Upload recipe/corpus_stats.json with huggingface_hub

recipe/corpus_stats.json

@@ -0,0 +1,32 @@
+{
+  "recipe_version": "hf-bytelevel-bpe-v2-iterator-eval",
+  "test_run": false,
+  "weights_target": {
+    "exotic": 0.7,
+    "eng": 0.1,
+    "spa": 0.1,
+    "code": 0.1
+  },
+  "weights_actual": {
+    "modern": 0.34789646559900145,
+    "old": 0.3521215369891937,
+    "eng": 0.0999832425438067,
+    "spa": 0.09999948567782432,
+    "code": 0.09999926919017384,
+    "exotic": 0.7000180025881951
+  },
+  "bytes": {
+    "modern": 159093371,
+    "old": 161025500,
+    "eng": 45722428,
+    "spa": 45729856,
+    "code": 45729757,
+    "exotic": 320118871
+  },
+  "total_bytes": 457300912,
+  "vocab_target": 256000,
+  "min_frequency": 2,
+  "special_token_count": 282,
+  "language_token_count": 248,
+  "external_to_exotic_ratio": 0.14285714285714288
+}