Upload tokenizer_parakeet_48k_261k/tokenizer_full_stats.json with huggingface_hub
Browse files
tokenizer_parakeet_48k_261k/tokenizer_full_stats.json
ADDED
|
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"en": {
|
| 3 |
+
"tokens": 453320060.0,
|
| 4 |
+
"words": 349957279.0,
|
| 5 |
+
"lines": 16916959.0,
|
| 6 |
+
"tokens_per_word": 1.2953582828605774,
|
| 7 |
+
"avg_tokens_per_line": 26.796781856597278
|
| 8 |
+
},
|
| 9 |
+
"hi": {
|
| 10 |
+
"tokens": 280684874.0,
|
| 11 |
+
"words": 210299384.0,
|
| 12 |
+
"lines": 10398436.0,
|
| 13 |
+
"tokens_per_word": 1.3346918505476935,
|
| 14 |
+
"avg_tokens_per_line": 26.992989522655137
|
| 15 |
+
},
|
| 16 |
+
"te": {
|
| 17 |
+
"tokens": 209728016.0,
|
| 18 |
+
"words": 119466611.0,
|
| 19 |
+
"lines": 8520112.0,
|
| 20 |
+
"tokens_per_word": 1.7555366662238372,
|
| 21 |
+
"avg_tokens_per_line": 24.61564073336125
|
| 22 |
+
},
|
| 23 |
+
"ml": {
|
| 24 |
+
"tokens": 151621954.0,
|
| 25 |
+
"words": 77461860.0,
|
| 26 |
+
"lines": 5431191.0,
|
| 27 |
+
"tokens_per_word": 1.9573755910328,
|
| 28 |
+
"avg_tokens_per_line": 27.916888579318975
|
| 29 |
+
},
|
| 30 |
+
"pa": {
|
| 31 |
+
"tokens": 149498418.0,
|
| 32 |
+
"words": 108515094.0,
|
| 33 |
+
"lines": 5036773.0,
|
| 34 |
+
"tokens_per_word": 1.377673948289627,
|
| 35 |
+
"avg_tokens_per_line": 29.68138885750857
|
| 36 |
+
},
|
| 37 |
+
"ta": {
|
| 38 |
+
"tokens": 132895268.0,
|
| 39 |
+
"words": 72855361.0,
|
| 40 |
+
"lines": 4533355.0,
|
| 41 |
+
"tokens_per_word": 1.8240973097367537,
|
| 42 |
+
"avg_tokens_per_line": 29.314992538638602
|
| 43 |
+
},
|
| 44 |
+
"kn": {
|
| 45 |
+
"tokens": 86369605.0,
|
| 46 |
+
"words": 45502553.0,
|
| 47 |
+
"lines": 2988443.0,
|
| 48 |
+
"tokens_per_word": 1.8981265732496373,
|
| 49 |
+
"avg_tokens_per_line": 28.90120541030898
|
| 50 |
+
},
|
| 51 |
+
"gu": {
|
| 52 |
+
"tokens": 75126588.0,
|
| 53 |
+
"words": 44453889.0,
|
| 54 |
+
"lines": 2626542.0,
|
| 55 |
+
"tokens_per_word": 1.6899891030906204,
|
| 56 |
+
"avg_tokens_per_line": 28.602850439855903
|
| 57 |
+
},
|
| 58 |
+
"bn": {
|
| 59 |
+
"tokens": 66371126.0,
|
| 60 |
+
"words": 39086017.0,
|
| 61 |
+
"lines": 2441433.0,
|
| 62 |
+
"tokens_per_word": 1.6980785225570567,
|
| 63 |
+
"avg_tokens_per_line": 27.18531534553682
|
| 64 |
+
},
|
| 65 |
+
"mr": {
|
| 66 |
+
"tokens": 61685551.0,
|
| 67 |
+
"words": 38015683.0,
|
| 68 |
+
"lines": 2256325.0,
|
| 69 |
+
"tokens_per_word": 1.6226342954301256,
|
| 70 |
+
"avg_tokens_per_line": 27.338947625009695
|
| 71 |
+
},
|
| 72 |
+
"or": {
|
| 73 |
+
"tokens": 38244695.0,
|
| 74 |
+
"words": 19515379.0,
|
| 75 |
+
"lines": 1241428.0,
|
| 76 |
+
"tokens_per_word": 1.9597208437509719,
|
| 77 |
+
"avg_tokens_per_line": 30.807018208063617
|
| 78 |
+
},
|
| 79 |
+
"as": {
|
| 80 |
+
"tokens": 23542984.0,
|
| 81 |
+
"words": 12977320.0,
|
| 82 |
+
"lines": 785702.0,
|
| 83 |
+
"tokens_per_word": 1.8141637872842775,
|
| 84 |
+
"avg_tokens_per_line": 29.96426635034657
|
| 85 |
+
}
|
| 86 |
+
}
|