": 4000,
-      "n_tokens": 488026,
-      "n_chars": 1453478,
+      "n_tokens": 465277,
+      "n_chars": 1395623,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 941,
-        "10 yıldız": 600,
-        "2 yıldız": 39,
-        "3 yıldız": 600,
-        "4 yıldız": 602,
-        "5 yıldız": 20,
-        "6 yıldız": 588,
-        "7 yıldız": 18,
-        "8 yıldız": 588,
-        "9 yıldız": 4
+        "nötr": 1217,
+        "olumlu": 1400,
+        "olumsuz": 1383
       }
     },
     {
       "haystack_id": "tr-core4000-4",
       "n_examples": 4000,
-      "n_tokens": 488940,
-      "n_chars": 1462029,
+      "n_tokens": 482723,
+      "n_chars": 1450651,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 111,
-        "10 yıldız": 588,
-        "2 yıldız": 589,
-        "3 yıldız": 36,
-        "4 yıldız": 475,
-        "5 yıldız": 274,
-        "6 yıldız": 29,
-        "7 yıldız": 600,
-        "8 yıldız": 698,
-        "9 yıldız": 600
+        "nötr": 1218,
+        "olumlu": 1382,
+        "olumsuz": 1400
       }
     },
     {
       "haystack_id": "tr-core4000-5",
       "n_examples": 4000,
-      "n_tokens": 505574,
-      "n_chars": 1513441,
+      "n_tokens": 472558,
+      "n_chars": 1417379,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 56,
-        "10 yıldız": 1033,
-        "2 yıldız": 61,
-        "3 yıldız": 600,
-        "4 yıldız": 7,
-        "5 yıldız": 600,
-        "6 yıldız": 589,
-        "7 yıldız": 297,
-        "8 yıldız": 170,
-        "9 yıldız": 587
+        "nötr": 1218,
+        "olumlu": 1382,
+        "olumsuz": 1400
       }
     },
     {
-      "haystack_id": "tr-core8000-0",
-      "n_examples": 8000,
-      "n_tokens": 987203,
-      "n_chars": 2949838,
+      "haystack_id": "tr-core6000-0",
+      "n_examples": 6000,
+      "n_tokens": 715567,
+      "n_chars": 2151527,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 1409,
-        "10 yıldız": 574,
-        "2 yıldız": 8,
-        "3 yıldız": 891,
-        "4 yıldız": 1184,
-        "5 yıldız": 1183,
-        "6 yıldız": 274,
-        "7 yıldız": 77,
-        "8 yıldız": 1200,
-        "9 yıldız": 1200
+        "nötr": 1819,
+        "olumlu": 2100,
+        "olumsuz": 2081
       }
     },
     {
-      "haystack_id": "tr-core8000-1",
-      "n_examples": 8000,
-      "n_tokens": 988829,
-      "n_chars": 2961073,
+      "haystack_id": "tr-core6000-1",
+      "n_examples": 6000,
+      "n_tokens": 720375,
+      "n_chars": 2162185,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 727,
-        "10 yıldız": 1185,
-        "2 yıldız": 12,
-        "3 yıldız": 1200,
-        "4 yıldız": 275,
-        "5 yıldız": 1200,
-        "6 yıldız": 10,
-        "7 yıldız": 1901,
-        "8 yıldız": 307,
-        "9 yıldız": 1183
+        "nötr": 1818,
+        "olumlu": 2100,
+        "olumsuz": 2082
       }
     },
     {
-      "haystack_id": "tr-core8000-2",
-      "n_examples": 8000,
-      "n_tokens": 1000973,
-      "n_chars": 2992916,
+      "haystack_id": "tr-core6000-2",
+      "n_examples": 6000,
+      "n_tokens": 715857,
+      "n_chars": 2151549,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 528,
-        "10 yıldız": 1186,
-        "2 yıldız": 13,
-        "3 yıldız": 1255,
-        "4 yıldız": 1000,
-        "5 yıldız": 1183,
-        "6 yıldız": 384,
-        "7 yıldız": 1200,
-        "8 yıldız": 51,
-        "9 yıldız": 1200
+        "nötr": 1821,
+        "olumlu": 2079,
+        "olumsuz": 2100
       }
     },
     {
-      "haystack_id": "tr-core8000-3",
-      "n_examples": 8000,
-      "n_tokens": 999926,
-      "n_chars": 2999191,
+      "haystack_id": "tr-core6000-3",
+      "n_examples": 6000,
+      "n_tokens": 716458,
+      "n_chars": 2151927,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 1134,
-        "10 yıldız": 632,
-        "2 yıldız": 9,
-        "3 yıldız": 7,
-        "4 yıldız": 1200,
-        "5 yıldız": 869,
-        "6 yıldız": 1200,
-        "7 yıldız": 582,
-        "8 yıldız": 1184,
-        "9 yıldız": 1183
+        "nötr": 1820,
+        "olumlu": 2080,
+        "olumsuz": 2100
       }
     },
     {
-      "haystack_id": "tr-core8000-4",
-      "n_examples": 8000,
-      "n_tokens": 989095,
-      "n_chars": 2952561,
+      "haystack_id": "tr-core6000-4",
+      "n_examples": 6000,
+      "n_tokens": 716263,
+      "n_chars": 2148309,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 1182,
-        "10 yıldız": 1200,
-        "3 yıldız": 840,
-        "4 yıldız": 2071,
-        "5 yıldız": 1186,
-        "6 yıldız": 213,
-        "8 yıldız": 1200,
-        "9 yıldız": 108
+        "nötr": 1824,
+        "olumlu": 2100,
+        "olumsuz": 2076
       }
     },
     {
-      "haystack_id": "tr-core8000-5",
-      "n_examples": 8000,
-      "n_tokens": 979615,
-      "n_chars": 2924572,
+      "haystack_id": "tr-core6000-5",
+      "n_examples": 6000,
+      "n_tokens": 714435,
+      "n_chars": 2141724,
       "drift_target": null,
       "drift_ok": false,
-      "n_questions": 2,
+      "n_questions": 1,
       "core_document": true,
       "label_counts": {
-        "1 yıldız": 1200,
-        "10 yıldız": 114,
-        "2 yıldız": 1518,
-        "3 yıldız": 1200,
-        "4 yıldız": 493,
-        "5 yıldız": 1186,
-        "6 yıldız": 354,
-        "7 yıldız": 1185,
-        "8 yıldız": 130,
-        "9 yıldız": 620
+        "nötr": 1820,
+        "olumlu": 2100,
+        "olumsuz": 2080
       }
     }
   ]

tr_intent/manifest.json

@@ -1,6 +1,6 @@
 {
-  "version": "0.9.0",
-  "date": "2026-09-28",
+  "version": "0.10.0",
+  "date": "2026-10-01",
   "environment": {
     "python": "3.12.9",
     "polars": "1.42.0",
@@ -17,6 +17,12 @@
     "drop_label_leakage": true,
     "leak_label_words": false,
     "min_class_support": 0,
+    "boilerplate_keywords": [],
+    "boilerplate_min_keywords": 3,
+    "boilerplate_weak_min": 2,
+    "boilerplate_regex": "",
+    "drop_text_regex": [],
+    "strip_html": false,
     "seed": 42,
     "haystack_target_tokens": [
       50000,
@@ -62,12 +68,14 @@
     "close_min_diff": 2,
     "close_exclude_labels": [],
     "close_ordinal_gap": 0,
+    "close_allowed_pairs_file": "experiments/separable_pairs/intent_shared.json",
     "core_doc_records": [],
     "core_docs_per_tier": 0,
     "core_share": 0.3,
     "core_pairs": [],
     "core_pairs_per_doc": 1,
     "core_seed_key": "",
+    "core_min_diff": 5,
     "entity_band_tol": 0.05,
     "label_band_tol": 0.15,
     "entity_jitter_sigma": 1.0,
@@ -117,9 +125,9 @@
   "proportion_unit": "per_mille",
   "top_k_k": 3,
   "tokenizer": "Qwen/Qwen3-8B",
-  "questions_written": 151,
+  "questions_written": 139,
   "kind_distribution": {
-    "close_comparison": 31,
+    "close_comparison": 19,
     "count": 60,
     "label_vs_label": 10,
     "least_common": 10,
@@ -140,7 +148,7 @@
       "n_chars": 146514,
       "drift_target": "qa_currency",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 14
     },
     {
       "haystack_id": "tr-50000-1",
@@ -149,7 +157,7 @@
       "n_chars": 147593,
       "drift_target": "takeaway_order",
       "drift_ok": true,
-      "n_questions": 15
+      "n_questions": 14
     },
     {
       "haystack_id": "tr-50000-2",
@@ -167,7 +175,7 @@
       "n_chars": 147295,
       "drift_target": "transport_query",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 14
     },
     {
       "haystack_id": "tr-50000-4",
@@ -176,7 +184,7 @@
       "n_chars": 146030,
       "drift_target": "general_quirky",
       "drift_ok": true,
-      "n_questions": 14
+      "n_questions": 13
     },
     {
       "haystack_id": "tr-100000-0",
@@ -185,7 +193,7 @@
       "n_chars": 294767,
       "drift_target": "calendar_query",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 15
     },
     {
       "haystack_id": "tr-100000-1",
@@ -194,7 +202,7 @@
       "n_chars": 296555,
       "drift_target": "recommendation_events",
       "drift_ok": true,
-      "n_questions": 15
+      "n_questions": 13
     },
     {
       "haystack_id": "tr-100000-2",
@@ -203,7 +211,7 @@
       "n_chars": 292813,
       "drift_target": "datetime_query",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 14
     },
     {
       "haystack_id": "tr-100000-3",
@@ -221,7 +229,7 @@
       "n_chars": 291372,
       "drift_target": "calendar_query",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 15
     }
   ]
 }

tr_intent/questions.jsonl

@@ -10,10 +10,8 @@
 {"uid": "tr_intent:tr-50000-0-q9", "dataset": "tr_intent", "id": "tr-50000-0-q9", "haystack_uid": "tr_intent:tr-50000-0", "haystack_id": "tr-50000-0", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "weather_query", "answer": "19", "question": "Bu kayıtlarda kaç tane 'weather_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-50000-0-q10", "dataset": "tr_intent", "id": "tr-50000-0-q10", "haystack_uid": "tr_intent:tr-50000-0", "haystack_id": "tr-50000-0", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "social_post", "answer": "28", "question": "Bu kayıtlarda kaç tane 'social_post' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-50000-0-q11", "dataset": "tr_intent", "id": "tr-50000-0-q11", "haystack_uid": "tr_intent:tr-50000-0", "haystack_id": "tr-50000-0", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "iot_hue_lightchange", "answer": "24", "question": "Bu kayıtlarda kaç tane 'iot_hue_lightchange' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent:tr-50000-0-q12", "dataset": "tr_intent", "id": "tr-50000-0-q12", "haystack_uid": "tr_intent:tr-50000-0", "haystack_id": "tr-50000-0", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "music_query", "label_b": "email_query", "candidates": ["music_query", "email_query"], "answer": "email_query", "answer_key": "email_query", "question": "Bu kayıtlarda hangisi daha çok: 'music_query' etiketli kayıtlar mı, 'email_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-50000-0-q13", "dataset": "tr_intent", "id": "tr-50000-0-q13", "haystack_uid": "tr_intent:tr-50000-0", "haystack_id": "tr-50000-0", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_stock", "label_b": "lists_createoradd", "candidates": ["qa_stock", "lists_createoradd"], "answer": "qa_stock", "answer_key": "qa_stock", "question": "Bu kayıtlarda hangisi daha çok: 'qa_stock' etiketli kayıtlar mı, 'lists_createoradd' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-50000-0-q14", "dataset": "tr_intent", "id": "tr-50000-0-q14", "haystack_uid": "tr_intent:tr-50000-0", "haystack_id": "tr-50000-0", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "audio_volume_mute", "label_b": "qa_maths", "candidates": ["audio_volume_mute", "qa_maths"], "answer": "qa_maths", "answer_key": "qa_maths", "question": "Bu kayıtlarda hangisi daha çok: 'audio_volume_mute' etiketli kayıtlar mı, 'qa_maths' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-50000-0-q15", "dataset": "tr_intent", "id": "tr-50000-0-q15", "haystack_uid": "tr_intent:tr-50000-0", "haystack_id": "tr-50000-0", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "iot_coffee", "label_b": "iot_cleaning", "candidates": ["iot_coffee", "iot_cleaning"], "answer": "iot_coffee", "answer_key": "iot_coffee", "question": "Bu kayıtlarda hangisi daha çok: 'iot_coffee' etiketli kayıtlar mı, 'iot_cleaning' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-50000-0-q12", "dataset": "tr_intent", "id": "tr-50000-0-q12", "haystack_uid": "tr_intent:tr-50000-0", "haystack_id": "tr-50000-0", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_stock", "label_b": "lists_createoradd", "candidates": ["qa_stock", "lists_createoradd"], "answer": "qa_stock", "answer_key": "qa_stock", "question": "Bu kayıtlarda hangisi daha çok: 'qa_stock' etiketli kayıtlar mı, 'lists_createoradd' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-50000-0-q13", "dataset": "tr_intent", "id": "tr-50000-0-q13", "haystack_uid": "tr_intent:tr-50000-0", "haystack_id": "tr-50000-0", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "iot_coffee", "label_b": "play_podcasts", "candidates": ["iot_coffee", "play_podcasts"], "answer": "play_podcasts", "answer_key": "play_podcasts", "question": "Bu kayıtlarda hangisi daha çok: 'iot_coffee' etiketli kayıtlar mı, 'play_podcasts' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent:tr-50000-1-q0", "dataset": "tr_intent", "id": "tr-50000-1-q0", "haystack_uid": "tr_intent:tr-50000-1", "haystack_id": "tr-50000-1", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "iot_hue_lightup", "answer": "27", "question": "Bu kayıtlarda kaç tane 'iot_hue_lightup' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-50000-1-q1", "dataset": "tr_intent", "id": "tr-50000-1-q1", "haystack_uid": "tr_intent:tr-50000-1", "haystack_id": "tr-50000-1", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "iot_hue_lightoff", "answer": "69", "question": "Bu kayıtlarda kaç tane 'iot_hue_lightoff' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-50000-1-q2", "dataset": "tr_intent", "id": "tr-50000-1-q2", "haystack_uid": "tr_intent:tr-50000-1", "haystack_id": "tr-50000-1", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "transport_taxi", "answer": "36", "question": "Bu kayıtlarda kaç tane 'transport_taxi' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -26,9 +24,8 @@
 {"uid": "tr_intent:tr-50000-1-q9", "dataset": "tr_intent", "id": "tr-50000-1-q9", "haystack_uid": "tr_intent:tr-50000-1", "haystack_id": "tr-50000-1", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "iot_cleaning", "answer": "19", "question": "Bu kayıtlarda kaç tane 'iot_cleaning' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-50000-1-q10", "dataset": "tr_intent", "id": "tr-50000-1-q10", "haystack_uid": "tr_intent:tr-50000-1", "haystack_id": "tr-50000-1", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "play_audiobook", "answer": "26", "question": "Bu kayıtlarda kaç tane 'play_audiobook' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-50000-1-q11", "dataset": "tr_intent", "id": "tr-50000-1-q11", "haystack_uid": "tr_intent:tr-50000-1", "haystack_id": "tr-50000-1", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "qa_stock", "answer": "8", "question": "Bu kayıtlarda kaç tane 'qa_stock' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent:tr-50000-1-q12", "dataset": "tr_intent", "id": "tr-50000-1-q12", "haystack_uid": "tr_intent:tr-50000-1", "haystack_id": "tr-50000-1", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "news_query", "label_b": "music_query", "candidates": ["news_query", "music_query"], "answer": "news_query", "answer_key": "news_query", "question": "Bu kayıtlarda hangisi daha çok: 'news_query' etiketli kayıtlar mı, 'music_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-50000-1-q13", "dataset": "tr_intent", "id": "tr-50000-1-q13", "haystack_uid": "tr_intent:tr-50000-1", "haystack_id": "tr-50000-1", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "calendar_remove", "label_b": "takeaway_order", "candidates": ["calendar_remove", "takeaway_order"], "answer": "takeaway_order", "answer_key": "takeaway_order", "question": "Bu kayıtlarda hangisi daha çok: 'calendar_remove' etiketli kayıtlar mı, 'takeaway_order' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-50000-1-q14", "dataset": "tr_intent", "id": "tr-50000-1-q14", "haystack_uid": "tr_intent:tr-50000-1", "haystack_id": "tr-50000-1", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "play_podcasts", "label_b": "email_querycontact", "candidates": ["play_podcasts", "email_querycontact"], "answer": "play_podcasts", "answer_key": "play_podcasts", "question": "Bu kayıtlarda hangisi daha çok: 'play_podcasts' etiketli kayıtlar mı, 'email_querycontact' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-50000-1-q12", "dataset": "tr_intent", "id": "tr-50000-1-q12", "haystack_uid": "tr_intent:tr-50000-1", "haystack_id": "tr-50000-1", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "play_podcasts", "label_b": "takeaway_query", "candidates": ["play_podcasts", "takeaway_query"], "answer": "play_podcasts", "answer_key": "play_podcasts", "question": "Bu kayıtlarda hangisi daha çok: 'play_podcasts' etiketli kayıtlar mı, 'takeaway_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-50000-1-q13", "dataset": "tr_intent", "id": "tr-50000-1-q13", "haystack_uid": "tr_intent:tr-50000-1", "haystack_id": "tr-50000-1", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "takeaway_order", "label_b": "calendar_remove", "candidates": ["takeaway_order", "calendar_remove"], "answer": "takeaway_order", "answer_key": "takeaway_order", "question": "Bu kayıtlarda hangisi daha çok: 'takeaway_order' etiketli kayıtlar mı, 'calendar_remove' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent:tr-50000-2-q0", "dataset": "tr_intent", "id": "tr-50000-2-q0", "haystack_uid": "tr_intent:tr-50000-2", "haystack_id": "tr-50000-2", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "cooking_recipe", "answer": "26", "question": "Bu kayıtlarda kaç tane 'cooking_recipe' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-50000-2-q1", "dataset": "tr_intent", "id": "tr-50000-2-q1", "haystack_uid": "tr_intent:tr-50000-2", "haystack_id": "tr-50000-2", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "calendar_remove", "answer": "58", "question": "Bu kayıtlarda kaç tane 'calendar_remove' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-50000-2-q2", "dataset": "tr_intent", "id": "tr-50000-2-q2", "haystack_uid": "tr_intent:tr-50000-2", "haystack_id": "tr-50000-2", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "email_query", "answer": "38", "question": "Bu kayıtlarda kaç tane 'email_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -55,10 +52,8 @@
