Upload best_models.json with huggingface_hub
Browse files- best_models.json +74 -0
best_models.json
ADDED
|
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bpe": {
|
| 3 |
+
"best_run": "v8000_mf2",
|
| 4 |
+
"out_dir": "results\\bpe\\v8000_mf2",
|
| 5 |
+
"metrics": {
|
| 6 |
+
"oov_rate": 0.0,
|
| 7 |
+
"avg_sequence_length": 96.0113,
|
| 8 |
+
"avg_processing_time_ms": 0.19588143825531007,
|
| 9 |
+
"compression_ratio": 96.0113,
|
| 10 |
+
"total_tokens_evaluated": 1920226,
|
| 11 |
+
"unk_count": 0,
|
| 12 |
+
"train_time_s": 105.87230825424194,
|
| 13 |
+
"config": {
|
| 14 |
+
"vocab_size": 8000,
|
| 15 |
+
"min_frequency": 2,
|
| 16 |
+
"continuing_subword_prefix": "##"
|
| 17 |
+
}
|
| 18 |
+
},
|
| 19 |
+
"score": 63.712785797135034
|
| 20 |
+
},
|
| 21 |
+
"wordpiece": {
|
| 22 |
+
"best_run": "v8000_mf1",
|
| 23 |
+
"out_dir": "results\\wordpiece\\v8000_mf1",
|
| 24 |
+
"metrics": {
|
| 25 |
+
"oov_rate": 0.0,
|
| 26 |
+
"avg_sequence_length": 95.39795,
|
| 27 |
+
"avg_processing_time_ms": 31.364226222038273,
|
| 28 |
+
"compression_ratio": 95.39795,
|
| 29 |
+
"total_tokens_evaluated": 1907959,
|
| 30 |
+
"unk_count": 0,
|
| 31 |
+
"train_time_s": 124.3489019870758,
|
| 32 |
+
"config": {
|
| 33 |
+
"vocab_size": 8000,
|
| 34 |
+
"min_frequency": 1
|
| 35 |
+
}
|
| 36 |
+
},
|
| 37 |
+
"score": 63.20955201220989
|
| 38 |
+
},
|
| 39 |
+
"unigram": {
|
| 40 |
+
"best_run": "v16000",
|
| 41 |
+
"out_dir": "results\\unigram\\v16000",
|
| 42 |
+
"metrics": {
|
| 43 |
+
"oov_rate": 0.0,
|
| 44 |
+
"avg_sequence_length": 90.8909,
|
| 45 |
+
"avg_processing_time_ms": 0.29166127443313594,
|
| 46 |
+
"compression_ratio": 90.8909,
|
| 47 |
+
"total_tokens_evaluated": 1817818,
|
| 48 |
+
"unk_count": 0,
|
| 49 |
+
"train_time_s": 614.1360929012299,
|
| 50 |
+
"config": {
|
| 51 |
+
"vocab_size": 16000
|
| 52 |
+
}
|
| 53 |
+
},
|
| 54 |
+
"score": 60.91625533579062
|
| 55 |
+
},
|
| 56 |
+
"spm": {
|
| 57 |
+
"best_run": "v32000",
|
| 58 |
+
"out_dir": "results\\spm_unigram\\v32000",
|
| 59 |
+
"metrics": {
|
| 60 |
+
"oov_rate": 0.0,
|
| 61 |
+
"avg_sequence_length": 86.6945,
|
| 62 |
+
"avg_processing_time_ms": 0.1026016116142273,
|
| 63 |
+
"compression_ratio": 86.6945,
|
| 64 |
+
"total_tokens_evaluated": 1733890,
|
| 65 |
+
"unk_count": 0,
|
| 66 |
+
"unk_piece_used": "[UNK]",
|
| 67 |
+
"train_time_s": 249.83488726615906,
|
| 68 |
+
"config": {
|
| 69 |
+
"vocab_size": 32000
|
| 70 |
+
}
|
| 71 |
+
},
|
| 72 |
+
"score": 61.78699439108904
|
| 73 |
+
}
|
| 74 |
+
}
|