Dataset Viewer
Auto-converted to Parquet Duplicate
id
stringlengths
46
111
messages
listlengths
2
6
filter_pass
bool
2 classes
reject_reasons
listlengths
0
3
target_lang_frac
float64
0.01
1
length_ratio
float64
0.11
8.5
src_tokens
int64
204
23k
trans_tokens
int64
251
24.6k
OpenThoughts3-full-filtered-math-decontam-v2_155331
[ { "content": "Součet krychłych mocnin tří kladných celých čísel je 300. Součin těchto tří celých čísel je 125. Kolik je součet převrácených hodnot těchto tří celých čísel?", "role": "user" }, { "content": "<think>Dobře, mám tady tento příklad: Součet krychłych mocnin tří kladných celých čísel je 300...
false
[ "truncated" ]
1
1.162933
15,043
17,494
correct-python-sft-187k-x16-thoughts-filtered-decontam-v2_b262a57e-d59a-46ed-8a27-6848743de3c4
[ { "content": "Vaším úkolem je vytvořit funkci, která simuluje jednoduchou datovou strukturu zásobníku (stack) pomocí seznamu. Zásobník by měl podporovat následující operace: 1. `push(item)`: Přidá prvek na vrchol zásobníku. 2. `pop()`: Odstraní a vrátí prvek z vrcholu zásobníku. 3. `peek()`: Vrátí prvek z vrcho...
true
[]
1
1.270428
4,112
5,224
if_qwq_reasoning_verified_filtered_decontam_v2_116820
[ { "content": "Navzdory integraci autentizačního protokolu do mé webové platformy založené na Django, neoprávněná vniknutí nadále představují zásadní problém. Mohli byste poskytnout vyčerpávající průvodce, který popsáše přesné techniky a osvědčené postupy pro vytvoření robustního a bezpečného autentizačního syst...
true
[]
1
1.548004
2,104
3,257
OpenThoughts3-full-filtered-math-decontam-v2_145630
[ { "content": "Určete největší možný kladný pěticiferný palindrom, který je dělitelný 9 a má pouze různé cifry.", "role": "user" }, { "content": "<think>Dobře, musím najít největší pěticiferný palindrom, který je dělitelný 9 a má všechny různé cifry. Hmm, okay. Pojďme to rozebrat krok za krokem. Nejp...
false
[ "truncated" ]
1
1.182199
14,314
16,922
OpenThoughts3-full-filtered-science-decontam-v2_1760
[ { "content": "Odštěpení atomu vodíku z ethanu jodovým radikálem je vysoce endotermická reakce (\\( \\Delta H^\\circ = 101 \\text{ kJ/mol} = 30 \\text{ kcal/mol} \\); viz tabulka 5.1 na straně 210 textu), takže jodové radikály budou spíše znovu tvořit \\( \\text{I}_2 \\), než aby odštěpily atom vodíku. O jaký ty...
true
[]
1
1.53688
4,474
6,876
correct-python-sft-187k-x16-thoughts-filtered-decontam-v2_20071180-9f69-4a21-86fb-76f2a5875de2
[ { "content": "Implementujte funkci `get_unique_genres(genres: List[str]) -> List[str]`, která pro daný seznam hudebních žánrů vrátí seznam unikátních žánrů přítomných ve vstupním seznamu. Výstupní seznam by měl zachovat pořadí prvních výskytů žánrů ve vstupním seznamu. Vstupní seznam může obsahovat duplikáty ne...
true
[]
1
1.280822
1,022
1,309
OpenThoughts3-full-filtered-math-decontam-v2_77008
[{"content":"Nechť $a,$ $b,$ $c$ jsou kladná reálná čísla. Definujme $f(a,b,c)$ jako kladné r(...TRUNCATED)
false
[ "truncated" ]
1
1.241407
13,906
17,263
correct-python-sft-187k-x16-thoughts-filtered-decontam-v2_d653eab4-5efc-4878-960f-329dddfeec1d
[{"content":"Zadaný seznam celých čísel využijte k nalezení maximálního součtu spojitého p(...TRUNCATED)
true
[]
1
1.250874
6,290
7,868
correct-python-sft-187k-x16-thoughts-filtered-decontam-v2_b831ea12-291d-4cef-b328-7d03b120e5a8
[{"content":"Je vám zadán řetězec představující posloupnost bajtů v hexadecimálním formát(...TRUNCATED)
true
[]
1
1.327575
1,505
1,998
SYNTHETIC-2-SFT-cn-fltrd-final-ngram-filtered-chinese-filtered_100646
[{"content":"Prosím o posouzení, zda je následující řešení otázky správné.\n\nOtázka:\nV(...TRUNCATED)
true
[]
1
1.086596
5,543
6,023
End of preview. Expand in Data Studio

Dolci-Think-SFT-translated

Machine translations of the Dolci-Think-SFT-32B dataset, produced with gemma-4-31B-it. The samples selected for translation are those where content_quality == "excellent" according to the propella annotations.

Columns

Each row is a translated conversation plus the result of a post-translation quality filter:

  • id — source record id.
  • messages — the translated conversation (list of {content, role}).
  • filter_pass — true if the row passed every filter check (the recommended clean subset).
  • reject_reasons — empty when filter_pass, otherwise any of: structure (broken <think>/turn mismatch), truncated (a trace/answer hit the generation token cap or has unbalanced delimiters), wrong_lang (target-language character fraction below 0.85), length (translation/source token ratio outside [0.60, 2.5]), degenerate (severe repetition).
  • target_lang_frac — fraction of code-stripped prose characters in the target language.
  • length_ratio — translated/source token ratio.
  • src_tokens, trans_tokens — token counts (gemma-4-31B-it tokenizer).

The dominant rejection axis is truncated: some long reasoning traces exceeded the generation cap and were cut off mid-thought (these are complete-quality translations that simply ran out of token budget). Filter the dataset with filter_pass == true for the clean subset, or re-threshold using the raw columns.

Language Rows Clean (filter_pass) Clean %
cs 96424 57993 60.1%
de 128405 73739 57.4%
el 75644 38583 51.0%
es 85744 47202 55.0%
fi 137705 83412 60.6%
fr 129597 72704 56.1%
it 201671 180710 89.6%
nl 173362 153934 88.8%
pl 163881 145298 88.7%
ro 146859 129301 88.0%
sv 141418 125170 88.5%
uk 134626 117252 87.1%
Downloads last month
1,692