 {"uid": "tr_intent:tr-50000-3-q9", "dataset": "tr_intent", "id": "tr-50000-3-q9", "haystack_uid": "tr_intent:tr-50000-3", "haystack_id": "tr-50000-3", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "calendar_query", "answer": "28", "question": "Bu kayıtlarda kaç tane 'calendar_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-50000-3-q10", "dataset": "tr_intent", "id": "tr-50000-3-q10", "haystack_uid": "tr_intent:tr-50000-3", "haystack_id": "tr-50000-3", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "datetime_query", "answer": "23", "question": "Bu kayıtlarda kaç tane 'datetime_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-50000-3-q11", "dataset": "tr_intent", "id": "tr-50000-3-q11", "haystack_uid": "tr_intent:tr-50000-3", "haystack_id": "tr-50000-3", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "play_radio", "answer": "12", "question": "Bu kayıtlarda kaç tane 'play_radio' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent:tr-50000-3-q12", "dataset": "tr_intent", "id": "tr-50000-3-q12", "haystack_uid": "tr_intent:tr-50000-3", "haystack_id": "tr-50000-3", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "alarm_query", "label_b": "cooking_recipe", "candidates": ["alarm_query", "cooking_recipe"], "answer": "alarm_query", "answer_key": "alarm_query", "question": "Bu kayıtlarda hangisi daha çok: 'alarm_query' etiketli kayıtlar mı, 'cooking_recipe' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-50000-3-q13", "dataset": "tr_intent", "id": "tr-50000-3-q13", "haystack_uid": "tr_intent:tr-50000-3", "haystack_id": "tr-50000-3", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "iot_coffee", "label_b": "general_quirky", "candidates": ["iot_coffee", "general_quirky"], "answer": "general_quirky", "answer_key": "general_quirky", "question": "Bu kayıtlarda hangisi daha çok: 'iot_coffee' etiketli kayıtlar mı, 'general_quirky' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-50000-3-q14", "dataset": "tr_intent", "id": "tr-50000-3-q14", "haystack_uid": "tr_intent:tr-50000-3", "haystack_id": "tr-50000-3", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "alarm_remove", "label_b": "news_query", "candidates": ["alarm_remove", "news_query"], "answer": "alarm_remove", "answer_key": "alarm_remove", "question": "Bu kayıtlarda hangisi daha çok: 'alarm_remove' etiketli kayıtlar mı, 'news_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-50000-3-q15", "dataset": "tr_intent", "id": "tr-50000-3-q15", "haystack_uid": "tr_intent:tr-50000-3", "haystack_id": "tr-50000-3", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "iot_cleaning", "label_b": "weather_query", "candidates": ["iot_cleaning", "weather_query"], "answer": "iot_cleaning", "answer_key": "iot_cleaning", "question": "Bu kayıtlarda hangisi daha çok: 'iot_cleaning' etiketli kayıtlar mı, 'weather_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-50000-3-q12", "dataset": "tr_intent", "id": "tr-50000-3-q12", "haystack_uid": "tr_intent:tr-50000-3", "haystack_id": "tr-50000-3", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "qa_currency", "candidates": ["qa_factoid", "qa_currency"], "answer": "qa_factoid", "answer_key": "qa_factoid", "question": "Bu kayıtlarda hangisi daha çok: 'qa_factoid' etiketli kayıtlar mı, 'qa_currency' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-50000-3-q13", "dataset": "tr_intent", "id": "tr-50000-3-q13", "haystack_uid": "tr_intent:tr-50000-3", "haystack_id": "tr-50000-3", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "calendar_remove", "label_b": "iot_coffee", "candidates": ["calendar_remove", "iot_coffee"], "answer": "calendar_remove", "answer_key": "calendar_remove", "question": "Bu kayıtlarda hangisi daha çok: 'calendar_remove' etiketli kayıtlar mı, 'iot_coffee' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent:tr-50000-4-q0", "dataset": "tr_intent", "id": "tr-50000-4-q0", "haystack_uid": "tr_intent:tr-50000-4", "haystack_id": "tr-50000-4", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "transport_traffic", "answer": "23", "question": "Bu kayıtlarda kaç tane 'transport_traffic' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-50000-4-q1", "dataset": "tr_intent", "id": "tr-50000-4-q1", "haystack_uid": "tr_intent:tr-50000-4", "haystack_id": "tr-50000-4", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "play_music", "answer": "89", "question": "Bu kayıtlarda kaç tane 'play_music' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-50000-4-q2", "dataset": "tr_intent", "id": "tr-50000-4-q2", "haystack_uid": "tr_intent:tr-50000-4", "haystack_id": "tr-50000-4", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "label": "iot_hue_lightchange", "answer": "47", "question": "Bu kayıtlarda kaç tane 'iot_hue_lightchange' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -71,8 +66,7 @@
 {"uid": "tr_intent:tr-50000-4-q9", "dataset": "tr_intent", "id": "tr-50000-4-q9", "haystack_uid": "tr_intent:tr-50000-4", "haystack_id": "tr-50000-4", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "qa_currency", "answer": "28", "question": "Bu kayıtlarda kaç tane 'qa_currency' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-50000-4-q10", "dataset": "tr_intent", "id": "tr-50000-4-q10", "haystack_uid": "tr_intent:tr-50000-4", "haystack_id": "tr-50000-4", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "play_podcasts", "answer": "9", "question": "Bu kayıtlarda kaç tane 'play_podcasts' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-50000-4-q11", "dataset": "tr_intent", "id": "tr-50000-4-q11", "haystack_uid": "tr_intent:tr-50000-4", "haystack_id": "tr-50000-4", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "count", "rare": true, "label": "transport_ticket", "answer": "13", "question": "Bu kayıtlarda kaç tane 'transport_ticket' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent:tr-50000-4-q12", "dataset": "tr_intent", "id": "tr-50000-4-q12", "haystack_uid": "tr_intent:tr-50000-4", "haystack_id": "tr-50000-4", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "audio_volume_mute", "label_b": "recommendation_events", "candidates": ["audio_volume_mute", "recommendation_events"], "answer": "recommendation_events", "answer_key": "recommendation_events", "question": "Bu kayıtlarda hangisi daha çok: 'audio_volume_mute' etiketli kayıtlar mı, 'recommendation_events' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-50000-4-q13", "dataset": "tr_intent", "id": "tr-50000-4-q13", "haystack_uid": "tr_intent:tr-50000-4", "haystack_id": "tr-50000-4", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_definition", "label_b": "play_audiobook", "candidates": ["qa_definition", "play_audiobook"], "answer": "qa_definition", "answer_key": "qa_definition", "question": "Bu kayıtlarda hangisi daha çok: 'qa_definition' etiketli kayıtlar mı, 'play_audiobook' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-50000-4-q12", "dataset": "tr_intent", "id": "tr-50000-4-q12", "haystack_uid": "tr_intent:tr-50000-4", "haystack_id": "tr-50000-4", "language": "tr", "target_tokens": 50000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_definition", "label_b": "play_audiobook", "candidates": ["qa_definition", "play_audiobook"], "answer": "qa_definition", "answer_key": "qa_definition", "question": "Bu kayıtlarda hangisi daha çok: 'qa_definition' etiketli kayıtlar mı, 'play_audiobook' etiketli kayıtlar mı? Sadece etiket adın�� yaz.", "role": "core"}
 {"uid": "tr_intent:tr-100000-0-q0", "dataset": "tr_intent", "id": "tr-100000-0-q0", "haystack_uid": "tr_intent:tr-100000-0", "haystack_id": "tr-100000-0", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "iot_hue_lightoff", "answer": "83", "question": "Bu kayıtlarda kaç tane 'iot_hue_lightoff' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-100000-0-q1", "dataset": "tr_intent", "id": "tr-100000-0-q1", "haystack_uid": "tr_intent:tr-100000-0", "haystack_id": "tr-100000-0", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "weather_query", "answer": "52", "question": "Bu kayıtlarda kaç tane 'weather_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-100000-0-q2", "dataset": "tr_intent", "id": "tr-100000-0-q2", "haystack_uid": "tr_intent:tr-100000-0", "haystack_id": "tr-100000-0", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "iot_cleaning", "answer": "57", "question": "Bu kayıtlarda kaç tane 'iot_cleaning' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -85,10 +79,9 @@
 {"uid": "tr_intent:tr-100000-0-q9", "dataset": "tr_intent", "id": "tr-100000-0-q9", "haystack_uid": "tr_intent:tr-100000-0", "haystack_id": "tr-100000-0", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "recommendation_movies", "answer": "19", "question": "Bu kayıtlarda kaç tane 'recommendation_movies' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-100000-0-q10", "dataset": "tr_intent", "id": "tr-100000-0-q10", "haystack_uid": "tr_intent:tr-100000-0", "haystack_id": "tr-100000-0", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "iot_hue_lightup", "answer": "29", "question": "Bu kayıtlarda kaç tane 'iot_hue_lightup' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-100000-0-q11", "dataset": "tr_intent", "id": "tr-100000-0-q11", "haystack_uid": "tr_intent:tr-100000-0", "haystack_id": "tr-100000-0", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "general_joke", "answer": "7", "question": "Bu kayıtlarda kaç tane 'general_joke' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent:tr-100000-0-q12", "dataset": "tr_intent", "id": "tr-100000-0-q12", "haystack_uid": "tr_intent:tr-100000-0", "haystack_id": "tr-100000-0", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "play_audiobook", "label_b": "play_music", "candidates": ["play_audiobook", "play_music"], "answer": "play_music", "answer_key": "play_music", "question": "Bu kayıtlarda hangisi daha çok: 'play_audiobook' etiketli kayıtlar mı, 'play_music' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-100000-0-q13", "dataset": "tr_intent", "id": "tr-100000-0-q13", "haystack_uid": "tr_intent:tr-100000-0", "haystack_id": "tr-100000-0", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "music_likeness", "label_b": "music_query", "candidates": ["music_likeness", "music_query"], "answer": "music_query", "answer_key": "music_query", "question": "Bu kayıtlarda hangisi daha çok: 'music_likeness' etiketli kayıtlar mı, 'music_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-100000-0-q14", "dataset": "tr_intent", "id": "tr-100000-0-q14", "haystack_uid": "tr_intent:tr-100000-0", "haystack_id": "tr-100000-0", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "takeaway_order", "label_b": "audio_volume_mute", "candidates": ["takeaway_order", "audio_volume_mute"], "answer": "takeaway_order", "answer_key": "takeaway_order", "question": "Bu kayıtlarda hangisi daha çok: 'takeaway_order' etiketli kayıtlar mı, 'audio_volume_mute' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-100000-0-q15", "dataset": "tr_intent", "id": "tr-100000-0-q15", "haystack_uid": "tr_intent:tr-100000-0", "haystack_id": "tr-100000-0", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "play_radio", "label_b": "transport_query", "candidates": ["play_radio", "transport_query"], "answer": "play_radio", "answer_key": "play_radio", "question": "Bu kayıtlarda hangisi daha çok: 'play_radio' etiketli kayıtlar mı, 'transport_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-100000-0-q12", "dataset": "tr_intent", "id": "tr-100000-0-q12", "haystack_uid": "tr_intent:tr-100000-0", "haystack_id": "tr-100000-0", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "play_audiobook", "label_b": "play_radio", "candidates": ["play_audiobook", "play_radio"], "answer": "play_radio", "answer_key": "play_radio", "question": "Bu kayıtlarda hangisi daha çok: 'play_audiobook' etiketli kayıtlar mı, 'play_radio' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-100000-0-q13", "dataset": "tr_intent", "id": "tr-100000-0-q13", "haystack_uid": "tr_intent:tr-100000-0", "haystack_id": "tr-100000-0", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "play_music", "label_b": "transport_query", "candidates": ["play_music", "transport_query"], "answer": "play_music", "answer_key": "play_music", "question": "Bu kayıtlarda hangisi daha çok: 'play_music' etiketli kayıtlar mı, 'transport_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-100000-0-q14", "dataset": "tr_intent", "id": "tr-100000-0-q14", "haystack_uid": "tr_intent:tr-100000-0", "haystack_id": "tr-100000-0", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "takeaway_order", "label_b": "lists_remove", "candidates": ["takeaway_order", "lists_remove"], "answer": "takeaway_order", "answer_key": "takeaway_order", "question": "Bu kayıtlarda hangisi daha çok: 'takeaway_order' etiketli kayıtlar mı, 'lists_remove' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent:tr-100000-1-q0", "dataset": "tr_intent", "id": "tr-100000-1-q0", "haystack_uid": "tr_intent:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "calendar_query", "answer": "103", "question": "Bu kayıtlarda kaç tane 'calendar_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-100000-1-q1", "dataset": "tr_intent", "id": "tr-100000-1-q1", "haystack_uid": "tr_intent:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "alarm_query", "answer": "144", "question": "Bu kayıtlarda kaç tane 'alarm_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-100000-1-q2", "dataset": "tr_intent", "id": "tr-100000-1-q2", "haystack_uid": "tr_intent:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "email_sendemail", "answer": "415", "question": "Bu kayıtlarda kaç tane 'email_sendemail' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -101,9 +94,7 @@
 {"uid": "tr_intent:tr-100000-1-q9", "dataset": "tr_intent", "id": "tr-100000-1-q9", "haystack_uid": "tr_intent:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "audio_volume_mute", "answer": "8", "question": "Bu kayıtlarda kaç tane 'audio_volume_mute' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-100000-1-q10", "dataset": "tr_intent", "id": "tr-100000-1-q10", "haystack_uid": "tr_intent:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "play_game", "answer": "15", "question": "Bu kayıtlarda kaç tane 'play_game' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-100000-1-q11", "dataset": "tr_intent", "id": "tr-100000-1-q11", "haystack_uid": "tr_intent:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "weather_query", "answer": "25", "question": "Bu kayıtlarda kaç tane 'weather_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent:tr-100000-1-q12", "dataset": "tr_intent", "id": "tr-100000-1-q12", "haystack_uid": "tr_intent:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "recommendation_locations", "label_b": "qa_factoid", "candidates": ["recommendation_locations", "qa_factoid"], "answer": "qa_factoid", "answer_key": "qa_factoid", "question": "Bu kayıtlarda hangisi daha çok: 'recommendation_locations' etiketli kayıtlar mı, 'qa_factoid' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-100000-1-q13", "dataset": "tr_intent", "id": "tr-100000-1-q13", "haystack_uid": "tr_intent:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "music_query", "label_b": "general_quirky", "candidates": ["music_query", "general_quirky"], "answer": "music_query", "answer_key": "music_query", "question": "Bu kayıtlarda hangisi daha çok: 'music_query' etiketli kayıtlar mı, 'general_quirky' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-100000-1-q14", "dataset": "tr_intent", "id": "tr-100000-1-q14", "haystack_uid": "tr_intent:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "social_post", "label_b": "alarm_set", "candidates": ["social_post", "alarm_set"], "answer": "alarm_set", "answer_key": "alarm_set", "question": "Bu kayıtlarda hangisi daha çok: 'social_post' etiketli kayıtlar mı, 'alarm_set' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-100000-1-q12", "dataset": "tr_intent", "id": "tr-100000-1-q12", "haystack_uid": "tr_intent:tr-100000-1", "haystack_id": "tr-100000-1", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "social_post", "label_b": "qa_factoid", "candidates": ["social_post", "qa_factoid"], "answer": "qa_factoid", "answer_key": "qa_factoid", "question": "Bu kayıtlarda hangisi daha çok: 'social_post' etiketli kayıtlar mı, 'qa_factoid' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent:tr-100000-2-q0", "dataset": "tr_intent", "id": "tr-100000-2-q0", "haystack_uid": "tr_intent:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "email_sendemail", "answer": "351", "question": "Bu kayıtlarda kaç tane 'email_sendemail' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-100000-2-q1", "dataset": "tr_intent", "id": "tr-100000-2-q1", "haystack_uid": "tr_intent:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "lists_remove", "answer": "169", "question": "Bu kayıtlarda kaç tane 'lists_remove' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-100000-2-q2", "dataset": "tr_intent", "id": "tr-100000-2-q2", "haystack_uid": "tr_intent:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "calendar_set", "answer": "121", "question": "Bu kayıtlarda kaç tane 'calendar_set' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -116,10 +107,8 @@
