Create tokenizer_config.json

tokenizer_config.json

@@ -0,0 +1,37 @@
+{
+  "tokenizer_class": "Video2ReactionTextTokenizer",
+  "tokenizer_config": {
+    "special_tokens": {
+      "<s>": 0,
+      "</s>": 1,
+      "<unk>": 2,
+      "<pad>": 3,
+      "<mask>": 4
+    },
+    "tokenizer_parameters": {
+      "tokenizer_model": "BPE",
+      "add_prefix_space": false,
+      "trim_offsets": true,
+      "model_max_length": 512,
+      "tokenization_method": "ByteLevel",
+      "tokenizer_version": "1.0.0"
+    },
+    "training_parameters": {
+      "vocab_size": 30000,
+      "min_frequency": 2,
+      "special_tokens_count": 5,
+      "max_length_single_sentence": 512,
+      "max_length_pair": 512,
+      "pad_token_id": 3,
+      "unk_token_id": 2,
+      "bos_token_id": 0,
+      "eos_token_id": 1
+    },
+    "model_information": {
+      "author": "Szilárd",
+      "description": "Custom tokenizer for Video2ReactionText model",
+      "license": "MIT",
+      "version": "1.0.0"
+    }
+  }
+}