Configure custom GPT for AutoModel loading

__init__.py

@@ -0,0 +1,10 @@
+# __init__.py
+
+from .model import GPTConfig, GPT
+
+try:
+    from transformers import AutoConfig, AutoModel
+    AutoConfig.register("gpt", GPTConfig)
+    AutoModel.register(GPTConfig, GPT)
+except ImportError:
+    pass

config.json

@@ -6,7 +6,12 @@
   "n_layer": 6,
   "n_head": 6,
   "n_embd": 384,
+  "dropout": 0.0,
+  "bias": true,
   "auto_map": {
+    "AutoConfig": "model.GPTConfig",
     "AutoModel": "model.GPT"
-  }
-}
+  },
+  "torch_dtype": "float32",
+  "transformers_version": "4.42.0"
+}