 {"uid": "tr_intent:tr-100000-2-q9", "dataset": "tr_intent", "id": "tr-100000-2-q9", "haystack_uid": "tr_intent:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "audio_volume_up", "answer": "11", "question": "Bu kayıtlarda kaç tane 'audio_volume_up' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-100000-2-q10", "dataset": "tr_intent", "id": "tr-100000-2-q10", "haystack_uid": "tr_intent:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "cooking_recipe", "answer": "14", "question": "Bu kayıtlarda kaç tane 'cooking_recipe' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-100000-2-q11", "dataset": "tr_intent", "id": "tr-100000-2-q11", "haystack_uid": "tr_intent:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "general_quirky", "answer": "6", "question": "Bu kayıtlarda kaç tane 'general_quirky' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent:tr-100000-2-q12", "dataset": "tr_intent", "id": "tr-100000-2-q12", "haystack_uid": "tr_intent:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "recommendation_events", "label_b": "play_game", "candidates": ["recommendation_events", "play_game"], "answer": "recommendation_events", "answer_key": "recommendation_events", "question": "Bu kayıtlarda hangisi daha çok: 'recommendation_events' etiketli kayıtlar mı, 'play_game' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-100000-2-q13", "dataset": "tr_intent", "id": "tr-100000-2-q13", "haystack_uid": "tr_intent:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "lists_query", "label_b": "transport_ticket", "candidates": ["lists_query", "transport_ticket"], "answer": "lists_query", "answer_key": "lists_query", "question": "Bu kayıtlarda hangisi daha çok: 'lists_query' etiketli kayıtlar mı, 'transport_ticket' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-100000-2-q14", "dataset": "tr_intent", "id": "tr-100000-2-q14", "haystack_uid": "tr_intent:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "play_podcasts", "candidates": ["qa_factoid", "play_podcasts"], "answer": "qa_factoid", "answer_key": "qa_factoid", "question": "Bu kayıtlarda hangisi daha çok: 'qa_factoid' etiketli kayıtlar mı, 'play_podcasts' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-100000-2-q15", "dataset": "tr_intent", "id": "tr-100000-2-q15", "haystack_uid": "tr_intent:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_currency", "label_b": "qa_definition", "candidates": ["qa_currency", "qa_definition"], "answer": "qa_currency", "answer_key": "qa_currency", "question": "Bu kayıtlarda hangisi daha çok: 'qa_currency' etiketli kayıtlar mı, 'qa_definition' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-100000-2-q12", "dataset": "tr_intent", "id": "tr-100000-2-q12", "haystack_uid": "tr_intent:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_currency", "label_b": "qa_definition", "candidates": ["qa_currency", "qa_definition"], "answer": "qa_currency", "answer_key": "qa_currency", "question": "Bu kayıtlarda hangisi daha çok: 'qa_currency' etiketli kayıtlar mı, 'qa_definition' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-100000-2-q13", "dataset": "tr_intent", "id": "tr-100000-2-q13", "haystack_uid": "tr_intent:tr-100000-2", "haystack_id": "tr-100000-2", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "play_podcasts", "candidates": ["qa_factoid", "play_podcasts"], "answer": "qa_factoid", "answer_key": "qa_factoid", "question": "Bu kayıtlarda hangisi daha çok: 'qa_factoid' etiketli kayıtlar mı, 'play_podcasts' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent:tr-100000-3-q0", "dataset": "tr_intent", "id": "tr-100000-3-q0", "haystack_uid": "tr_intent:tr-100000-3", "haystack_id": "tr-100000-3", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "recommendation_movies", "answer": "66", "question": "Bu kayıtlarda kaç tane 'recommendation_movies' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-100000-3-q1", "dataset": "tr_intent", "id": "tr-100000-3-q1", "haystack_uid": "tr_intent:tr-100000-3", "haystack_id": "tr-100000-3", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "calendar_set", "answer": "585", "question": "Bu kayıtlarda kaç tane 'calendar_set' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent:tr-100000-3-q2", "dataset": "tr_intent", "id": "tr-100000-3-q2", "haystack_uid": "tr_intent:tr-100000-3", "haystack_id": "tr-100000-3", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "label": "qa_stock", "answer": "198", "question": "Bu kayıtlarda kaç tane 'qa_stock' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -145,7 +134,6 @@
 {"uid": "tr_intent:tr-100000-4-q9", "dataset": "tr_intent", "id": "tr-100000-4-q9", "haystack_uid": "tr_intent:tr-100000-4", "haystack_id": "tr-100000-4", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "email_query", "answer": "17", "question": "Bu kayıtlarda kaç tane 'email_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-100000-4-q10", "dataset": "tr_intent", "id": "tr-100000-4-q10", "haystack_uid": "tr_intent:tr-100000-4", "haystack_id": "tr-100000-4", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "email_querycontact", "answer": "23", "question": "Bu kayıtlarda kaç tane 'email_querycontact' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent:tr-100000-4-q11", "dataset": "tr_intent", "id": "tr-100000-4-q11", "haystack_uid": "tr_intent:tr-100000-4", "haystack_id": "tr-100000-4", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "count", "rare": true, "label": "email_sendemail", "answer": "25", "question": "Bu kayıtlarda kaç tane 'email_sendemail' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent:tr-100000-4-q12", "dataset": "tr_intent", "id": "tr-100000-4-q12", "haystack_uid": "tr_intent:tr-100000-4", "haystack_id": "tr-100000-4", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "recommendation_events", "label_b": "calendar_remove", "candidates": ["recommendation_events", "calendar_remove"], "answer": "calendar_remove", "answer_key": "calendar_remove", "question": "Bu kayıtlarda hangisi daha çok: 'recommendation_events' etiketli kayıtlar mı, 'calendar_remove' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-100000-4-q13", "dataset": "tr_intent", "id": "tr-100000-4-q13", "haystack_uid": "tr_intent:tr-100000-4", "haystack_id": "tr-100000-4", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "lists_createoradd", "label_b": "cooking_recipe", "candidates": ["lists_createoradd", "cooking_recipe"], "answer": "lists_createoradd", "answer_key": "lists_createoradd", "question": "Bu kayıtlarda hangisi daha çok: 'lists_createoradd' etiketli kayıtlar mı, 'cooking_recipe' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-100000-4-q14", "dataset": "tr_intent", "id": "tr-100000-4-q14", "haystack_uid": "tr_intent:tr-100000-4", "haystack_id": "tr-100000-4", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "iot_coffee", "label_b": "transport_taxi", "candidates": ["iot_coffee", "transport_taxi"], "answer": "iot_coffee", "answer_key": "iot_coffee", "question": "Bu kayıtlarda hangisi daha çok: 'iot_coffee' etiketli kayıtlar mı, 'transport_taxi' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent:tr-100000-4-q15", "dataset": "tr_intent", "id": "tr-100000-4-q15", "haystack_uid": "tr_intent:tr-100000-4", "haystack_id": "tr-100000-4", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "play_podcasts", "label_b": "play_audiobook", "candidates": ["play_podcasts", "play_audiobook"], "answer": "play_podcasts", "answer_key": "play_podcasts", "question": "Bu kayıtlarda hangisi daha çok: 'play_podcasts' etiketli kayıtlar mı, 'play_audiobook' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-100000-4-q12", "dataset": "tr_intent", "id": "tr-100000-4-q12", "haystack_uid": "tr_intent:tr-100000-4", "haystack_id": "tr-100000-4", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "transport_taxi", "label_b": "iot_coffee", "candidates": ["transport_taxi", "iot_coffee"], "answer": "iot_coffee", "answer_key": "iot_coffee", "question": "Bu kayıtlarda hangisi daha çok: 'transport_taxi' etiketli kayıtlar mı, 'iot_coffee' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-100000-4-q13", "dataset": "tr_intent", "id": "tr-100000-4-q13", "haystack_uid": "tr_intent:tr-100000-4", "haystack_id": "tr-100000-4", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "cooking_recipe", "label_b": "lists_createoradd", "candidates": ["cooking_recipe", "lists_createoradd"], "answer": "lists_createoradd", "answer_key": "lists_createoradd", "question": "Bu kayıtlarda hangisi daha çok: 'cooking_recipe' etiketli kayıtlar mı, 'lists_createoradd' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent:tr-100000-4-q14", "dataset": "tr_intent", "id": "tr-100000-4-q14", "haystack_uid": "tr_intent:tr-100000-4", "haystack_id": "tr-100000-4", "language": "tr", "target_tokens": 100000, "target_records": null, "kind": "close_comparison", "label": null, "label_a": "play_podcasts", "label_b": "calendar_remove", "candidates": ["play_podcasts", "calendar_remove"], "answer": "calendar_remove", "answer_key": "calendar_remove", "question": "Bu kayıtlarda hangisi daha çok: 'play_podcasts' etiketli kayıtlar mı, 'calendar_remove' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}

tr_intent_paired/manifest.json

@@ -1,6 +1,6 @@
 {
-  "version": "0.9.0",
-  "date": "2026-09-29",
+  "version": "0.10.0",
+  "date": "2026-10-01",
   "environment": {
     "python": "3.12.9",
     "polars": "1.42.0",
@@ -17,6 +17,12 @@
     "drop_label_leakage": true,
     "leak_label_words": false,
     "min_class_support": 0,
+    "boilerplate_keywords": [],
+    "boilerplate_min_keywords": 3,
+    "boilerplate_weak_min": 2,
+    "boilerplate_regex": "",
+    "drop_text_regex": [],
+    "strip_html": false,
     "seed": 42,
     "haystack_target_tokens": [
       50000,
@@ -65,6 +71,7 @@
     "close_min_diff": 2,
     "close_exclude_labels": [],
     "close_ordinal_gap": 0,
+    "close_allowed_pairs_file": "experiments/separable_pairs/intent_paired_shared.json",
     "core_doc_records": [
       3000
     ],
@@ -112,8 +119,8 @@
     },
     "core3000": {
       "n_haystacks": 10,
-      "mean_jaccard": 0.1514,
-      "max_jaccard": 0.1978,
+      "mean_jaccard": 0.1473,
+      "max_jaccard": 0.1795,
       "pool_fraction_per_haystack": 0.199
     }
   },
@@ -129,9 +136,9 @@
   "proportion_unit": "per_mille",
   "top_k_k": 3,
   "tokenizer": "Qwen/Qwen3-8B",
-  "questions_written": 175,
+  "questions_written": 161,
   "kind_distribution": {
-    "close_comparison": 55,
+    "close_comparison": 41,
     "count": 58,
     "label_vs_label": 11,
     "least_common": 10,
@@ -161,7 +168,7 @@
       "n_chars": 142996,
       "drift_target": "transport_taxi",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 14
     },
     {
       "haystack_id": "tr-r3000-2",
@@ -170,7 +177,7 @@
       "n_chars": 138288,
       "drift_target": "music_likeness",
       "drift_ok": true,
-      "n_questions": 14
+      "n_questions": 12
     },
     {
       "haystack_id": "tr-r3000-3",
@@ -179,7 +186,7 @@
       "n_chars": 139309,
       "drift_target": "iot_hue_lightoff",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 14
     },
     {
       "haystack_id": "tr-r3000-4",
@@ -188,7 +195,7 @@
       "n_chars": 141284,
       "drift_target": "email_querycontact",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 14
     },
     {
       "haystack_id": "tr-r6000-0",
@@ -197,7 +204,7 @@
       "n_chars": 287816,
       "drift_target": "lists_query",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 15
     },
     {
       "haystack_id": "tr-r6000-1",
@@ -206,7 +213,7 @@
       "n_chars": 289234,
       "drift_target": "social_post",
       "drift_ok": true,
-      "n_questions": 15
+      "n_questions": 14
     },
     {
       "haystack_id": "tr-r6000-2",
@@ -215,7 +222,7 @@
       "n_chars": 293894,
       "drift_target": "email_query",
       "drift_ok": true,
-      "n_questions": 15
+      "n_questions": 13
     },
     {
       "haystack_id": "tr-r6000-3",
@@ -224,7 +231,7 @@
       "n_chars": 291420,
       "drift_target": "transport_query",
       "drift_ok": true,
-      "n_questions": 16
+      "n_questions": 15
     },
     {
       "haystack_id": "tr-r6000-4",
@@ -233,13 +240,13 @@
       "n_chars": 278131,
       "drift_target": "lists_query",
       "drift_ok": true,
-      "n_questions": 15
+      "n_questions": 14
     },
     {
       "haystack_id": "tr-core3000-0",
       "n_examples": 3000,
-      "n_tokens": 47598,
-      "n_chars": 139242,
+      "n_tokens": 48786,
+      "n_chars": 144102,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -255,7 +262,7 @@
         "calendar_set": 31,
         "cooking_recipe": 20,
         "datetime_query": 31,
-        "email_query": 204,
+        "email_query": 210,
         "email_querycontact": 40,
         "email_sendemail": 29,
         "general_joke": 56,
@@ -266,25 +273,25 @@
         "iot_hue_lightoff": 68,
         "iot_hue_lightup": 66,
         "lists_createoradd": 202,
-        "lists_query": 210,
-        "lists_remove": 40,
-        "music_likeness": 45,
-        "music_query": 97,
-        "news_query": 64,
-        "play_audiobook": 49,
-        "play_game": 56,
-        "play_music": 61,
-        "play_podcasts": 31,
-        "play_radio": 77,
-        "qa_currency": 26,
-        "qa_definition": 45,
+        "lists_query": 40,
+        "lists_remove": 45,
+        "music_likeness": 97,
+        "music_query": 64,
+        "news_query": 49,
+        "play_audiobook": 56,
+        "play_game": 61,
+        "play_music": 31,
+        "play_podcasts": 77,
+        "play_radio": 26,
+        "qa_currency": 45,
+        "qa_definition": 43,
         "qa_factoid": 43,
-        "qa_maths": 43,
-        "qa_stock": 41,
-        "recommendation_events": 66,
-        "recommendation_locations": 24,
-        "recommendation_movies": 63,
-        "social_post": 53,
+        "qa_maths": 41,
+        "qa_stock": 66,
+        "recommendation_events": 24,
+        "recommendation_locations": 63,
+        "recommendation_movies": 53,
+        "social_post": 204,
         "social_query": 26,
         "takeaway_order": 57,
         "takeaway_query": 53,
@@ -298,8 +305,8 @@
     {
       "haystack_id": "tr-core3000-1",
       "n_examples": 3000,
-      "n_tokens": 48591,
-      "n_chars": 143615,
+      "n_tokens": 49491,
+      "n_chars": 146021,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -310,27 +317,27 @@
         "alarm_set": 110,
         "audio_volume_mute": 56,
         "audio_volume_up": 45,
-        "calendar_query": 210,
-        "calendar_remove": 28,
+        "calendar_query": 203,
+        "calendar_remove": 210,
         "calendar_set": 203,
-        "cooking_recipe": 61,
-        "datetime_query": 42,
-        "email_query": 17,
-        "email_querycontact": 47,
