fix(config): missing config not being documented and fix model_ override (#2317)

* fix(config): missing config not being documented and fix model_ space override

* fix: delete redundant field

docs/config.qmd

@@ -46,6 +46,10 @@ overrides_of_model_config:
     type: # linear | dynamic
     factor: # float
 
+# optional overrides the base model loading from_pretrained
+overrides_of_model_kwargs:
+  # use_cache: False
+
 # optional overrides to the bnb 4bit quantization configuration
 # https://huggingface.co/docs/transformers/main/main_classes/quantization#transformers.BitsAndBytesConfig
 bnb_config_kwargs:

src/axolotl/utils/config/models/input/v0_4_1/__init__.py

@@ -115,6 +115,9 @@ class RemappedParameters(BaseModel):
     overrides_of_model_config: Optional[Dict[str, Any]] = Field(
         default=None, alias="model_config"
     )
+    overrides_of_model_kwargs: Optional[Dict[str, Any]] = Field(
+        default=None, alias="model_kwargs"
+    )
     type_of_model: Optional[str] = Field(default=None, alias="model_type")
     revision_of_model: Optional[str] = Field(default=None, alias="model_revision")
 
@@ -426,8 +429,6 @@ class ModelInputConfig(BaseModel):
     )
     trust_remote_code: Optional[bool] = None
 
-    model_kwargs: Optional[Dict[str, Any]] = None
-
     @field_validator("trust_remote_code")
     @classmethod
     def hint_trust_remote_code(cls, trust_remote_code):

src/axolotl/utils/models.py

@@ -357,8 +357,8 @@ class ModelLoader:
 
         # init model kwargs
         self.model_kwargs: Dict[str, Any] = {}
-        if cfg.model_kwargs:
-            for key, val in cfg.model_kwargs.items():
+        if cfg.overrides_of_model_kwargs:
+            for key, val in cfg.overrides_of_model_kwargs.items():
                 self.model_kwargs[key] = val
 
         # init model