Transformers
English
Sanskrit
tokenizer
sentencepiece
unigram
english
math
custom_code
mume-tokenizer-32k / eval /verification.json
ks2812's picture
v0.1.0: spm_shared_unigram_32000_v1
e6d904a verified
Raw History Blame Contribute Delete
18 kB
{
"transformers": "5.18.0",
"tokenizers": "0.23.2",
"sentencepiece": "0.2.2",
"autotokenizer_trust_remote_code": "MumeTokenizer",
"autotokenizer_default": "TokenizersBackend",
"sets": {
"fineweb_val": {
"group": "english",
"records": 14983,
"words": 7770317,
"utf8_bytes": 46515251,
"tokens": 11831667,
"tokens_per_word": 1.5227,
"bytes_per_token": 3.9314,
"ids_identical_exact": 14983,
"ids_identical_fast": 14959,
"roundtrip_exact": 14983,
"roundtrip_spm": 14983,
"roundtrip_fast": 14983,
"fast_differs_at": [
210,
492,
699,
1298,
3648
],
"gpt2_tokens": 10470777,
"gpt2_tokens_per_word": 1.3475
},
"fineweb_val_clean": {
"group": "english",
"records": 11048,
"words": 4782318,
"utf8_bytes": 28445808,
"tokens": 7177975,
"tokens_per_word": 1.5009,
"bytes_per_token": 3.9629,
"ids_identical_exact": 11048,
"ids_identical_fast": 11041,
"roundtrip_exact": 11048,
"roundtrip_spm": 11048,
"roundtrip_fast": 11048,
"fast_differs_at": [
155,
2866,
3549,
4095,
7837
],
"gpt2_tokens": 6392995,
"gpt2_tokens_per_word": 1.3368
},
"wikitext103_test": {
"group": "english",
"records": 62,
"words": 241211,
"utf8_bytes": 1287656,
"tokens": 363932,
"tokens_per_word": 1.5088,
"bytes_per_token": 3.5382,
"ids_identical_exact": 62,
"ids_identical_fast": 62,
"roundtrip_exact": 62,
"roundtrip_spm": 62,
"roundtrip_fast": 62,
"fast_differs_at": [],
"gpt2_tokens": 283287,
"gpt2_tokens_per_word": 1.1744
},
"enwik8_test": {
"group": "english",
"records": 1,
"words": 648280,
"utf8_bytes": 5000000,
"tokens": 1717932,
"tokens_per_word": 2.65,
"bytes_per_token": 2.9105,
"ids_identical_exact": 1,
"ids_identical_fast": 0,
"roundtrip_exact": 1,
"roundtrip_spm": 1,
"roundtrip_fast": 1,
"fast_differs_at": [
0
],
"gpt2_tokens": 1480280,
"gpt2_tokens_per_word": 2.2834
},
"text8_test": {
"group": "english",
"records": 1,
"words": 855233,
"utf8_bytes": 5000000,
"tokens": 1145389,
"tokens_per_word": 1.3393,
"bytes_per_token": 4.3653,
"ids_identical_exact": 1,
"ids_identical_fast": 0,
"roundtrip_exact": 1,
"roundtrip_spm": 1,
"roundtrip_fast": 1,
"fast_differs_at": [
0
],
"gpt2_tokens": 1003900,
"gpt2_tokens_per_word": 1.1738
},
"owm_val": {
"group": "math",
"records": 2704,
"words": 3157209,
"utf8_bytes": 20001168,
"tokens": 6124528,
"tokens_per_word": 1.9399,
"bytes_per_token": 3.2657,
"ids_identical_exact": 2704,
"ids_identical_fast": 2670,
"roundtrip_exact": 2703,
"roundtrip_spm": 2703,
"roundtrip_fast": 2703,
"fast_differs_at": [
13,
81,
82,
297,
431
],
"gpt2_tokens": 5540067,
"gpt2_tokens_per_word": 1.7547
},
"owm_val_clean": {
"group": "math",
"records": 1126,
"words": 768667,
"utf8_bytes": 4780645,
"tokens": 1436589,
"tokens_per_word": 1.8689,
"bytes_per_token": 3.3278,
"ids_identical_exact": 1126,
"ids_identical_fast": 1119,
"roundtrip_exact": 1126,
"roundtrip_spm": 1126,
"roundtrip_fast": 1126,
"fast_differs_at": [
7,
36,
235,
313,
547
],
"gpt2_tokens": 1307204,
"gpt2_tokens_per_word": 1.7006
},
"gsm8k_test": {
"group": "math",
"records": 1319,
"words": 130627,
"utf8_bytes": 652230,
"tokens": 224032,
"tokens_per_word": 1.7151,
"bytes_per_token": 2.9113,
"ids_identical_exact": 1319,
"ids_identical_fast": 1319,
"roundtrip_exact": 1319,
"roundtrip_spm": 1319,
"roundtrip_fast": 1319,
"fast_differs_at": [],
"gpt2_tokens": 173987,
"gpt2_tokens_per_word": 1.3319
},
"math_test": {
"group": "math",