-        "email_sendemail": 16,
-        "general_joke": 33,
-        "general_quirky": 46,
-        "iot_cleaning": 85,
-        "iot_coffee": 11,
-        "iot_hue_lightchange": 45,
-        "iot_hue_lightoff": 36,
-        "iot_hue_lightup": 86,
-        "lists_createoradd": 34,
-        "lists_query": 210,
+        "cooking_recipe": 28,
+        "datetime_query": 61,
+        "email_query": 42,
+        "email_querycontact": 17,
+        "email_sendemail": 47,
+        "general_joke": 16,
+        "general_quirky": 33,
+        "iot_cleaning": 46,
+        "iot_coffee": 85,
+        "iot_hue_lightchange": 11,
+        "iot_hue_lightoff": 45,
+        "iot_hue_lightup": 36,
+        "lists_createoradd": 86,
+        "lists_query": 34,
         "lists_remove": 42,
         "music_likeness": 40,
         "music_query": 88,
-        "news_query": 203,
+        "news_query": 210,
         "play_audiobook": 21,
         "play_game": 32,
         "play_music": 77,
@@ -358,8 +365,8 @@
     {
       "haystack_id": "tr-core3000-2",
       "n_examples": 3000,
-      "n_tokens": 48049,
-      "n_chars": 140804,
+      "n_tokens": 49673,
+      "n_chars": 145619,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -372,21 +379,21 @@
         "audio_volume_up": 73,
         "calendar_query": 89,
         "calendar_remove": 64,
-        "calendar_set": 37,
-        "cooking_recipe": 11,
-        "datetime_query": 39,
-        "email_query": 34,
-        "email_querycontact": 49,
-        "email_sendemail": 56,
-        "general_joke": 69,
-        "general_quirky": 56,
-        "iot_cleaning": 17,
-        "iot_coffee": 74,
-        "iot_hue_lightchange": 69,
-        "iot_hue_lightoff": 83,
-        "iot_hue_lightup": 41,
-        "lists_createoradd": 210,
-        "lists_query": 203,
+        "calendar_set": 210,
+        "cooking_recipe": 37,
+        "datetime_query": 11,
+        "email_query": 39,
+        "email_querycontact": 34,
+        "email_sendemail": 49,
+        "general_joke": 56,
+        "general_quirky": 69,
+        "iot_cleaning": 56,
+        "iot_coffee": 17,
+        "iot_hue_lightchange": 74,
+        "iot_hue_lightoff": 69,
+        "iot_hue_lightup": 83,
+        "lists_createoradd": 204,
+        "lists_query": 41,
         "lists_remove": 30,
         "music_likeness": 42,
         "music_query": 66,
@@ -395,14 +402,14 @@
         "play_game": 59,
         "play_music": 58,
         "play_podcasts": 15,
-        "play_radio": 58,
-        "qa_currency": 21,
-        "qa_definition": 72,
-        "qa_factoid": 48,
-        "qa_maths": 61,
-        "qa_stock": 132,
-        "recommendation_events": 41,
-        "recommendation_locations": 204,
+        "play_radio": 203,
+        "qa_currency": 58,
+        "qa_definition": 21,
+        "qa_factoid": 72,
+        "qa_maths": 48,
+        "qa_stock": 61,
+        "recommendation_events": 132,
+        "recommendation_locations": 41,
         "recommendation_movies": 22,
         "social_post": 21,
         "social_query": 19,
@@ -418,8 +425,8 @@
     {
       "haystack_id": "tr-core3000-3",
       "n_examples": 3000,
-      "n_tokens": 47913,
-      "n_chars": 139294,
+      "n_tokens": 48026,
+      "n_chars": 140186,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -430,38 +437,38 @@
         "alarm_set": 32,
         "audio_volume_mute": 40,
         "audio_volume_up": 51,
-        "calendar_query": 55,
-        "calendar_remove": 69,
-        "calendar_set": 36,
-        "cooking_recipe": 46,
+        "calendar_query": 203,
+        "calendar_remove": 55,
+        "calendar_set": 69,
+        "cooking_recipe": 210,
         "datetime_query": 204,
-        "email_query": 35,
-        "email_querycontact": 64,
-        "email_sendemail": 49,
-        "general_joke": 60,
-        "general_quirky": 33,
-        "iot_cleaning": 53,
-        "iot_coffee": 43,
-        "iot_hue_lightchange": 81,
-        "iot_hue_lightoff": 84,
-        "iot_hue_lightup": 34,
-        "lists_createoradd": 117,
-        "lists_query": 210,
-        "lists_remove": 24,
-        "music_likeness": 31,
-        "music_query": 74,
-        "news_query": 54,
-        "play_audiobook": 40,
-        "play_game": 35,
-        "play_music": 30,
-        "play_podcasts": 67,
-        "play_radio": 203,
-        "qa_currency": 37,
-        "qa_definition": 51,
-        "qa_factoid": 14,
-        "qa_maths": 115,
-        "qa_stock": 45,
-        "recommendation_events": 210,
+        "email_query": 36,
+        "email_querycontact": 46,
+        "email_sendemail": 35,
+        "general_joke": 64,
+        "general_quirky": 49,
+        "iot_cleaning": 60,
+        "iot_coffee": 33,
+        "iot_hue_lightchange": 53,
+        "iot_hue_lightoff": 43,
+        "iot_hue_lightup": 81,
+        "lists_createoradd": 84,
+        "lists_query": 34,
+        "lists_remove": 103,
+        "music_likeness": 24,
+        "music_query": 31,
+        "news_query": 74,
+        "play_audiobook": 54,
+        "play_game": 40,
+        "play_music": 35,
+        "play_podcasts": 30,
+        "play_radio": 210,
+        "qa_currency": 67,
+        "qa_definition": 37,
+        "qa_factoid": 51,
+        "qa_maths": 14,
+        "qa_stock": 129,
+        "recommendation_events": 45,
         "recommendation_locations": 49,
         "recommendation_movies": 34,
         "social_post": 34,
@@ -598,8 +605,8 @@
     {
       "haystack_id": "tr-core3000-6",
       "n_examples": 3000,
-      "n_tokens": 47699,
-      "n_chars": 139046,
+      "n_tokens": 48066,
+      "n_chars": 140965,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 2,
@@ -619,25 +626,25 @@
         "email_querycontact": 37,
         "email_sendemail": 20,
         "general_joke": 49,
-        "general_quirky": 210,
-        "iot_cleaning": 45,
-        "iot_coffee": 66,
-        "iot_hue_lightchange": 48,
-        "iot_hue_lightoff": 33,
-        "iot_hue_lightup": 34,
-        "lists_createoradd": 21,
-        "lists_query": 210,
-        "lists_remove": 16,
-        "music_likeness": 90,
-        "music_query": 29,
-        "news_query": 33,
+        "general_quirky": 45,
+        "iot_cleaning": 66,
+        "iot_coffee": 48,
+        "iot_hue_lightchange": 33,
+        "iot_hue_lightoff": 34,
+        "iot_hue_lightup": 21,
+        "lists_createoradd": 203,
+        "lists_query": 16,
+        "lists_remove": 90,
+        "music_likeness": 29,
+        "music_query": 33,
+        "news_query": 210,
         "play_audiobook": 52,
         "play_game": 28,
         "play_music": 39,
-        "play_podcasts": 203,
+        "play_podcasts": 204,
         "play_radio": 61,
         "qa_currency": 24,
-        "qa_definition": 204,
+        "qa_definition": 210,
         "qa_factoid": 26,
         "qa_maths": 115,
         "qa_stock": 35,

tr_intent_paired/questions.jsonl

@@ -11,9 +11,9 @@
 {"uid": "tr_intent_paired:tr-r3000-0-q10", "dataset": "tr_intent_paired", "id": "tr-r3000-0-q10", "haystack_uid": "tr_intent_paired:tr-r3000-0", "haystack_id": "tr-r3000-0", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "lists_remove", "answer": "7", "question": "Bu kayıtlarda kaç tane 'lists_remove' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r3000-0-q11", "dataset": "tr_intent_paired", "id": "tr-r3000-0-q11", "haystack_uid": "tr_intent_paired:tr-r3000-0", "haystack_id": "tr-r3000-0", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "general_quirky", "answer": "21", "question": "Bu kayıtlarda kaç tane 'general_quirky' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r3000-0-q12", "dataset": "tr_intent_paired", "id": "tr-r3000-0-q12", "haystack_uid": "tr_intent_paired:tr-r3000-0", "haystack_id": "tr-r3000-0", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "qa_stock", "label_b": "iot_hue_lightchange", "candidates": ["qa_stock", "iot_hue_lightchange"], "answer": "qa_stock", "answer_key": "qa_stock", "question": "Bu kayıtlarda hangisi daha çok: 'qa_stock' etiketli kayıtlar mı, 'iot_hue_lightchange' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r3000-0-q13", "dataset": "tr_intent_paired", "id": "tr-r3000-0-q13", "haystack_uid": "tr_intent_paired:tr-r3000-0", "haystack_id": "tr-r3000-0", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "iot_coffee", "label_b": "takeaway_query", "candidates": ["iot_coffee", "takeaway_query"], "answer": "takeaway_query", "answer_key": "takeaway_query", "question": "Bu kayıtlarda hangisi daha çok: 'iot_coffee' etiketli kayıtlar mı, 'takeaway_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r3000-0-q14", "dataset": "tr_intent_paired", "id": "tr-r3000-0-q14", "haystack_uid": "tr_intent_paired:tr-r3000-0", "haystack_id": "tr-r3000-0", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_query", "label_b": "cooking_recipe", "candidates": ["calendar_query", "cooking_recipe"], "answer": "calendar_query", "answer_key": "calendar_query", "question": "Bu kayıtlarda hangisi daha çok: 'calendar_query' etiketli kayıtlar mı, 'cooking_recipe' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r3000-0-q15", "dataset": "tr_intent_paired", "id": "tr-r3000-0-q15", "haystack_uid": "tr_intent_paired:tr-r3000-0", "haystack_id": "tr-r3000-0", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "social_post", "label_b": "iot_hue_lightoff", "candidates": ["social_post", "iot_hue_lightoff"], "answer": "social_post", "answer_key": "social_post", "question": "Bu kayıtlarda hangisi daha çok: 'social_post' etiketli kayıtlar mı, 'iot_hue_lightoff' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r3000-0-q13", "dataset": "tr_intent_paired", "id": "tr-r3000-0-q13", "haystack_uid": "tr_intent_paired:tr-r3000-0", "haystack_id": "tr-r3000-0", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "social_post", "label_b": "iot_hue_lightoff", "candidates": ["social_post", "iot_hue_lightoff"], "answer": "social_post", "answer_key": "social_post", "question": "Bu kayıtlarda hangisi daha çok: 'social_post' etiketli kayıtlar mı, 'iot_hue_lightoff' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r3000-0-q14", "dataset": "tr_intent_paired", "id": "tr-r3000-0-q14", "haystack_uid": "tr_intent_paired:tr-r3000-0", "haystack_id": "tr-r3000-0", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "takeaway_query", "label_b": "iot_coffee", "candidates": ["takeaway_query", "iot_coffee"], "answer": "takeaway_query", "answer_key": "takeaway_query", "question": "Bu kayıtlarda hangisi daha çok: 'takeaway_query' etiketli kayıtlar mı, 'iot_coffee' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r3000-0-q15", "dataset": "tr_intent_paired", "id": "tr-r3000-0-q15", "haystack_uid": "tr_intent_paired:tr-r3000-0", "haystack_id": "tr-r3000-0", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "cooking_recipe", "label_b": "takeaway_order", "candidates": ["cooking_recipe", "takeaway_order"], "answer": "cooking_recipe", "answer_key": "cooking_recipe", "question": "Bu kayıtlarda hangisi daha çok: 'cooking_recipe' etiketli kayıtlar mı, 'takeaway_order' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent_paired:tr-r3000-1-q0", "dataset": "tr_intent_paired", "id": "tr-r3000-1-q0", "haystack_uid": "tr_intent_paired:tr-r3000-1", "haystack_id": "tr-r3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "calendar_query", "answer": "27", "question": "Bu kayıtlarda kaç tane 'calendar_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r3000-1-q1", "dataset": "tr_intent_paired", "id": "tr-r3000-1-q1", "haystack_uid": "tr_intent_paired:tr-r3000-1", "haystack_id": "tr-r3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "social_query", "answer": "138", "question": "Bu kayıtlarda kaç tane 'social_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r3000-1-q2", "dataset": "tr_intent_paired", "id": "tr-r3000-1-q2", "haystack_uid": "tr_intent_paired:tr-r3000-1", "haystack_id": "tr-r3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "cooking_recipe", "answer": "35", "question": "Bu kayıtlarda kaç tane 'cooking_recipe' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -26,10 +26,8 @@
 {"uid": "tr_intent_paired:tr-r3000-1-q9", "dataset": "tr_intent_paired", "id": "tr-r3000-1-q9", "haystack_uid": "tr_intent_paired:tr-r3000-1", "haystack_id": "tr-r3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "iot_coffee", "answer": "14", "question": "Bu kayıtlarda kaç tane 'iot_coffee' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r3000-1-q10", "dataset": "tr_intent_paired", "id": "tr-r3000-1-q10", "haystack_uid": "tr_intent_paired:tr-r3000-1", "haystack_id": "tr-r3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "recommendation_locations", "answer": "5", "question": "Bu kayıtlarda kaç tane 'recommendation_locations' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r3000-1-q11", "dataset": "tr_intent_paired", "id": "tr-r3000-1-q11", "haystack_uid": "tr_intent_paired:tr-r3000-1", "haystack_id": "tr-r3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "calendar_remove", "answer": "30", "question": "Bu kayıtlarda kaç tane 'calendar_remove' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent_paired:tr-r3000-1-q12", "dataset": "tr_intent_paired", "id": "tr-r3000-1-q12", "haystack_uid": "tr_intent_paired:tr-r3000-1", "haystack_id": "tr-r3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "datetime_query", "label_b": "social_post", "candidates": ["datetime_query", "social_post"], "answer": "social_post", "answer_key": "social_post", "question": "Bu kayıtlarda hangisi daha çok: 'datetime_query' etiketli kayıtlar mı, 'social_post' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r3000-1-q13", "dataset": "tr_intent_paired", "id": "tr-r3000-1-q13", "haystack_uid": "tr_intent_paired:tr-r3000-1", "haystack_id": "tr-r3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "transport_ticket", "label_b": "weather_query", "candidates": ["transport_ticket", "weather_query"], "answer": "transport_ticket", "answer_key": "transport_ticket", "question": "Bu kayıtlarda hangisi daha çok: 'transport_ticket' etiketli kayıtlar mı, 'weather_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r3000-1-q14", "dataset": "tr_intent_paired", "id": "tr-r3000-1-q14", "haystack_uid": "tr_intent_paired:tr-r3000-1", "haystack_id": "tr-r3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "email_sendemail", "label_b": "music_likeness", "candidates": ["email_sendemail", "music_likeness"], "answer": "music_likeness", "answer_key": "music_likeness", "question": "Bu kayıtlarda hangisi daha çok: 'email_sendemail' etiketli kayıtlar mı, 'music_likeness' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r3000-1-q15", "dataset": "tr_intent_paired", "id": "tr-r3000-1-q15", "haystack_uid": "tr_intent_paired:tr-r3000-1", "haystack_id": "tr-r3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "transport_traffic", "label_b": "transport_query", "candidates": ["transport_traffic", "transport_query"], "answer": "transport_traffic", "answer_key": "transport_traffic", "question": "Bu kayıtlarda hangisi daha çok: 'transport_traffic' etiketli kayıtlar mı, 'transport_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r3000-1-q12", "dataset": "tr_intent_paired", "id": "tr-r3000-1-q12", "haystack_uid": "tr_intent_paired:tr-r3000-1", "haystack_id": "tr-r3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "social_post", "label_b": "datetime_query", "candidates": ["social_post", "datetime_query"], "answer": "social_post", "answer_key": "social_post", "question": "Bu kayıtlarda hangisi daha çok: 'social_post' etiketli kayıtlar mı, 'datetime_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r3000-1-q13", "dataset": "tr_intent_paired", "id": "tr-r3000-1-q13", "haystack_uid": "tr_intent_paired:tr-r3000-1", "haystack_id": "tr-r3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "email_query", "label_b": "weather_query", "candidates": ["email_query", "weather_query"], "answer": "email_query", "answer_key": "email_query", "question": "Bu kayıtlarda hangisi daha çok: 'email_query' etiketli kayıtlar mı, 'weather_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent_paired:tr-r3000-2-q0", "dataset": "tr_intent_paired", "id": "tr-r3000-2-q0", "haystack_uid": "tr_intent_paired:tr-r3000-2", "haystack_id": "tr-r3000-2", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "qa_definition", "answer": "195", "question": "Bu kayıtlarda kaç tane 'qa_definition' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r3000-2-q1", "dataset": "tr_intent_paired", "id": "tr-r3000-2-q1", "haystack_uid": "tr_intent_paired:tr-r3000-2", "haystack_id": "tr-r3000-2", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "iot_hue_lightchange", "answer": "36", "question": "Bu kayıtlarda kaç tane 'iot_hue_lightchange' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r3000-2-q2", "dataset": "tr_intent_paired", "id": "tr-r3000-2-q2", "haystack_uid": "tr_intent_paired:tr-r3000-2", "haystack_id": "tr-r3000-2", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "email_query", "answer": "16", "question": "Bu kayıtlarda kaç tane 'email_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -42,8 +40,6 @@
