Upload recipe/corpus_stats.json with huggingface_hub
Browse files- recipe/corpus_stats.json +32 -0
recipe/corpus_stats.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"recipe_version": "hf-bytelevel-bpe-v1",
|
| 3 |
+
"test_run": true,
|
| 4 |
+
"weights_target": {
|
| 5 |
+
"exotic": 0.7,
|
| 6 |
+
"eng": 0.1,
|
| 7 |
+
"spa": 0.1,
|
| 8 |
+
"code": 0.1
|
| 9 |
+
},
|
| 10 |
+
"weights_actual": {
|
| 11 |
+
"modern": 0.22659476422696032,
|
| 12 |
+
"old": 0.47387660725836755,
|
| 13 |
+
"eng": 0.10006682589867377,
|
| 14 |
+
"spa": 0.10003697198211248,
|
| 15 |
+
"code": 0.09942483063388591,
|
| 16 |
+
"exotic": 0.7004713714853279
|
| 17 |
+
},
|
| 18 |
+
"bytes": {
|
| 19 |
+
"modern": 25813993,
|
| 20 |
+
"old": 53984687,
|
| 21 |
+
"eng": 11399753,
|
| 22 |
+
"spa": 11396352,
|
| 23 |
+
"code": 11326616,
|
| 24 |
+
"exotic": 79798680
|
| 25 |
+
},
|
| 26 |
+
"total_bytes": 113921401,
|
| 27 |
+
"vocab_target": 256000,
|
| 28 |
+
"min_frequency": 2,
|
| 29 |
+
"special_token_count": 126,
|
| 30 |
+
"language_token_count": 92,
|
| 31 |
+
"external_to_exotic_ratio": 0.14285714285714288
|
| 32 |
+
}
|