Alexandre/add base model option

cohere/client.py

@@ -1082,6 +1082,7 @@ def create_custom_model(
         name: str,
         model_type: CUSTOM_MODEL_TYPE,
         dataset: CustomModelDataset,
+        base_model: Optional[str] = None,
         hyperparameters: Optional[HyperParametersInput] = None,
     ) -> CustomModel:
         """Create a new custom model
@@ -1090,6 +1091,11 @@ def create_custom_model(
             name (str): name of your custom model, has to be unique across your organization
             model_type (GENERATIVE, CLASSIFY, RERANK): type of custom model
             dataset (InMemoryDataset, CsvDataset, JsonlDataset, TextDataset): A dataset for your training. Consists of a train and optional eval file.
+            base_model (str): base model to use for your custom model.
+                For generative and classify models, `base_model` has to be None (no option available for now)
+                For rerank models, you can choose between `english` and `multilingual`. Defaults to `english` if not specified.
+                    The English model is better for English, while the multilingual model should be picked if a non-negligible part of queries/documents
+                    will be in other languages
             hyperparameters (HyperParametersInput): adjust hyperparameters for your custom model. Only for generative custom models.
         Returns:
             str: the id of the custom model that was created
@@ -1112,12 +1118,13 @@ def create_custom_model(
 
         """
         internal_custom_model_type = CUSTOM_MODEL_PRODUCT_MAPPING[model_type]
+
         json = {
             "name": name,
             "settings": {
                 "trainFiles": [],
                 "evalFiles": [],
-                "baseModel": "medium",
+                "baseModel": base_model,
                 "finetuneType": internal_custom_model_type,
             },
         }

cohere/client_async.py

@@ -826,6 +826,7 @@ async def create_custom_model(
         name: str,
         model_type: CUSTOM_MODEL_TYPE,
         dataset: CustomModelDataset,
+        base_model: Optional[str] = None,
         hyperparameters: Optional[HyperParametersInput] = None,
     ) -> AsyncCustomModel:
         """Create a new custom model
@@ -834,6 +835,11 @@ async def create_custom_model(
             name (str): name of your custom model, has to be unique across your organization
             model_type (GENERATIVE, CLASSIFY, RERANK): type of custom model
             dataset (InMemoryDataset, CsvDataset, JsonlDataset, TextDataset): A dataset for your training. Consists of a train and optional eval file.
+            base_model (str): base model to use for your custom model.
+                For generative and classify models, `base_model` has to be None (no option available for now)
+                For rerank models, you can choose between `english` and `multilingual`. Defaults to `english` if not specified.
+                    The English model is better for English, while the multilingual model should be picked if a non-negligible part of queries/documents
+                    will be in other languages
             hyperparameters (HyperParametersInput): adjust hyperparameters for your custom model. Only for generative custom models.
         Returns:
             str: the id of the custom model that was created
@@ -856,12 +862,13 @@ async def create_custom_model(
 
         """
         internal_custom_model_type = CUSTOM_MODEL_PRODUCT_MAPPING[model_type]
+
         json = {
             "name": name,
             "settings": {
                 "trainFiles": [],
                 "evalFiles": [],
-                "baseModel": "medium",
+                "baseModel": base_model,
                 "finetuneType": internal_custom_model_type,
             },
         }

cohere/responses/custom_model.py

@@ -85,6 +85,7 @@ def __init__(
         model_type: CUSTOM_MODEL_TYPE,
         created_at: datetime,
         completed_at: Optional[datetime],
+        base_model: Optional[str] = None,
         model_id: Optional[str] = None,
         hyperparameters: Optional[HyperParameters] = None,
     ) -> None:
@@ -95,6 +96,7 @@ def __init__(
         self.model_type = model_type
         self.created_at = created_at
         self.completed_at = completed_at
+        self.base_model = base_model
         self.model_id = model_id
         self.hyperparameters = hyperparameters
         self._wait_fn = wait_fn
@@ -109,6 +111,7 @@ def from_dict(cls, data: Dict[str, Any], wait_fn) -> "BaseCustomModel":
             model_type=REVERSE_CUSTOM_MODEL_PRODUCT_MAPPING[data["settings"]["finetuneType"]],
             created_at=_parse_date(data["created_at"]),
             completed_at=_parse_date(data["completed_at"]) if "completed_at" in data else None,
+            base_model=data["settings"]["baseModel"],
             model_id=data["model"]["route"] if "model" in data else None,
             hyperparameters=HyperParameters.from_response(data["settings"]["hyperparameters"])
             if data["settings"]["hyperparameters"]