 {"uid": "tr_intent_paired:tr-r3000-2-q9", "dataset": "tr_intent_paired", "id": "tr-r3000-2-q9", "haystack_uid": "tr_intent_paired:tr-r3000-2", "haystack_id": "tr-r3000-2", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "play_audiobook", "answer": "13", "question": "Bu kayıtlarda kaç tane 'play_audiobook' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r3000-2-q10", "dataset": "tr_intent_paired", "id": "tr-r3000-2-q10", "haystack_uid": "tr_intent_paired:tr-r3000-2", "haystack_id": "tr-r3000-2", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "play_radio", "answer": "10", "question": "Bu kayıtlarda kaç tane 'play_radio' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r3000-2-q11", "dataset": "tr_intent_paired", "id": "tr-r3000-2-q11", "haystack_uid": "tr_intent_paired:tr-r3000-2", "haystack_id": "tr-r3000-2", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "transport_taxi", "answer": "28", "question": "Bu kayıtlarda kaç tane 'transport_taxi' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent_paired:tr-r3000-2-q12", "dataset": "tr_intent_paired", "id": "tr-r3000-2-q12", "haystack_uid": "tr_intent_paired:tr-r3000-2", "haystack_id": "tr-r3000-2", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "music_query", "label_b": "qa_stock", "candidates": ["music_query", "qa_stock"], "answer": "qa_stock", "answer_key": "qa_stock", "question": "Bu kayıtlarda hangisi daha çok: 'music_query' etiketli kayıtlar mı, 'qa_stock' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r3000-2-q13", "dataset": "tr_intent_paired", "id": "tr-r3000-2-q13", "haystack_uid": "tr_intent_paired:tr-r3000-2", "haystack_id": "tr-r3000-2", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "news_query", "label_b": "qa_maths", "candidates": ["news_query", "qa_maths"], "answer": "qa_maths", "answer_key": "qa_maths", "question": "Bu kayıtlarda hangisi daha çok: 'news_query' etiketli kayıtlar mı, 'qa_maths' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent_paired:tr-r3000-3-q0", "dataset": "tr_intent_paired", "id": "tr-r3000-3-q0", "haystack_uid": "tr_intent_paired:tr-r3000-3", "haystack_id": "tr-r3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "iot_hue_lightup", "answer": "96", "question": "Bu kayıtlarda kaç tane 'iot_hue_lightup' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r3000-3-q1", "dataset": "tr_intent_paired", "id": "tr-r3000-3-q1", "haystack_uid": "tr_intent_paired:tr-r3000-3", "haystack_id": "tr-r3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "music_query", "answer": "15", "question": "Bu kayıtlarda kaç tane 'music_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r3000-3-q2", "dataset": "tr_intent_paired", "id": "tr-r3000-3-q2", "haystack_uid": "tr_intent_paired:tr-r3000-3", "haystack_id": "tr-r3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "transport_taxi", "answer": "89", "question": "Bu kayıtlarda kaç tane 'transport_taxi' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -56,10 +52,8 @@
 {"uid": "tr_intent_paired:tr-r3000-3-q9", "dataset": "tr_intent_paired", "id": "tr-r3000-3-q9", "haystack_uid": "tr_intent_paired:tr-r3000-3", "haystack_id": "tr-r3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "takeaway_query", "answer": "18", "question": "Bu kayıtlarda kaç tane 'takeaway_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r3000-3-q10", "dataset": "tr_intent_paired", "id": "tr-r3000-3-q10", "haystack_uid": "tr_intent_paired:tr-r3000-3", "haystack_id": "tr-r3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "play_radio", "answer": "15", "question": "Bu kayıtlarda kaç tane 'play_radio' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r3000-3-q11", "dataset": "tr_intent_paired", "id": "tr-r3000-3-q11", "haystack_uid": "tr_intent_paired:tr-r3000-3", "haystack_id": "tr-r3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "audio_volume_up", "answer": "24", "question": "Bu kayıtlarda kaç tane 'audio_volume_up' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent_paired:tr-r3000-3-q12", "dataset": "tr_intent_paired", "id": "tr-r3000-3-q12", "haystack_uid": "tr_intent_paired:tr-r3000-3", "haystack_id": "tr-r3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "general_joke", "label_b": "qa_maths", "candidates": ["general_joke", "qa_maths"], "answer": "qa_maths", "answer_key": "qa_maths", "question": "Bu kayıtlarda hangisi daha çok: 'general_joke' etiketli kayıtlar mı, 'qa_maths' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r3000-3-q13", "dataset": "tr_intent_paired", "id": "tr-r3000-3-q13", "haystack_uid": "tr_intent_paired:tr-r3000-3", "haystack_id": "tr-r3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "alarm_remove", "label_b": "iot_coffee", "candidates": ["alarm_remove", "iot_coffee"], "answer": "iot_coffee", "answer_key": "iot_coffee", "question": "Bu kayıtlarda hangisi daha çok: 'alarm_remove' etiketli kayıtlar mı, 'iot_coffee' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r3000-3-q14", "dataset": "tr_intent_paired", "id": "tr-r3000-3-q14", "haystack_uid": "tr_intent_paired:tr-r3000-3", "haystack_id": "tr-r3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_createoradd", "label_b": "iot_hue_lightchange", "candidates": ["lists_createoradd", "iot_hue_lightchange"], "answer": "lists_createoradd", "answer_key": "lists_createoradd", "question": "Bu kayıtlarda hangisi daha çok: 'lists_createoradd' etiketli kayıtlar mı, 'iot_hue_lightchange' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r3000-3-q15", "dataset": "tr_intent_paired", "id": "tr-r3000-3-q15", "haystack_uid": "tr_intent_paired:tr-r3000-3", "haystack_id": "tr-r3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "weather_query", "label_b": "general_quirky", "candidates": ["weather_query", "general_quirky"], "answer": "weather_query", "answer_key": "weather_query", "question": "Bu kayıtlarda hangisi daha çok: 'weather_query' etiketli kayıtlar mı, 'general_quirky' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r3000-3-q12", "dataset": "tr_intent_paired", "id": "tr-r3000-3-q12", "haystack_uid": "tr_intent_paired:tr-r3000-3", "haystack_id": "tr-r3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "iot_coffee", "label_b": "qa_factoid", "candidates": ["iot_coffee", "qa_factoid"], "answer": "qa_factoid", "answer_key": "qa_factoid", "question": "Bu kayıtlarda hangisi daha çok: 'iot_coffee' etiketli kayıtlar mı, 'qa_factoid' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r3000-3-q13", "dataset": "tr_intent_paired", "id": "tr-r3000-3-q13", "haystack_uid": "tr_intent_paired:tr-r3000-3", "haystack_id": "tr-r3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_createoradd", "label_b": "iot_hue_lightchange", "candidates": ["lists_createoradd", "iot_hue_lightchange"], "answer": "lists_createoradd", "answer_key": "lists_createoradd", "question": "Bu kayıtlarda hangisi daha çok: 'lists_createoradd' etiketli kayıtlar mı, 'iot_hue_lightchange' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent_paired:tr-r3000-4-q0", "dataset": "tr_intent_paired", "id": "tr-r3000-4-q0", "haystack_uid": "tr_intent_paired:tr-r3000-4", "haystack_id": "tr-r3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "iot_coffee", "answer": "60", "question": "Bu kayıtlarda kaç tane 'iot_coffee' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r3000-4-q1", "dataset": "tr_intent_paired", "id": "tr-r3000-4-q1", "haystack_uid": "tr_intent_paired:tr-r3000-4", "haystack_id": "tr-r3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "lists_createoradd", "answer": "33", "question": "Bu kayıtlarda kaç tane 'lists_createoradd' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r3000-4-q2", "dataset": "tr_intent_paired", "id": "tr-r3000-4-q2", "haystack_uid": "tr_intent_paired:tr-r3000-4", "haystack_id": "tr-r3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "label": "iot_hue_lightoff", "answer": "55", "question": "Bu kayıtlarda kaç tane 'iot_hue_lightoff' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -72,10 +66,8 @@
 {"uid": "tr_intent_paired:tr-r3000-4-q9", "dataset": "tr_intent_paired", "id": "tr-r3000-4-q9", "haystack_uid": "tr_intent_paired:tr-r3000-4", "haystack_id": "tr-r3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "calendar_remove", "answer": "6", "question": "Bu kayıtlarda kaç tane 'calendar_remove' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r3000-4-q10", "dataset": "tr_intent_paired", "id": "tr-r3000-4-q10", "haystack_uid": "tr_intent_paired:tr-r3000-4", "haystack_id": "tr-r3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "social_query", "answer": "15", "question": "Bu kayıtlarda kaç tane 'social_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r3000-4-q11", "dataset": "tr_intent_paired", "id": "tr-r3000-4-q11", "haystack_uid": "tr_intent_paired:tr-r3000-4", "haystack_id": "tr-r3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "count", "rare": true, "label": "lists_query", "answer": "19", "question": "Bu kayıtlarda kaç tane 'lists_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent_paired:tr-r3000-4-q12", "dataset": "tr_intent_paired", "id": "tr-r3000-4-q12", "haystack_uid": "tr_intent_paired:tr-r3000-4", "haystack_id": "tr-r3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "iot_coffee", "candidates": ["qa_factoid", "iot_coffee"], "answer": "qa_factoid", "answer_key": "qa_factoid", "question": "Bu kayıtlarda hangisi daha çok: 'qa_factoid' etiketli kayıtlar mı, 'iot_coffee' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r3000-4-q13", "dataset": "tr_intent_paired", "id": "tr-r3000-4-q13", "haystack_uid": "tr_intent_paired:tr-r3000-4", "haystack_id": "tr-r3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "email_querycontact", "label_b": "play_radio", "candidates": ["email_querycontact", "play_radio"], "answer": "play_radio", "answer_key": "play_radio", "question": "Bu kayıtlarda hangisi daha çok: 'email_querycontact' etiketli kayıtlar mı, 'play_radio' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r3000-4-q14", "dataset": "tr_intent_paired", "id": "tr-r3000-4-q14", "haystack_uid": "tr_intent_paired:tr-r3000-4", "haystack_id": "tr-r3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "cooking_recipe", "label_b": "play_audiobook", "candidates": ["cooking_recipe", "play_audiobook"], "answer": "cooking_recipe", "answer_key": "cooking_recipe", "question": "Bu kayıtlarda hangisi daha çok: 'cooking_recipe' etiketli kayıtlar mı, 'play_audiobook' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r3000-4-q15", "dataset": "tr_intent_paired", "id": "tr-r3000-4-q15", "haystack_uid": "tr_intent_paired:tr-r3000-4", "haystack_id": "tr-r3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "play_music", "label_b": "recommendation_movies", "candidates": ["play_music", "recommendation_movies"], "answer": "play_music", "answer_key": "play_music", "question": "Bu kayıtlarda hangisi daha çok: 'play_music' etiketli kayıtlar mı, 'recommendation_movies' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r3000-4-q12", "dataset": "tr_intent_paired", "id": "tr-r3000-4-q12", "haystack_uid": "tr_intent_paired:tr-r3000-4", "haystack_id": "tr-r3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "play_audiobook", "candidates": ["qa_factoid", "play_audiobook"], "answer": "qa_factoid", "answer_key": "qa_factoid", "question": "Bu kayıtlarda hangisi daha çok: 'qa_factoid' etiketli kayıtlar mı, 'play_audiobook' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r3000-4-q13", "dataset": "tr_intent_paired", "id": "tr-r3000-4-q13", "haystack_uid": "tr_intent_paired:tr-r3000-4", "haystack_id": "tr-r3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "cooking_recipe", "label_b": "iot_coffee", "candidates": ["cooking_recipe", "iot_coffee"], "answer": "cooking_recipe", "answer_key": "cooking_recipe", "question": "Bu kayıtlarda hangisi daha çok: 'cooking_recipe' etiketli kayıtlar mı, 'iot_coffee' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent_paired:tr-r6000-0-q0", "dataset": "tr_intent_paired", "id": "tr-r6000-0-q0", "haystack_uid": "tr_intent_paired:tr-r6000-0", "haystack_id": "tr-r6000-0", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "iot_cleaning", "answer": "108", "question": "Bu kayıtlarda kaç tane 'iot_cleaning' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r6000-0-q1", "dataset": "tr_intent_paired", "id": "tr-r6000-0-q1", "haystack_uid": "tr_intent_paired:tr-r6000-0", "haystack_id": "tr-r6000-0", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "takeaway_query", "answer": "14", "question": "Bu kayıtlarda kaç tane 'takeaway_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r6000-0-q2", "dataset": "tr_intent_paired", "id": "tr-r6000-0-q2", "haystack_uid": "tr_intent_paired:tr-r6000-0", "haystack_id": "tr-r6000-0", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "play_podcasts", "answer": "74", "question": "Bu kayıtlarda kaç tane 'play_podcasts' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -88,10 +80,9 @@
 {"uid": "tr_intent_paired:tr-r6000-0-q9", "dataset": "tr_intent_paired", "id": "tr-r6000-0-q9", "haystack_uid": "tr_intent_paired:tr-r6000-0", "haystack_id": "tr-r6000-0", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "play_radio", "answer": "20", "question": "Bu kayıtlarda kaç tane 'play_radio' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r6000-0-q10", "dataset": "tr_intent_paired", "id": "tr-r6000-0-q10", "haystack_uid": "tr_intent_paired:tr-r6000-0", "haystack_id": "tr-r6000-0", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "email_sendemail", "answer": "213", "question": "Bu kayıtlarda kaç tane 'email_sendemail' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r6000-0-q11", "dataset": "tr_intent_paired", "id": "tr-r6000-0-q11", "haystack_uid": "tr_intent_paired:tr-r6000-0", "haystack_id": "tr-r6000-0", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "recommendation_movies", "answer": "98", "question": "Bu kayıtlarda kaç tane 'recommendation_movies' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
