fixed padding token for finetuning

config.json

@@ -19,7 +19,7 @@
   "num_attention_heads": 32,
   "num_hidden_layers": 40,
   "num_key_value_heads": 8,
-  "pad_token_id": 0,
+  "pad_token_id": 49152,
   "residual_multiplier": 0.22,
   "rms_norm_eps": 1e-05,
   "rope_scaling": null,

generation_config.json

@@ -2,6 +2,6 @@
   "_from_model_config": true,
   "bos_token_id": 0,
   "eos_token_id": 0,
-  "pad_token_id": 0,
+  "pad_token_id": 49152,
   "transformers_version": "4.46.0"
 }

special_tokens_map.json

@@ -18,7 +18,8 @@
     "<commit_before>",
     "<commit_msg>",
     "<commit_after>",
-    "<reponame>"
+    "<reponame>",
+    "<|finetune_left_pad_id|>"
   ],
   "bos_token": {
     "content": "<|endoftext|>",
@@ -35,7 +36,7 @@
     "single_word": false
   },
   "pad_token": {
-    "content": "<|endoftext|>",
+    "content": "<|finetune_left_pad_id|>",
     "lstrip": false,
     "normalized": false,
     "rstrip": false,

tokenizer.json

@@ -173,6 +173,15 @@
       "rstrip": false,
       "normalized": false,
       "special": true
+    },
+    {
+      "id": 49152,
+      "content": "<|finetune_left_pad_id|>",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": false,
+      "special": true
     }
   ],
   "normalizer": null,
@@ -49359,7 +49368,8 @@
       "loot": 49148,
       "mpath": 49149,
       "ĠSIP": 49150,
-      "getOptions": 49151
+      "getOptions": 49151,
+      "<|finetune_left_pad_id|>":49152
     },
     "merges": [
       "Ġ Ġ",

tokenizer_config.json

@@ -152,6 +152,14 @@
       "rstrip": false,
       "single_word": false,
       "special": true
+    },
+    "49152": {
+      "content": "<|finetune_left_pad_id|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
     }
   },
   "additional_special_tokens": [
@@ -173,15 +181,16 @@
     "<commit_before>",
     "<commit_msg>",
     "<commit_after>",
-    "<reponame>"
+    "<reponame>",
+    "<|finetune_left_pad_id|>"
   ],
   "bos_token": "<|endoftext|>",
   "clean_up_tokenization_spaces": true,
   "eos_token": "<|endoftext|>",
   "model_max_length": 9223372036854775807,
-  "pad_token": "<|endoftext|>",
+  "pad_token": "<|finetune_left_pad_id|>",
   "padding_side": "left",
   "tokenizer_class": "GPT2Tokenizer",
   "unk_token": "<|endoftext|>",
-  "vocab_size": 49152
+  "vocab_size": 49153
 }