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": 0.40473279904504394,
|
| 12 |
+
"old": 0.2955810913595943,
|
| 13 |
+
"eng": 0.1000143045303409,
|
| 14 |
+
"spa": 0.10001878784659393,
|
| 15 |
+
"code": 0.09965301721842694,
|
| 16 |
+
"exotic": 0.7003138904046382
|
| 17 |
+
},
|
| 18 |
+
"bytes": {
|
| 19 |
+
"modern": 25006263,
|
| 20 |
+
"old": 18262366,
|
| 21 |
+
"eng": 6179346,
|
| 22 |
+
"spa": 6179623,
|
| 23 |
+
"code": 6157024,
|
| 24 |
+
"exotic": 43268629
|
| 25 |
+
},
|
| 26 |
+
"total_bytes": 61784622,
|
| 27 |
+
"vocab_target": 256000,
|
| 28 |
+
"min_frequency": 2,
|
| 29 |
+
"special_token_count": 100,
|
| 30 |
+
"language_token_count": 66,
|
| 31 |
+
"external_to_exotic_ratio": 0.14285714285714288
|
| 32 |
+
}
|