-{"uid": "tr_intent_paired:tr-r6000-0-q12", "dataset": "tr_intent_paired", "id": "tr-r6000-0-q12", "haystack_uid": "tr_intent_paired:tr-r6000-0", "haystack_id": "tr-r6000-0", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "social_query", "label_b": "qa_currency", "candidates": ["social_query", "qa_currency"], "answer": "qa_currency", "answer_key": "qa_currency", "question": "Bu kayıtlarda hangisi daha çok: 'social_query' etiketli kayıtlar mı, 'qa_currency' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r6000-0-q13", "dataset": "tr_intent_paired", "id": "tr-r6000-0-q13", "haystack_uid": "tr_intent_paired:tr-r6000-0", "haystack_id": "tr-r6000-0", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "music_query", "label_b": "iot_hue_lightoff", "candidates": ["music_query", "iot_hue_lightoff"], "answer": "iot_hue_lightoff", "answer_key": "iot_hue_lightoff", "question": "Bu kayıtlarda hangisi daha çok: 'music_query' etiketli kayıtlar mı, 'iot_hue_lightoff' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r6000-0-q14", "dataset": "tr_intent_paired", "id": "tr-r6000-0-q14", "haystack_uid": "tr_intent_paired:tr-r6000-0", "haystack_id": "tr-r6000-0", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "recommendation_events", "label_b": "alarm_set", "candidates": ["recommendation_events", "alarm_set"], "answer": "recommendation_events", "answer_key": "recommendation_events", "question": "Bu kayıtlarda hangisi daha çok: 'recommendation_events' etiketli kayıtlar mı, 'alarm_set' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r6000-0-q15", "dataset": "tr_intent_paired", "id": "tr-r6000-0-q15", "haystack_uid": "tr_intent_paired:tr-r6000-0", "haystack_id": "tr-r6000-0", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "qa_definition", "label_b": "music_likeness", "candidates": ["qa_definition", "music_likeness"], "answer": "qa_definition", "answer_key": "qa_definition", "question": "Bu kayıtlarda hangisi daha çok: 'qa_definition' etiketli kayıtlar mı, 'music_likeness' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r6000-0-q12", "dataset": "tr_intent_paired", "id": "tr-r6000-0-q12", "haystack_uid": "tr_intent_paired:tr-r6000-0", "haystack_id": "tr-r6000-0", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "calendar_set", "label_b": "transport_query", "candidates": ["calendar_set", "transport_query"], "answer": "transport_query", "answer_key": "transport_query", "question": "Bu kayıtlarda hangisi daha çok: 'calendar_set' etiketli kayıtlar mı, 'transport_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r6000-0-q13", "dataset": "tr_intent_paired", "id": "tr-r6000-0-q13", "haystack_uid": "tr_intent_paired:tr-r6000-0", "haystack_id": "tr-r6000-0", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "weather_query", "candidates": ["qa_factoid", "weather_query"], "answer": "weather_query", "answer_key": "weather_query", "question": "Bu kayıtlarda hangisi daha çok: 'qa_factoid' etiketli kayıtlar mı, 'weather_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r6000-0-q14", "dataset": "tr_intent_paired", "id": "tr-r6000-0-q14", "haystack_uid": "tr_intent_paired:tr-r6000-0", "haystack_id": "tr-r6000-0", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "email_query", "label_b": "qa_currency", "candidates": ["email_query", "qa_currency"], "answer": "email_query", "answer_key": "email_query", "question": "Bu kayıtlarda hangisi daha çok: 'email_query' etiketli kayıtlar mı, 'qa_currency' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent_paired:tr-r6000-1-q0", "dataset": "tr_intent_paired", "id": "tr-r6000-1-q0", "haystack_uid": "tr_intent_paired:tr-r6000-1", "haystack_id": "tr-r6000-1", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "audio_volume_mute", "answer": "104", "question": "Bu kayıtlarda kaç tane 'audio_volume_mute' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r6000-1-q1", "dataset": "tr_intent_paired", "id": "tr-r6000-1-q1", "haystack_uid": "tr_intent_paired:tr-r6000-1", "haystack_id": "tr-r6000-1", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "audio_volume_up", "answer": "112", "question": "Bu kayıtlarda kaç tane 'audio_volume_up' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r6000-1-q2", "dataset": "tr_intent_paired", "id": "tr-r6000-1-q2", "haystack_uid": "tr_intent_paired:tr-r6000-1", "haystack_id": "tr-r6000-1", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "iot_coffee", "answer": "148", "question": "Bu kayıtlarda kaç tane 'iot_coffee' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -104,9 +95,8 @@
 {"uid": "tr_intent_paired:tr-r6000-1-q9", "dataset": "tr_intent_paired", "id": "tr-r6000-1-q9", "haystack_uid": "tr_intent_paired:tr-r6000-1", "haystack_id": "tr-r6000-1", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "takeaway_query", "answer": "21", "question": "Bu kayıtlarda kaç tane 'takeaway_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r6000-1-q10", "dataset": "tr_intent_paired", "id": "tr-r6000-1-q10", "haystack_uid": "tr_intent_paired:tr-r6000-1", "haystack_id": "tr-r6000-1", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "takeaway_order", "answer": "9", "question": "Bu kayıtlarda kaç tane 'takeaway_order' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r6000-1-q11", "dataset": "tr_intent_paired", "id": "tr-r6000-1-q11", "haystack_uid": "tr_intent_paired:tr-r6000-1", "haystack_id": "tr-r6000-1", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "news_query", "answer": "13", "question": "Bu kayıtlarda kaç tane 'news_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent_paired:tr-r6000-1-q12", "dataset": "tr_intent_paired", "id": "tr-r6000-1-q12", "haystack_uid": "tr_intent_paired:tr-r6000-1", "haystack_id": "tr-r6000-1", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "iot_hue_lightoff", "label_b": "weather_query", "candidates": ["iot_hue_lightoff", "weather_query"], "answer": "iot_hue_lightoff", "answer_key": "iot_hue_lightoff", "question": "Bu kayıtlarda hangisi daha çok: 'iot_hue_lightoff' etiketli kayıtlar mı, 'weather_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r6000-1-q13", "dataset": "tr_intent_paired", "id": "tr-r6000-1-q13", "haystack_uid": "tr_intent_paired:tr-r6000-1", "haystack_id": "tr-r6000-1", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "alarm_query", "label_b": "music_likeness", "candidates": ["alarm_query", "music_likeness"], "answer": "alarm_query", "answer_key": "alarm_query", "question": "Bu kayıtlarda hangisi daha çok: 'alarm_query' etiketli kayıtlar mı, 'music_likeness' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r6000-1-q14", "dataset": "tr_intent_paired", "id": "tr-r6000-1-q14", "haystack_uid": "tr_intent_paired:tr-r6000-1", "haystack_id": "tr-r6000-1", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "recommendation_locations", "label_b": "email_query", "candidates": ["recommendation_locations", "email_query"], "answer": "email_query", "answer_key": "email_query", "question": "Bu kayıtlarda hangisi daha çok: 'recommendation_locations' etiketli kayıtlar mı, 'email_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r6000-1-q12", "dataset": "tr_intent_paired", "id": "tr-r6000-1-q12", "haystack_uid": "tr_intent_paired:tr-r6000-1", "haystack_id": "tr-r6000-1", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "email_query", "label_b": "transport_ticket", "candidates": ["email_query", "transport_ticket"], "answer": "transport_ticket", "answer_key": "transport_ticket", "question": "Bu kayıtlarda hangisi daha çok: 'email_query' etiketli kayıtlar mı, 'transport_ticket' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r6000-1-q13", "dataset": "tr_intent_paired", "id": "tr-r6000-1-q13", "haystack_uid": "tr_intent_paired:tr-r6000-1", "haystack_id": "tr-r6000-1", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "weather_query", "label_b": "iot_hue_lightoff", "candidates": ["weather_query", "iot_hue_lightoff"], "answer": "iot_hue_lightoff", "answer_key": "iot_hue_lightoff", "question": "Bu kayıtlarda hangisi daha çok: 'weather_query' etiketli kayıtlar mı, 'iot_hue_lightoff' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent_paired:tr-r6000-2-q0", "dataset": "tr_intent_paired", "id": "tr-r6000-2-q0", "haystack_uid": "tr_intent_paired:tr-r6000-2", "haystack_id": "tr-r6000-2", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "email_querycontact", "answer": "134", "question": "Bu kayıtlarda kaç tane 'email_querycontact' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r6000-2-q1", "dataset": "tr_intent_paired", "id": "tr-r6000-2-q1", "haystack_uid": "tr_intent_paired:tr-r6000-2", "haystack_id": "tr-r6000-2", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "qa_stock", "answer": "92", "question": "Bu kayıtlarda kaç tane 'qa_stock' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r6000-2-q2", "dataset": "tr_intent_paired", "id": "tr-r6000-2-q2", "haystack_uid": "tr_intent_paired:tr-r6000-2", "haystack_id": "tr-r6000-2", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "takeaway_order", "answer": "100", "question": "Bu kayıtlarda kaç tane 'takeaway_order' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -119,9 +109,7 @@
 {"uid": "tr_intent_paired:tr-r6000-2-q9", "dataset": "tr_intent_paired", "id": "tr-r6000-2-q9", "haystack_uid": "tr_intent_paired:tr-r6000-2", "haystack_id": "tr-r6000-2", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "second_most", "label": null, "candidates": ["alarm_set", "music_query", "play_audiobook", "qa_currency", "qa_stock"], "answer": "qa_stock", "question": "Bu kayıtlarda ikinci en sık görülen etiket hangisi? Etiketler: 'alarm_set', 'music_query', 'play_audiobook', 'qa_currency', 'qa_stock'. Sadece etiket adını yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r6000-2-q10", "dataset": "tr_intent_paired", "id": "tr-r6000-2-q10", "haystack_uid": "tr_intent_paired:tr-r6000-2", "haystack_id": "tr-r6000-2", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "label_vs_label", "label": null, "label_a": "qa_currency", "label_b": "audio_volume_mute", "candidates": ["qa_currency", "audio_volume_mute"], "answer": "daha az", "answer_key": "less", "question": "'qa_currency' etiketli kayıtlar 'audio_volume_mute' etiketli kayıtlardan daha mı çok, daha mı az, yoksa eşit sayıda mı? 'daha çok', 'daha az' veya 'eşit' yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r6000-2-q11", "dataset": "tr_intent_paired", "id": "tr-r6000-2-q11", "haystack_uid": "tr_intent_paired:tr-r6000-2", "haystack_id": "tr-r6000-2", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "label_vs_label", "label": null, "label_a": "qa_definition", "label_b": "lists_createoradd", "candidates": ["qa_definition", "lists_createoradd"], "answer": "daha az", "answer_key": "less", "question": "'qa_definition' etiketli kayıtlar 'lists_createoradd' etiketli kayıtlardan daha mı çok, daha mı az, yoksa eşit sayıda mı? 'daha çok', 'daha az' veya 'eşit' yaz.", "role": "control"}
-{"uid": "tr_intent_paired:tr-r6000-2-q12", "dataset": "tr_intent_paired", "id": "tr-r6000-2-q12", "haystack_uid": "tr_intent_paired:tr-r6000-2", "haystack_id": "tr-r6000-2", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "social_query", "label_b": "datetime_query", "candidates": ["social_query", "datetime_query"], "answer": "datetime_query", "answer_key": "datetime_query", "question": "Bu kayıtlarda hangisi daha çok: 'social_query' etiketli kayıtlar mı, 'datetime_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r6000-2-q13", "dataset": "tr_intent_paired", "id": "tr-r6000-2-q13", "haystack_uid": "tr_intent_paired:tr-r6000-2", "haystack_id": "tr-r6000-2", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "news_query", "label_b": "audio_volume_mute", "candidates": ["news_query", "audio_volume_mute"], "answer": "audio_volume_mute", "answer_key": "audio_volume_mute", "question": "Bu kayıtlarda hangisi daha çok: 'news_query' etiketli kayıtlar mı, 'audio_volume_mute' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r6000-2-q14", "dataset": "tr_intent_paired", "id": "tr-r6000-2-q14", "haystack_uid": "tr_intent_paired:tr-r6000-2", "haystack_id": "tr-r6000-2", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "alarm_set", "candidates": ["qa_factoid", "alarm_set"], "answer": "alarm_set", "answer_key": "alarm_set", "question": "Bu kayıtlarda hangisi daha çok: 'qa_factoid' etiketli kayıtlar mı, 'alarm_set' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r6000-2-q12", "dataset": "tr_intent_paired", "id": "tr-r6000-2-q12", "haystack_uid": "tr_intent_paired:tr-r6000-2", "haystack_id": "tr-r6000-2", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "qa_factoid", "label_b": "alarm_set", "candidates": ["qa_factoid", "alarm_set"], "answer": "alarm_set", "answer_key": "alarm_set", "question": "Bu kayıtlarda hangisi daha çok: 'qa_factoid' etiketli kayıtlar mı, 'alarm_set' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent_paired:tr-r6000-3-q0", "dataset": "tr_intent_paired", "id": "tr-r6000-3-q0", "haystack_uid": "tr_intent_paired:tr-r6000-3", "haystack_id": "tr-r6000-3", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "qa_currency", "answer": "72", "question": "Bu kayıtlarda kaç tane 'qa_currency' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r6000-3-q1", "dataset": "tr_intent_paired", "id": "tr-r6000-3-q1", "haystack_uid": "tr_intent_paired:tr-r6000-3", "haystack_id": "tr-r6000-3", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "general_quirky", "answer": "304", "question": "Bu kayıtlarda kaç tane 'general_quirky' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r6000-3-q2", "dataset": "tr_intent_paired", "id": "tr-r6000-3-q2", "haystack_uid": "tr_intent_paired:tr-r6000-3", "haystack_id": "tr-r6000-3", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "transport_traffic", "answer": "138", "question": "Bu kayıtlarda kaç tane 'transport_traffic' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -134,10 +122,9 @@
 {"uid": "tr_intent_paired:tr-r6000-3-q9", "dataset": "tr_intent_paired", "id": "tr-r6000-3-q9", "haystack_uid": "tr_intent_paired:tr-r6000-3", "haystack_id": "tr-r6000-3", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "play_audiobook", "answer": "22", "question": "Bu kayıtlarda kaç tane 'play_audiobook' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r6000-3-q10", "dataset": "tr_intent_paired", "id": "tr-r6000-3-q10", "haystack_uid": "tr_intent_paired:tr-r6000-3", "haystack_id": "tr-r6000-3", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "play_podcasts", "answer": "14", "question": "Bu kayıtlarda kaç tane 'play_podcasts' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r6000-3-q11", "dataset": "tr_intent_paired", "id": "tr-r6000-3-q11", "haystack_uid": "tr_intent_paired:tr-r6000-3", "haystack_id": "tr-r6000-3", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "music_likeness", "answer": "24", "question": "Bu kayıtlarda kaç tane 'music_likeness' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent_paired:tr-r6000-3-q12", "dataset": "tr_intent_paired", "id": "tr-r6000-3-q12", "haystack_uid": "tr_intent_paired:tr-r6000-3", "haystack_id": "tr-r6000-3", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "iot_coffee", "label_b": "social_query", "candidates": ["iot_coffee", "social_query"], "answer": "iot_coffee", "answer_key": "iot_coffee", "question": "Bu kayıtlarda hangisi daha çok: 'iot_coffee' etiketli kayıtlar mı, 'social_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r6000-3-q13", "dataset": "tr_intent_paired", "id": "tr-r6000-3-q13", "haystack_uid": "tr_intent_paired:tr-r6000-3", "haystack_id": "tr-r6000-3", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "calendar_query", "label_b": "calendar_set", "candidates": ["calendar_query", "calendar_set"], "answer": "calendar_set", "answer_key": "calendar_set", "question": "Bu kayıtlarda hangisi daha çok: 'calendar_query' etiketli kayıtlar mı, 'calendar_set' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r6000-3-q14", "dataset": "tr_intent_paired", "id": "tr-r6000-3-q14", "haystack_uid": "tr_intent_paired:tr-r6000-3", "haystack_id": "tr-r6000-3", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "transport_taxi", "label_b": "calendar_remove", "candidates": ["transport_taxi", "calendar_remove"], "answer": "transport_taxi", "answer_key": "transport_taxi", "question": "Bu kayıtlarda hangisi daha çok: 'transport_taxi' etiketli kayıtlar mı, 'calendar_remove' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r6000-3-q15", "dataset": "tr_intent_paired", "id": "tr-r6000-3-q15", "haystack_uid": "tr_intent_paired:tr-r6000-3", "haystack_id": "tr-r6000-3", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "takeaway_query", "label_b": "email_query", "candidates": ["takeaway_query", "email_query"], "answer": "email_query", "answer_key": "email_query", "question": "Bu kayıtlarda hangisi daha çok: 'takeaway_query' etiketli kayıtlar mı, 'email_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r6000-3-q12", "dataset": "tr_intent_paired", "id": "tr-r6000-3-q12", "haystack_uid": "tr_intent_paired:tr-r6000-3", "haystack_id": "tr-r6000-3", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "calendar_set", "label_b": "calendar_query", "candidates": ["calendar_set", "calendar_query"], "answer": "calendar_set", "answer_key": "calendar_set", "question": "Bu kayıtlarda hangisi daha çok: 'calendar_set' etiketli kayıtlar mı, 'calendar_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r6000-3-q13", "dataset": "tr_intent_paired", "id": "tr-r6000-3-q13", "haystack_uid": "tr_intent_paired:tr-r6000-3", "haystack_id": "tr-r6000-3", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "transport_taxi", "label_b": "calendar_remove", "candidates": ["transport_taxi", "calendar_remove"], "answer": "transport_taxi", "answer_key": "transport_taxi", "question": "Bu kayıtlarda hangisi daha çok: 'transport_taxi' etiketli kayıtlar mı, 'calendar_remove' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r6000-3-q14", "dataset": "tr_intent_paired", "id": "tr-r6000-3-q14", "haystack_uid": "tr_intent_paired:tr-r6000-3", "haystack_id": "tr-r6000-3", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "email_query", "label_b": "takeaway_query", "candidates": ["email_query", "takeaway_query"], "answer": "email_query", "answer_key": "email_query", "question": "Bu kayıtlarda hangisi daha çok: 'email_query' etiketli kayıtlar mı, 'takeaway_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
