Upload morphel_metrics_report.json with huggingface_hub
Browse files- morphel_metrics_report.json +54 -0
morphel_metrics_report.json
ADDED
|
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"tokenizer": "MorpheL",
|
| 3 |
+
"config": {
|
| 4 |
+
"vocab_size": 8097,
|
| 5 |
+
"top_k": 4,
|
| 6 |
+
"temperature": 1.0,
|
| 7 |
+
"mi_threshold": 0.0,
|
| 8 |
+
"min_frequency": 2,
|
| 9 |
+
"special_tokens": [
|
| 10 |
+
"<s>",
|
| 11 |
+
"<pad>",
|
| 12 |
+
"</s>",
|
| 13 |
+
"<unk>",
|
| 14 |
+
"<mask>"
|
| 15 |
+
],
|
| 16 |
+
"dataset": "facebook/xnli/tr",
|
| 17 |
+
"corpus_lines": 800404,
|
| 18 |
+
"mi_index_words": 97975
|
| 19 |
+
},
|
| 20 |
+
"metrics": {
|
| 21 |
+
"fertility": 1.8496005424179818,
|
| 22 |
+
"tokens_per_char": 0.24694732958592314,
|
| 23 |
+
"avg_seq_len": 21.2776,
|
| 24 |
+
"vocab_coverage": 1.0,
|
| 25 |
+
"oov_rate": 0.0,
|
| 26 |
+
"fallback_event_rate": 0.2401350927803399,
|
| 27 |
+
"char_shatter_rate": 0.0,
|
| 28 |
+
"n_sentences": 15000,
|
| 29 |
+
"n_word_types": 26764
|
| 30 |
+
},
|
| 31 |
+
"baselines": {
|
| 32 |
+
"BPE": {
|
| 33 |
+
"fertility": null,
|
| 34 |
+
"tokens_per_char": null,
|
| 35 |
+
"avg_seq_len": null,
|
| 36 |
+
"vocab_coverage": null,
|
| 37 |
+
"oov_rate": null
|
| 38 |
+
},
|
| 39 |
+
"WordPiece": {
|
| 40 |
+
"fertility": null,
|
| 41 |
+
"tokens_per_char": null,
|
| 42 |
+
"avg_seq_len": null,
|
| 43 |
+
"vocab_coverage": null,
|
| 44 |
+
"oov_rate": null
|
| 45 |
+
},
|
| 46 |
+
"Unigram": {
|
| 47 |
+
"fertility": null,
|
| 48 |
+
"tokens_per_char": null,
|
| 49 |
+
"avg_seq_len": null,
|
| 50 |
+
"vocab_coverage": null,
|
| 51 |
+
"oov_rate": null
|
| 52 |
+
}
|
| 53 |
+
}
|
| 54 |
+
}
|