VER5 — CPU merge full model + base tokenizer (regex fix ready)

finetune_config.json

@@ -14,6 +14,5 @@
   "lora_dropout": 0.1,
   "max_len": 512,
   "dataset": "data/sql2nl_rag_hybrid_final_100ea.json",
-  "schema_dir": "data/bird/BIRD_SQLite",
-  "notes": "CPU merge; save base tokenizer to avoid regex/tokenizer mismatch"
+  "schema_dir": "data/bird/BIRD_SQLite"
 }

model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0c983db988d6ea8e1ce285abdb1a5cfbc109b4b54eb966abfb6efd65958da96f
+oid sha256:4038dda123153c929dfeac80ee81b1eab6652988ab4209a8164e990fee8ef7b3
 size 4965799096

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2dbb010ae27795cfd6d5deb00dd7436e810d77bf3c900fe0f8e8d6cb3119fda6
+oid sha256:66b996d36a1a86c85153d5cd251ae389067b4cceec65855d242441de9c988d1e
 size 1459729952

tokenizer.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b
-size 17209920
+oid sha256:c33847c2b3ac674227d54e9cdae85d500e7634c104440a5955a3fdef6372f3cb
+size 17210037

tokenizer_config.json

@@ -2053,6 +2053,7 @@
   "clean_up_tokenization_spaces": true,
   "eos_token": "<|eot_id|>",
   "extra_special_tokens": {},
+  "fix_mistral_regex": true,
   "model_input_names": [
     "input_ids",
     "attention_mask"