CRAYON-tokenizer / benchmark_results.json
Phase-Technologies's picture
Upload folder using huggingface_hub
708f4a3 verified
{
"date": "2026-02-02T21:46:22.756992",
"test_text_bytes": 30800,
"iterations": 10,
"results": [
{
"name": "CRAYON (CPU - code)",
"status": "OK",
"vocab_size": "~250k",
"avg_tokens": 30800.0,
"token_count": 30800,
"load_time_ms": 128.97940003313124,
"avg_time_ms": 1.2961800443008542,
"min_time_ms": 1.007900107651949,
"max_time_ms": 2.303199842572212,
"tokens_per_sec": 23762130.990539353,
"mb_per_sec": 22.661334028758386
},
{
"name": "CRAYON (CPU - science)",
"status": "OK",
"vocab_size": "~250k",
"avg_tokens": 24900.0,
"token_count": 24900,
"load_time_ms": 3.807599889114499,
"avg_time_ms": 1.3703399803489447,
"min_time_ms": 0.9711999446153641,
"max_time_ms": 2.43859994225204,
"tokens_per_sec": 18170673.232243754,
"mb_per_sec": 21.43494998798246
},
{
"name": "CRAYON (CPU - lite)",
"status": "OK",
"vocab_size": "50k",
"avg_tokens": 15700.0,
"token_count": 15700,
"load_time_ms": 20.62970004044473,
"avg_time_ms": 1.5809000004082918,
"min_time_ms": 1.2891001533716917,
"max_time_ms": 1.9415998831391335,
"tokens_per_sec": 9931051.929878697,
"mb_per_sec": 18.580029690509473
},
{
"name": "tiktoken (p50k/GPT-3)",
"status": "OK",
"vocab_size": 50000,
"avg_tokens": 11900.0,
"token_count": 11900,
"load_time_ms": 0.008899951353669167,
"avg_time_ms": 28.1568999402225,
"min_time_ms": 21.030299831181765,
"max_time_ms": 55.71989994496107,
"tokens_per_sec": 422631.7536825385,
"mb_per_sec": 1.0431961262664624
},
{
"name": "tiktoken (cl100k/GPT-4)",
"status": "OK",
"vocab_size": 100000,
"avg_tokens": 9000.0,
"token_count": 9000,
"load_time_ms": 0.011600088328123093,
"avg_time_ms": 23.468929948285222,
"min_time_ms": 20.06639982573688,
"max_time_ms": 35.85169999860227,
"tokens_per_sec": 383485.74135386146,
"mb_per_sec": 1.2515768298783763
},
{
"name": "HF T5 (SentencePiece)",
"status": "OK",
"vocab_size": 32000,
"avg_tokens": 12601.0,
"token_count": 12601,
"load_time_ms": 1777.7703001629561,
"avg_time_ms": 32.928459998220205,
"min_time_ms": 32.26630017161369,
"max_time_ms": 34.046499989926815,
"tokens_per_sec": 382678.08457125194,
"mb_per_sec": 0.8920298412649765
},
{
"name": "HF LLaMA (SP-BPE)",
"status": "OK",
"vocab_size": 32000,
"avg_tokens": 11401.0,
"token_count": 11401,
"load_time_ms": 1174.7749999631196,
"avg_time_ms": 39.65424003545195,
"min_time_ms": 30.960500007495284,
"max_time_ms": 45.88270001113415,
"tokens_per_sec": 287510.2382445661,
"mb_per_sec": 0.7407321113467842
},
{
"name": "HF GPT-2 (BPE)",
"status": "OK",
"vocab_size": 50257,
"avg_tokens": 15700.0,
"token_count": 15700,
"load_time_ms": 1819.5615001022816,
"avg_time_ms": 73.55678000021726,
"min_time_ms": 61.30379997193813,
"max_time_ms": 98.4288000036031,
"tokens_per_sec": 213440.55571700702,
"mb_per_sec": 0.3993264651501295
},
{
"name": "HF BERT (WordPiece)",
"status": "OK",
"vocab_size": 30522,
"avg_tokens": 11402.0,
"token_count": 11402,
"load_time_ms": 1832.9594999086112,
"avg_time_ms": 58.81147999316454,
"min_time_ms": 50.545900128781796,
"max_time_ms": 68.34379979409277,
"tokens_per_sec": 193873.71311392295,
"mb_per_sec": 0.49944617868359115
}
]
}