"records": 5000,
"words": 577396,
"utf8_bytes": 3565468,
"tokens": 1531141,
"tokens_per_word": 2.6518,
"bytes_per_token": 2.3286,
"ids_identical_exact": 5000,
"ids_identical_fast": 4997,
"roundtrip_exact": 5000,
"roundtrip_spm": 5000,
"roundtrip_fast": 5000,
"fast_differs_at": [
180,
510,
1083
],
"gpt2_tokens": 1457380,
"gpt2_tokens_per_word": 2.5241
},
"math_test_clean": {
"group": "math",
"records": 3754,
"words": 393713,
"utf8_bytes": 2418527,
"tokens": 1014818,
"tokens_per_word": 2.5776,
"bytes_per_token": 2.3832,
"ids_identical_exact": 3754,
"ids_identical_fast": 3751,
"roundtrip_exact": 3754,
"roundtrip_spm": 3754,
"roundtrip_fast": 3754,
"fast_differs_at": [
140,
393,
838
],
"gpt2_tokens": 958028,
"gpt2_tokens_per_word": 2.4333
},
"sanskrit_dcs_gold_slp1": {
"group": "sanskrit",
"records": 3000,
"words": 16596,
"utf8_bytes": 139276,
"tokens": 39143,
"tokens_per_word": 2.3586,
"bytes_per_token": 3.5581,
"ids_identical_exact": 3000,
"ids_identical_fast": 3000,
"roundtrip_exact": 3000,
"roundtrip_spm": 3000,
"roundtrip_fast": 3000,
"fast_differs_at": []
},
"sanskrit_dcs_gold_devanagari": {
"group": "sanskrit",
"records": 3000,
"words": 16596,
"utf8_bytes": 370794,
"tokens": 164598,
"tokens_per_word": 9.9179,
"bytes_per_token": 2.2527,
"ids_identical_exact": 3000,
"ids_identical_fast": 3000,
"roundtrip_exact": 3000,
"roundtrip_spm": 3000,
"roundtrip_fast": 3000,
"fast_differs_at": []
},
"sanskrit_gita_slp1": {
"group": "sanskrit",
"records": 700,
"words": 7814,
"utf8_bytes": 67003,
"tokens": 18460,
"tokens_per_word": 2.3624,
"bytes_per_token": 3.6296,
"ids_identical_exact": 700,
"ids_identical_fast": 700,
"roundtrip_exact": 700,
"roundtrip_spm": 700,
"roundtrip_fast": 700,
"fast_differs_at": []
},
"sanskrit_gita_devanagari": {
"group": "sanskrit",
"records": 700,
"words": 7814,
"utf8_bytes": 166845,
"tokens": 72265,
"tokens_per_word": 9.2481,
"bytes_per_token": 2.3088,
"ids_identical_exact": 700,
"ids_identical_fast": 700,
"roundtrip_exact": 700,
"roundtrip_spm": 700,
"roundtrip_fast": 700,
"fast_differs_at": []
},
"sanskrit_ood_slp1": {
"group": "sanskrit",
"records": 2000,
"words": 132896,
"utf8_bytes": 1220801,
"tokens": 347994,
"tokens_per_word": 2.6185,
"bytes_per_token": 3.5081,
"ids_identical_exact": 2000,
"ids_identical_fast": 2000,
"roundtrip_exact": 2000,
"roundtrip_spm": 2000,
"roundtrip_fast": 2000,
"fast_differs_at": []
},
"sanskrit_ood_devanagari": {
"group": "sanskrit",
"records": 2000,
"words": 132896,
"utf8_bytes": 2921785,
"tokens": 1288809,
"tokens_per_word": 9.6979,
"bytes_per_token": 2.267,
"ids_identical_exact": 2000,
"ids_identical_fast": 2000,
"roundtrip_exact": 2000,
"roundtrip_spm": 2000,
"roundtrip_fast": 2000,
"fast_differs_at": []
},
"sanskrit_prose_slp1": {
"group": "sanskrit",
"records": 2470,
"words": 55243,
"utf8_bytes": 420458,
"tokens": 118897,
"tokens_per_word": 2.1523,
"bytes_per_token": 3.5363,
"ids_identical_exact": 2470,
"ids_identical_fast": 2469,
"roundtrip_exact": 2470,
"roundtrip_spm": 2470,
"roundtrip_fast": 2470,
"fast_differs_at": [
1495
]
},
"sanskrit_prose_devanagari": {
"group": "sanskrit",
"records": 2470,
"words": 55243,
"utf8_bytes": 1014824,
"tokens": 464012,
"tokens_per_word": 8.3995,
"bytes_per_token": 2.1871,
"ids_identical_exact": 2470,
"ids_identical_fast": 2470,
"roundtrip_exact": 2470,
"roundtrip_spm": 2470,
"roundtrip_fast": 2470,
"fast_differs_at": []
},
"sanskrit_vedic_slp1": {
"group": "sanskrit",
"records": 1000,
"words": 4079,
"utf8_bytes": 29125,
"tokens": 13336,
"tokens_per_word": 3.2694,
"bytes_per_token": 2.1839,