 {"uid": "tr_intent_paired:tr-r6000-4-q0", "dataset": "tr_intent_paired", "id": "tr-r6000-4-q0", "haystack_uid": "tr_intent_paired:tr-r6000-4", "haystack_id": "tr-r6000-4", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "social_query", "answer": "45", "question": "Bu kayıtlarda kaç tane 'social_query' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r6000-4-q1", "dataset": "tr_intent_paired", "id": "tr-r6000-4-q1", "haystack_uid": "tr_intent_paired:tr-r6000-4", "haystack_id": "tr-r6000-4", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "iot_hue_lightchange", "answer": "54", "question": "Bu kayıtlarda kaç tane 'iot_hue_lightchange' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
 {"uid": "tr_intent_paired:tr-r6000-4-q2", "dataset": "tr_intent_paired", "id": "tr-r6000-4-q2", "haystack_uid": "tr_intent_paired:tr-r6000-4", "haystack_id": "tr-r6000-4", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "label": "iot_hue_lightoff", "answer": "15", "question": "Bu kayıtlarda kaç tane 'iot_hue_lightoff' etiketli kayıt var? Sadece sayıyı yaz.", "role": "control"}
@@ -150,23 +137,22 @@
 {"uid": "tr_intent_paired:tr-r6000-4-q9", "dataset": "tr_intent_paired", "id": "tr-r6000-4-q9", "haystack_uid": "tr_intent_paired:tr-r6000-4", "haystack_id": "tr-r6000-4", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "iot_hue_lightup", "answer": "30", "question": "Bu kayıtlarda kaç tane 'iot_hue_lightup' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r6000-4-q10", "dataset": "tr_intent_paired", "id": "tr-r6000-4-q10", "haystack_uid": "tr_intent_paired:tr-r6000-4", "haystack_id": "tr-r6000-4", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "cooking_recipe", "answer": "29", "question": "Bu kayıtlarda kaç tane 'cooking_recipe' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
 {"uid": "tr_intent_paired:tr-r6000-4-q11", "dataset": "tr_intent_paired", "id": "tr-r6000-4-q11", "haystack_uid": "tr_intent_paired:tr-r6000-4", "haystack_id": "tr-r6000-4", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "count", "rare": true, "label": "calendar_set", "answer": "22", "question": "Bu kayıtlarda kaç tane 'calendar_set' etiketli kayıt var? Sadece sayıyı yaz.", "role": "retrieval"}
-{"uid": "tr_intent_paired:tr-r6000-4-q12", "dataset": "tr_intent_paired", "id": "tr-r6000-4-q12", "haystack_uid": "tr_intent_paired:tr-r6000-4", "haystack_id": "tr-r6000-4", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "qa_currency", "label_b": "play_game", "candidates": ["qa_currency", "play_game"], "answer": "qa_currency", "answer_key": "qa_currency", "question": "Bu kayıtlarda hangisi daha çok: 'qa_currency' etiketli kayıtlar mı, 'play_game' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r6000-4-q13", "dataset": "tr_intent_paired", "id": "tr-r6000-4-q13", "haystack_uid": "tr_intent_paired:tr-r6000-4", "haystack_id": "tr-r6000-4", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "play_music", "label_b": "weather_query", "candidates": ["play_music", "weather_query"], "answer": "play_music", "answer_key": "play_music", "question": "Bu kayıtlarda hangisi daha çok: 'play_music' etiketli kayıtlar mı, 'weather_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-r6000-4-q14", "dataset": "tr_intent_paired", "id": "tr-r6000-4-q14", "haystack_uid": "tr_intent_paired:tr-r6000-4", "haystack_id": "tr-r6000-4", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "lists_query", "label_b": "lists_remove", "candidates": ["lists_query", "lists_remove"], "answer": "lists_query", "answer_key": "lists_query", "question": "Bu kayıtlarda hangisi daha çok: 'lists_query' etiketli kayıtlar mı, 'lists_remove' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
-{"uid": "tr_intent_paired:tr-core3000-0-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-0-q0", "haystack_uid": "tr_intent_paired:tr-core3000-0", "haystack_id": "tr-core3000-0", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_query", "label_b": "lists_createoradd", "candidates": ["lists_query", "lists_createoradd"], "answer": "lists_query", "answer_key": "lists_query", "question": "Bu kayıtlarda hangisi daha çok: 'lists_query' etiketli kayıtlar mı, 'lists_createoradd' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "tr_intent_paired:tr-core3000-0-q1", "dataset": "tr_intent_paired", "id": "tr-core3000-0-q1", "haystack_uid": "tr_intent_paired:tr-core3000-0", "haystack_id": "tr-core3000-0", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "email_query", "label_b": "weather_query", "candidates": ["email_query", "weather_query"], "answer": "weather_query", "answer_key": "weather_query", "question": "Bu kayıtlarda hangisi daha çok: 'email_query' etiketli kayıtlar mı, 'weather_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "tr_intent_paired:tr-core3000-1-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-1-q0", "haystack_uid": "tr_intent_paired:tr-core3000-1", "haystack_id": "tr-core3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_query", "label_b": "news_query", "candidates": ["lists_query", "news_query"], "answer": "lists_query", "answer_key": "lists_query", "question": "Bu kayıtlarda hangisi daha çok: 'lists_query' etiketli kayıtlar mı, 'news_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "tr_intent_paired:tr-core3000-1-q1", "dataset": "tr_intent_paired", "id": "tr-core3000-1-q1", "haystack_uid": "tr_intent_paired:tr-core3000-1", "haystack_id": "tr-core3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_set", "label_b": "calendar_query", "candidates": ["calendar_set", "calendar_query"], "answer": "calendar_query", "answer_key": "calendar_query", "question": "Bu kayıtlarda hangisi daha çok: 'calendar_set' etiketli kayıtlar mı, 'calendar_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "tr_intent_paired:tr-core3000-2-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-2-q0", "haystack_uid": "tr_intent_paired:tr-core3000-2", "haystack_id": "tr-core3000-2", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_createoradd", "label_b": "recommendation_locations", "candidates": ["lists_createoradd", "recommendation_locations"], "answer": "lists_createoradd", "answer_key": "lists_createoradd", "question": "Bu kayıtlarda hangisi daha çok: 'lists_createoradd' etiketli kayıtlar mı, 'recommendation_locations' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "tr_intent_paired:tr-core3000-2-q1", "dataset": "tr_intent_paired", "id": "tr-core3000-2-q1", "haystack_uid": "tr_intent_paired:tr-core3000-2", "haystack_id": "tr-core3000-2", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_query", "label_b": "transport_query", "candidates": ["lists_query", "transport_query"], "answer": "transport_query", "answer_key": "transport_query", "question": "Bu kayıtlarda hangisi daha çok: 'lists_query' etiketli kayıtlar mı, 'transport_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "tr_intent_paired:tr-core3000-3-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-3-q0", "haystack_uid": "tr_intent_paired:tr-core3000-3", "haystack_id": "tr-core3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "recommendation_events", "label_b": "datetime_query", "candidates": ["recommendation_events", "datetime_query"], "answer": "recommendation_events", "answer_key": "recommendation_events", "question": "Bu kayıtlarda hangisi daha çok: 'recommendation_events' etiketli kayıtlar mı, 'datetime_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "tr_intent_paired:tr-core3000-3-q1", "dataset": "tr_intent_paired", "id": "tr-core3000-3-q1", "haystack_uid": "tr_intent_paired:tr-core3000-3", "haystack_id": "tr-core3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "play_radio", "label_b": "lists_query", "candidates": ["play_radio", "lists_query"], "answer": "lists_query", "answer_key": "lists_query", "question": "Bu kayıtlarda hangisi daha çok: 'play_radio' etiketli kayıtlar mı, 'lists_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "tr_intent_paired:tr-r6000-4-q12", "dataset": "tr_intent_paired", "id": "tr-r6000-4-q12", "haystack_uid": "tr_intent_paired:tr-r6000-4", "haystack_id": "tr-r6000-4", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "weather_query", "label_b": "play_music", "candidates": ["weather_query", "play_music"], "answer": "play_music", "answer_key": "play_music", "question": "Bu kayıtlarda hangisi daha çok: 'weather_query' etiketli kayıtlar mı, 'play_music' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-r6000-4-q13", "dataset": "tr_intent_paired", "id": "tr-r6000-4-q13", "haystack_uid": "tr_intent_paired:tr-r6000-4", "haystack_id": "tr-r6000-4", "language": "tr", "target_tokens": null, "target_records": 6000, "kind": "close_comparison", "label": null, "label_a": "qa_currency", "label_b": "email_sendemail", "candidates": ["qa_currency", "email_sendemail"], "answer": "email_sendemail", "answer_key": "email_sendemail", "question": "Bu kayıtlarda hangisi daha çok: 'qa_currency' etiketli kayıtlar mı, 'email_sendemail' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core"}
+{"uid": "tr_intent_paired:tr-core3000-0-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-0-q0", "haystack_uid": "tr_intent_paired:tr-core3000-0", "haystack_id": "tr-core3000-0", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "email_query", "label_b": "lists_createoradd", "candidates": ["email_query", "lists_createoradd"], "answer": "email_query", "answer_key": "email_query", "question": "Bu kayıtlarda hangisi daha çok: 'email_query' etiketli kayıtlar mı, 'lists_createoradd' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "tr_intent_paired:tr-core3000-0-q1", "dataset": "tr_intent_paired", "id": "tr-core3000-0-q1", "haystack_uid": "tr_intent_paired:tr-core3000-0", "haystack_id": "tr-core3000-0", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "social_post", "label_b": "weather_query", "candidates": ["social_post", "weather_query"], "answer": "weather_query", "answer_key": "weather_query", "question": "Bu kayıtlarda hangisi daha çok: 'social_post' etiketli kayıtlar mı, 'weather_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "tr_intent_paired:tr-core3000-1-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-1-q0", "haystack_uid": "tr_intent_paired:tr-core3000-1", "haystack_id": "tr-core3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "news_query", "label_b": "calendar_set", "candidates": ["news_query", "calendar_set"], "answer": "news_query", "answer_key": "news_query", "question": "Bu kayıtlarda hangisi daha çok: 'news_query' etiketli kayıtlar mı, 'calendar_set' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "tr_intent_paired:tr-core3000-1-q1", "dataset": "tr_intent_paired", "id": "tr-core3000-1-q1", "haystack_uid": "tr_intent_paired:tr-core3000-1", "haystack_id": "tr-core3000-1", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_query", "label_b": "calendar_remove", "candidates": ["calendar_query", "calendar_remove"], "answer": "calendar_remove", "answer_key": "calendar_remove", "question": "Bu kayıtlarda hangisi daha çok: 'calendar_query' etiketli kayıtlar mı, 'calendar_remove' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "tr_intent_paired:tr-core3000-2-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-2-q0", "haystack_uid": "tr_intent_paired:tr-core3000-2", "haystack_id": "tr-core3000-2", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "transport_query", "label_b": "lists_createoradd", "candidates": ["transport_query", "lists_createoradd"], "answer": "transport_query", "answer_key": "transport_query", "question": "Bu kayıtlarda hangisi daha çok: 'transport_query' etiketli kayıtlar mı, 'lists_createoradd' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "tr_intent_paired:tr-core3000-2-q1", "dataset": "tr_intent_paired", "id": "tr-core3000-2-q1", "haystack_uid": "tr_intent_paired:tr-core3000-2", "haystack_id": "tr-core3000-2", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "play_radio", "label_b": "calendar_set", "candidates": ["play_radio", "calendar_set"], "answer": "calendar_set", "answer_key": "calendar_set", "question": "Bu kayıtlarda hangisi daha çok: 'play_radio' etiketli kayıtlar mı, 'calendar_set' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "tr_intent_paired:tr-core3000-3-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-3-q0", "haystack_uid": "tr_intent_paired:tr-core3000-3", "haystack_id": "tr-core3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "play_radio", "label_b": "datetime_query", "candidates": ["play_radio", "datetime_query"], "answer": "play_radio", "answer_key": "play_radio", "question": "Bu kayıtlarda hangisi daha çok: 'play_radio' etiketli kayıtlar mı, 'datetime_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "tr_intent_paired:tr-core3000-3-q1", "dataset": "tr_intent_paired", "id": "tr-core3000-3-q1", "haystack_uid": "tr_intent_paired:tr-core3000-3", "haystack_id": "tr-core3000-3", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_query", "label_b": "cooking_recipe", "candidates": ["calendar_query", "cooking_recipe"], "answer": "cooking_recipe", "answer_key": "cooking_recipe", "question": "Bu kayıtlarda hangisi daha çok: 'calendar_query' etiketli kayıtlar mı, 'cooking_recipe' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "tr_intent_paired:tr-core3000-4-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-4-q0", "haystack_uid": "tr_intent_paired:tr-core3000-4", "haystack_id": "tr-core3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "social_post", "label_b": "transport_query", "candidates": ["social_post", "transport_query"], "answer": "social_post", "answer_key": "social_post", "question": "Bu kayıtlarda hangisi daha çok: 'social_post' etiketli kayıtlar mı, 'transport_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "tr_intent_paired:tr-core3000-4-q1", "dataset": "tr_intent_paired", "id": "tr-core3000-4-q1", "haystack_uid": "tr_intent_paired:tr-core3000-4", "haystack_id": "tr-core3000-4", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "alarm_set", "label_b": "calendar_set", "candidates": ["alarm_set", "calendar_set"], "answer": "calendar_set", "answer_key": "calendar_set", "question": "Bu kayıtlarda hangisi daha çok: 'alarm_set' etiketli kayıtlar mı, 'calendar_set' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "tr_intent_paired:tr-core3000-5-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-5-q0", "haystack_uid": "tr_intent_paired:tr-core3000-5", "haystack_id": "tr-core3000-5", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_remove", "label_b": "weather_query", "candidates": ["calendar_remove", "weather_query"], "answer": "calendar_remove", "answer_key": "calendar_remove", "question": "Bu kayıtlarda hangisi daha çok: 'calendar_remove' etiketli kayıtlar mı, 'weather_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "tr_intent_paired:tr-core3000-5-q1", "dataset": "tr_intent_paired", "id": "tr-core3000-5-q1", "haystack_uid": "tr_intent_paired:tr-core3000-5", "haystack_id": "tr-core3000-5", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "news_query", "label_b": "cooking_recipe", "candidates": ["news_query", "cooking_recipe"], "answer": "cooking_recipe", "answer_key": "cooking_recipe", "question": "Bu kayıtlarda hangisi daha çok: 'news_query' etiketli kayıtlar mı, 'cooking_recipe' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "tr_intent_paired:tr-core3000-6-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-6-q0", "haystack_uid": "tr_intent_paired:tr-core3000-6", "haystack_id": "tr-core3000-6", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "general_quirky", "label_b": "qa_definition", "candidates": ["general_quirky", "qa_definition"], "answer": "general_quirky", "answer_key": "general_quirky", "question": "Bu kayıtlarda hangisi daha çok: 'general_quirky' etiketli kayıtlar mı, 'qa_definition' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
-{"uid": "tr_intent_paired:tr-core3000-6-q1", "dataset": "tr_intent_paired", "id": "tr-core3000-6-q1", "haystack_uid": "tr_intent_paired:tr-core3000-6", "haystack_id": "tr-core3000-6", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "play_podcasts", "label_b": "lists_query", "candidates": ["play_podcasts", "lists_query"], "answer": "lists_query", "answer_key": "lists_query", "question": "Bu kayıtlarda hangisi daha çok: 'play_podcasts' etiketli kayıtlar mı, 'lists_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "tr_intent_paired:tr-core3000-6-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-6-q0", "haystack_uid": "tr_intent_paired:tr-core3000-6", "haystack_id": "tr-core3000-6", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "qa_definition", "label_b": "play_podcasts", "candidates": ["qa_definition", "play_podcasts"], "answer": "qa_definition", "answer_key": "qa_definition", "question": "Bu kayıtlarda hangisi daha çok: 'qa_definition' etiketli kayıtlar mı, 'play_podcasts' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
