Upload train_aviation.py with huggingface_hub

train_aviation.py

@@ -40,28 +40,28 @@ from trl import SFTTrainer, SFTConfig
 from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig, AutoConfig
 
 # Register 'ministral3' config to handle nested text_config
-print("🔧 Registering ministral3 config...")
+print("🔧 Registering ministral3 config (Monkey Patch Strategy)...")
 try:
-    # Use standard MistralConfig (not MinistralConfig) as base
-    from transformers import MistralConfig, MistralModel, AutoModel, AutoConfig
-    class Ministral3CompatConfig(MistralConfig):
-        model_type = "ministral3"
+    from transformers import MinistralConfig, AutoConfig
     
-    # 1. Register Config
-    AutoConfig.register("ministral3", Ministral3CompatConfig)
-    print("   Registered ministral3 -> Ministral3CompatConfig (subclass of MistralConfig)")
-
-    # 2. Register Model for this config
-    AutoModel.register(Ministral3CompatConfig, MistralModel)
-    print("   Registered Ministral3CompatConfig -> MistralModel")
+    # Monkey patch the model_type to match what the config.json has
+    # This allows us to use the native class which is already registered with AutoModel
+    print(f"   Original MinistralConfig.model_type: {MinistralConfig.model_type}")
+    MinistralConfig.model_type = "ministral3"
+    print(f"   Patched MinistralConfig.model_type: {MinistralConfig.model_type}")
+    
+    # Register the patched class for the "ministral3" key
+    AutoConfig.register("ministral3", MinistralConfig)
+    print("   Registered ministral3 -> MinistralConfig (native, patched)")
 
 except Exception as e:
-    print(f"   ❌ Failed to register ministral3 config/model: {e}")
+    print(f"   ❌ Failed to patch/register ministral3 config: {e}")
 
 # Register Mistral3Config to a model class
 # ... (rest of registration kept as is)
 # ... (rest of registration kept as is)
 # ... (rest of registration kept as is)
+# ... (rest of registration kept as is)
 print("🔧 Registering Mistral3 model class...")
 try:
     from transformers.models.mistral3.configuration_mistral3 import Mistral3Config