"ids_identical_exact": 1000,
"ids_identical_fast": 1000,
"roundtrip_exact": 1000,
"roundtrip_spm": 1000,
"roundtrip_fast": 1000,
"fast_differs_at": []
},
"sanskrit_vedic_devanagari": {
"group": "sanskrit",
"records": 1000,
"words": 4079,
"utf8_bytes": 77606,
"tokens": 41505,
"tokens_per_word": 10.1753,
"bytes_per_token": 1.8698,
"ids_identical_exact": 1000,
"ids_identical_fast": 1000,
"roundtrip_exact": 1000,
"roundtrip_spm": 1000,
"roundtrip_fast": 1000,
"fast_differs_at": []
}
},
"edge_cases": [
{
"text": "",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": " ",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": " ",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "a",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": " a",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "a ",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": " two leading spaces",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "trailing spaces ",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "tab\tseparated\tvalues",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "line one\nline two\n\nafter a blank line\n",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "\n",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "\r\n windows line ends \r\n",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "2026",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "x2y",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "3.14159",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "1,234,567.89 and -42",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "$12 for 3 x 4 = 12",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "Let $f(x) = x^2 + 1$. Then \\frac{a}{b} \\geq \\sqrt{2}",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "\\begin{align} a &= b \\\\ c &= d \\end{align}",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "</s> <s> <unk> literal special-token text",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "<0x41>",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "▁ already has a meta-symbol",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": false
},
{
"text": "Ünïcödé café naïve façade",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "日本語のテキスト",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "emoji 🙂👍🏽",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "Ελληνικά and русский",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "संस्कृतं नाम दैवी वाक् । ॥ १२३ ॥",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "saMskftaM nAma dEvI vAk",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "zero​width",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "combining é vs é",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": " no-break space",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "word word word word word word word word word word word word ",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "012345678901234567890123456789012345678901234567890123456789",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
},
{
"text": "''' \"\"\" ``` ~~~",
"ids_equal_exact": true,
"ids_equal_fast": true,
"decode_equal_spm_decode_exact": true,
"decode_equal_spm_decode_fast": true,
"spm_roundtrip": true
}
],
"edge_cases_ids_equal_exact": "35/35",
"edge_cases_ids_equal_fast": "35/35",
"edge_cases_decode_equal_spm_decode_exact": "35/35",
"edge_cases_decode_equal_spm_decode_fast": "35/35",
"edge_cases_spm_roundtrip": "34/35",
"all_texts": 58338,
"all_ids_identical_exact": "58338/58338",
"all_ids_identical_fast": "58257/58338",
"seconds": 99.9
}