+{"uid": "tr_intent_paired:tr-core3000-6-q1", "dataset": "tr_intent_paired", "id": "tr-core3000-6-q1", "haystack_uid": "tr_intent_paired:tr-core3000-6", "haystack_id": "tr-core3000-6", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "lists_createoradd", "label_b": "news_query", "candidates": ["lists_createoradd", "news_query"], "answer": "news_query", "answer_key": "news_query", "question": "Bu kayıtlarda hangisi daha çok: 'lists_createoradd' etiketli kayıtlar mı, 'news_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "tr_intent_paired:tr-core3000-7-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-7-q0", "haystack_uid": "tr_intent_paired:tr-core3000-7", "haystack_id": "tr-core3000-7", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_remove", "label_b": "email_query", "candidates": ["calendar_remove", "email_query"], "answer": "calendar_remove", "answer_key": "calendar_remove", "question": "Bu kayıtlarda hangisi daha çok: 'calendar_remove' etiketli kayıtlar mı, 'email_query' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "tr_intent_paired:tr-core3000-7-q1", "dataset": "tr_intent_paired", "id": "tr-core3000-7-q1", "haystack_uid": "tr_intent_paired:tr-core3000-7", "haystack_id": "tr-core3000-7", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_set", "label_b": "alarm_set", "candidates": ["calendar_set", "alarm_set"], "answer": "alarm_set", "answer_key": "alarm_set", "question": "Bu kayıtlarda hangisi daha çok: 'calendar_set' etiketli kayıtlar mı, 'alarm_set' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}
 {"uid": "tr_intent_paired:tr-core3000-8-q0", "dataset": "tr_intent_paired", "id": "tr-core3000-8-q0", "haystack_uid": "tr_intent_paired:tr-core3000-8", "haystack_id": "tr-core3000-8", "language": "tr", "target_tokens": null, "target_records": 3000, "kind": "close_comparison", "label": null, "label_a": "calendar_set", "label_b": "play_podcasts", "candidates": ["calendar_set", "play_podcasts"], "answer": "calendar_set", "answer_key": "calendar_set", "question": "Bu kayıtlarda hangisi daha çok: 'calendar_set' etiketli kayıtlar mı, 'play_podcasts' etiketli kayıtlar mı? Sadece etiket adını yaz.", "role": "core", "core_document": true}

vitamins_tr/haystacks.jsonl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8d63b3c3f9c329021ea065c52db4d276f4fe9b4493b708640755e5e670c5d5ba
-size 31189419
+oid sha256:e98aa32441b85d35175d038cfd788188da68a43229361f5093363d50a02c41ed
+size 31051259

vitamins_tr/manifest.json

@@ -1,6 +1,6 @@
 {
-  "version": "0.9.0",
-  "date": "2026-09-29",
+  "version": "0.10.0",
+  "date": "2026-10-01",
   "environment": {
     "python": "3.12.9",
     "polars": "1.42.0",
@@ -17,6 +17,17 @@
     "drop_label_leakage": true,
     "leak_label_words": false,
     "min_class_support": 0,
+    "boilerplate_keywords": [],
+    "boilerplate_min_keywords": 3,
+    "boilerplate_weak_min": 2,
+    "boilerplate_regex": "",
+    "drop_text_regex": [
+      "\\b\\d{1,3}([.,]\\d)?\\s*/\\s*(10|5|100)\\b",
+      "\\b\\d{1,2}\\s*(yıldız|puan)",
+      "\\bpuan[ıi]m\\b",
+      "\\b(on|dokuz|sekiz|yedi|altı|beş|dört|üç|iki|bir|sıfır)\\s+(yıldız|puan)"
+    ],
+    "strip_html": true,
     "seed": 42,
     "haystack_target_tokens": [
       100000,
@@ -78,6 +89,7 @@
       "nötr"
     ],
     "close_ordinal_gap": 0,
+    "close_allowed_pairs_file": "experiments/separable_pairs/vitamins_tr.json",
     "core_doc_records": [
       2500,
       6000,
@@ -115,57 +127,58 @@
     "out_dir": "vitamins_tr_out",
     "dataset": ""
   },
-  "source_hash_first1000": "ec88c923ecabf044",
-  "rows_after_cleaning": 38968,
+  "source_hash_first1000": "8dddf5a9ff8b5af5",
+  "rows_after_cleaning": 38529,
   "cleaning": {
-    "rows_dropped_label_leakage": 177,
-    "label_leakage_rate": 0.004522,
+    "rows_dropped_text_regex": 449,
+    "rows_dropped_label_leakage": 170,
+    "label_leakage_rate": 0.004393,
     "classes_dropped_low_support": [],
     "min_class_support": 0
   },
   "ranking_feasibility": {
     "n_classes": 3,
-    "min_class_pool": 7989,
-    "max_records": 23967,
+    "min_class_pool": 7875,
+    "max_records": 23625,
     "binding": true,
-    "ranking_length_ceiling_tokens": 855621
+    "ranking_length_ceiling_tokens": 793469
   },
   "tier_overlap": {
     "100000": {
       "n_haystacks": 10,
-      "mean_jaccard": 0.0412,
-      "max_jaccard": 0.111,
-      "pool_fraction_per_haystack": 0.0616
+      "mean_jaccard": 0.0426,
+      "max_jaccard": 0.1165,
+      "pool_fraction_per_haystack": 0.0631
     },
     "250000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.0882,
-      "max_jaccard": 0.1151,
-      "pool_fraction_per_haystack": 0.1559
+      "mean_jaccard": 0.0906,
+      "max_jaccard": 0.12,
+      "pool_fraction_per_haystack": 0.1591
     },
     "500000": {
       "n_haystacks": 5,
-      "mean_jaccard": 0.1887,
-      "max_jaccard": 0.2511,
-      "pool_fraction_per_haystack": 0.3193
+      "mean_jaccard": 0.1943,
+      "max_jaccard": 0.2526,
+      "pool_fraction_per_haystack": 0.3263
     },
     "core12000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.186,
-      "max_jaccard": 0.1906,
-      "pool_fraction_per_haystack": 0.3079
+      "mean_jaccard": 0.1887,
+      "max_jaccard": 0.1919,
+      "pool_fraction_per_haystack": 0.3115
     },
     "core2500": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0341,
-      "max_jaccard": 0.0382,
-      "pool_fraction_per_haystack": 0.0642
+      "mean_jaccard": 0.0346,
+      "max_jaccard": 0.0384,
+      "pool_fraction_per_haystack": 0.0649
     },
     "core6000": {
       "n_haystacks": 6,
-      "mean_jaccard": 0.0853,
-      "max_jaccard": 0.0908,
-      "pool_fraction_per_haystack": 0.154
+      "mean_jaccard": 0.0866,
+      "max_jaccard": 0.0907,
+      "pool_fraction_per_haystack": 0.1557
     }
   },
   "max_pool_fraction": 0.35,
@@ -178,7 +191,7 @@
   "tiers_built_over_pool_cap": [
     {
       "tier": 500000,
-      "pool_fraction": 0.3774
+      "pool_fraction": 0.4057
     }
   ],
   "label_space": 3,
@@ -188,12 +201,12 @@
   "questions_written": 187,
   "kind_distribution": {
     "close_comparison": 18,
-    "count": 51,
+    "count": 52,
     "label_vs_label": 20,
-    "least_common": 17,
+    "least_common": 16,
     "most_common": 20,
-    "proportion": 42,
-    "second_most": 19
+    "proportion": 43,
+    "second_most": 18
   },
   "rare_count_questions": 0,
   "rare_band": [
@@ -203,189 +216,189 @@
   "haystacks": [
     {
       "haystack_id": "tr-100000-0",
-      "n_examples": 2586,
-      "n_tokens": 99266,
-      "n_chars": 286183,
+      "n_examples": 2619,
+      "n_tokens": 99223,
+      "n_chars": 285723,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-100000-1",
-      "n_examples": 2390,
-      "n_tokens": 99288,
-      "n_chars": 286701,
+      "n_examples": 2450,
+      "n_tokens": 99284,
+      "n_chars": 286844,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-100000-2",
-      "n_examples": 2324,
-      "n_tokens": 99267,
-      "n_chars": 289374,
+      "n_examples": 2373,
+      "n_tokens": 99277,
+      "n_chars": 289096,
       "drift_target": "olumlu",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-100000-3",
-      "n_examples": 2470,
-      "n_tokens": 99217,
-      "n_chars": 287214,
-      "drift_target": "olumsuz",
+      "n_examples": 2511,
+      "n_tokens": 99281,
+      "n_chars": 287583,
+      "drift_target": "olumlu",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-100000-4",
-      "n_examples": 2406,
-      "n_tokens": 99267,
-      "n_chars": 286115,
+      "n_examples": 2405,
+      "n_tokens": 99262,
+      "n_chars": 285949,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-100000-5",
-      "n_examples": 2373,
-      "n_tokens": 99264,
-      "n_chars": 285899,
-      "drift_target": "nötr",
+      "n_examples": 2352,
+      "n_tokens": 99311,
+      "n_chars": 285812,
+      "drift_target": "olumsuz",
       "drift_ok": true,
-      "n_questions": 8
+      "n_questions": 7
     },
     {
       "haystack_id": "tr-100000-6",
-      "n_examples": 2320,
-      "n_tokens": 99204,
-      "n_chars": 288176,
-      "drift_target": "olumlu",
+      "n_examples": 2377,
+      "n_tokens": 99243,
+      "n_chars": 287916,
+      "drift_target": "olumsuz",
       "drift_ok": true,
-      "n_questions": 9
+      "n_questions": 10
     },
     {
       "haystack_id": "tr-100000-7",
-      "n_examples": 2426,
-      "n_tokens": 99285,
-      "n_chars": 286928,
-      "drift_target": "olumsuz",
+      "n_examples": 2484,
+      "n_tokens": 99258,
+      "n_chars": 286397,
+      "drift_target": "nötr",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-100000-8",
-      "n_examples": 2246,
-      "n_tokens": 99216,
-      "n_chars": 286602,
-      "drift_target": "olumsuz",
+      "n_examples": 2213,
+      "n_tokens": 99265,
+      "n_chars": 287351,
+      "drift_target": "nötr",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-100000-9",
-      "n_examples": 2480,
-      "n_tokens": 99278,
-      "n_chars": 286312,
-      "drift_target": "olumlu",
+      "n_examples": 2510,
+      "n_tokens": 99268,
+      "n_chars": 287218,
+      "drift_target": "nötr",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-250000-0",
-      "n_examples": 6573,
-      "n_tokens": 248130,
-      "n_chars": 718886,
+      "n_examples": 6553,
+      "n_tokens": 248145,
+      "n_chars": 720312,
       "drift_target": "olumlu",
       "drift_ok": true,
       "n_questions": 7
     },
     {
       "haystack_id": "tr-250000-1",
-      "n_examples": 5770,
-      "n_tokens": 248236,
-      "n_chars": 715675,
+      "n_examples": 5817,
+      "n_tokens": 248280,
+      "n_chars": 714674,
       "drift_target": "nötr",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "tr-250000-2",
-      "n_examples": 5830,
-      "n_tokens": 248104,
-      "n_chars": 714413,
+      "n_examples": 5908,
+      "n_tokens": 248159,
+      "n_chars": 718262,
       "drift_target": "nötr",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "tr-250000-3",
-      "n_examples": 5977,
-      "n_tokens": 248186,
-      "n_chars": 716583,
+      "n_examples": 6011,
+      "n_tokens": 248006,
+      "n_chars": 715207,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "tr-250000-4",
-      "n_examples": 6229,
-      "n_tokens": 248174,
-      "n_chars": 716681,
+      "n_examples": 6352,
+      "n_tokens": 248224,
+      "n_chars": 716413,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 9
     },
     {
       "haystack_id": "tr-500000-0",
-      "n_examples": 12064,
-      "n_tokens": 496377,
-      "n_chars": 1431675,
+      "n_examples": 12221,
+      "n_tokens": 496369,
+      "n_chars": 1431330,
       "drift_target": "olumlu",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "tr-500000-1",
-      "n_examples": 12271,
-      "n_tokens": 496506,
-      "n_chars": 1432692,
+      "n_examples": 12394,
+      "n_tokens": 496474,
+      "n_chars": 1433278,
       "drift_target": "olumsuz",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "tr-500000-2",
-      "n_examples": 12523,
-      "n_tokens": 496526,
-      "n_chars": 1437530,
-      "drift_target": "olumlu",
+      "n_examples": 12647,
+      "n_tokens": 496451,
+      "n_chars": 1436743,
+      "drift_target": "nötr",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "tr-500000-3",
-      "n_examples": 12898,
-      "n_tokens": 496317,
-      "n_chars": 1442793,
-      "drift_target": "olumlu",
+      "n_examples": 12946,
+      "n_tokens": 496397,
+      "n_chars": 1438684,
+      "drift_target": "nötr",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "tr-500000-4",
-      "n_examples": 12465,
-      "n_tokens": 496364,
-      "n_chars": 1445970,
-      "drift_target": "olumsuz",
+      "n_examples": 12660,
+      "n_tokens": 496346,
+      "n_chars": 1445025,
+      "drift_target": "nötr",
       "drift_ok": true,
       "n_questions": 8
     },
     {
       "haystack_id": "tr-core2500-0",
       "n_examples": 2500,
-      "n_tokens": 100929,
-      "n_chars": 292099,
+      "n_tokens": 101562,
+      "n_chars": 294572,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -399,8 +412,8 @@
     {
       "haystack_id": "tr-core2500-1",
       "n_examples": 2500,
-      "n_tokens": 101832,
-      "n_chars": 294483,
+      "n_tokens": 101856,
+      "n_chars": 294673,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -414,8 +427,8 @@
     {
       "haystack_id": "tr-core2500-2",
       "n_examples": 2500,
-      "n_tokens": 100977,
-      "n_chars": 292553,
+      "n_tokens": 101398,
+      "n_chars": 293720,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -429,8 +442,8 @@
     {
       "haystack_id": "tr-core2500-3",
       "n_examples": 2500,
-      "n_tokens": 100853,
-      "n_chars": 292348,
+      "n_tokens": 100445,
+      "n_chars": 291026,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -444,8 +457,8 @@
     {
       "haystack_id": "tr-core2500-4",
       "n_examples": 2500,
-      "n_tokens": 100795,
-      "n_chars": 292590,
+      "n_tokens": 100329,
+      "n_chars": 290393,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -459,8 +472,8 @@
     {
       "haystack_id": "tr-core2500-5",
       "n_examples": 2500,
-      "n_tokens": 99502,
-      "n_chars": 288711,
+      "n_tokens": 99923,
+      "n_chars": 288851,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -474,8 +487,8 @@
     {
       "haystack_id": "tr-core6000-0",
       "n_examples": 6000,
-      "n_tokens": 243334,
-      "n_chars": 704120,
+      "n_tokens": 242362,
+      "n_chars": 702013,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -489,8 +502,8 @@
     {
       "haystack_id": "tr-core6000-1",
       "n_examples": 6000,
-      "n_tokens": 242057,
-      "n_chars": 701449,
+      "n_tokens": 238923,
+      "n_chars": 692193,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -504,8 +517,8 @@
     {
       "haystack_id": "tr-core6000-2",
       "n_examples": 6000,
-      "n_tokens": 239879,
-      "n_chars": 694266,
+      "n_tokens": 240745,
+      "n_chars": 698622,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -519,8 +532,8 @@
     {
       "haystack_id": "tr-core6000-3",
       "n_examples": 6000,
-      "n_tokens": 241772,
-      "n_chars": 700132,
+      "n_tokens": 240919,
+      "n_chars": 697895,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -534,8 +547,8 @@
     {
       "haystack_id": "tr-core6000-4",
       "n_examples": 6000,
-      "n_tokens": 242337,
-      "n_chars": 700602,
+      "n_tokens": 239953,
+      "n_chars": 693241,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -549,8 +562,8 @@
     {
       "haystack_id": "tr-core6000-5",
       "n_examples": 6000,
-      "n_tokens": 242302,
-      "n_chars": 701220,
+      "n_tokens": 239852,
+      "n_chars": 695101,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -564,8 +577,8 @@
     {
       "haystack_id": "tr-core12000-0",
       "n_examples": 12000,
-      "n_tokens": 487553,
-      "n_chars": 1411874,
+      "n_tokens": 480991,
+      "n_chars": 1391289,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -579,8 +592,8 @@
     {
       "haystack_id": "tr-core12000-1",
       "n_examples": 12000,
-      "n_tokens": 481270,
-      "n_chars": 1393905,
+      "n_tokens": 479833,
+      "n_chars": 1389305,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -594,8 +607,8 @@
     {
       "haystack_id": "tr-core12000-2",
       "n_examples": 12000,
-      "n_tokens": 485100,
-      "n_chars": 1404452,
+      "n_tokens": 483697,
+      "n_chars": 1399628,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -609,8 +622,8 @@
     {
       "haystack_id": "tr-core12000-3",
       "n_examples": 12000,
-      "n_tokens": 487963,
-      "n_chars": 1413865,
+      "n_tokens": 480354,
+      "n_chars": 1390198,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -624,8 +637,8 @@
     {
       "haystack_id": "tr-core12000-4",
       "n_examples": 12000,
-      "n_tokens": 487506,
-      "n_chars": 1411082,
+      "n_tokens": 481607,
+      "n_chars": 1393713,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,
@@ -639,8 +652,8 @@
     {
       "haystack_id": "tr-core12000-5",
       "n_examples": 12000,
-      "n_tokens": 489875,
-      "n_chars": 1416086,
+      "n_tokens": 477671,
+      "n_chars": 1383138,
       "drift_target": null,
       "drift_ok": false,
       "n_questions": 1,