Add madlad400 support.

src/config.py

@@ -5,7 +5,7 @@ from typing import List, Dict, Literal
 
 
 class ModelConfig:
-    def __init__(self, name: str, url: str, path: str = None, type: str = "whisper", tokenizer_url: str = None):
+    def __init__(self, name: str, url: str, path: str = None, type: str = "whisper", tokenizer_url: str = None, revision: str = None, model_file: str = None,):
         """
         Initialize a model configuration.
 
@@ -13,12 +13,19 @@ class ModelConfig:
         url: URL to download the model from
         path: Path to the model file. If not set, the model will be downloaded from the URL.
         type: Type of model. Can be whisper or huggingface.
+        revision: [by transformers] The specific model version to use.
+            It can be a branch name, a tag name, or a commit id, 
+            since we use a git-based system for storing models and other artifacts on huggingface.co, 
+            so revision can be any identifier allowed by git.
+        model_file: The name of the model file in repo or directory.[from marella/ctransformers] 
         """
         self.name = name
         self.url = url
         self.path = path
         self.type = type
         self.tokenizer_url = tokenizer_url
+        self.revision = revision
+        self.model_file = model_file
 
 VAD_INITIAL_PROMPT_MODE_VALUES=["prepend_all_segments", "prepend_first_segment", "json_prompt_mode"]
 
@@ -43,7 +50,7 @@ class VadInitialPromptMode(Enum):
             return None
 
 class ApplicationConfig:
-    def __init__(self, models: Dict[Literal["whisper", "m2m100", "nllb", "mt5", "ALMA"], List[ModelConfig]],
+    def __init__(self, models: Dict[Literal["whisper", "m2m100", "nllb", "mt5", "ALMA", "madlad400"], List[ModelConfig]],
                  input_audio_max_duration: int = 600, share: bool = False, server_name: str = None, server_port: int = 7860, 
                  queue_concurrency_count: int = 1, delete_uploaded_files: bool = True,
                  whisper_implementation: str = "whisper", default_model_name: str = "medium", 
@@ -169,7 +176,7 @@ class ApplicationConfig:
             # Load using json5
             data = json5.load(f)
             data_models = data.pop("models", [])
-            models: Dict[Literal["whisper", "m2m100", "nllb", "mt5", "ALMA"], List[ModelConfig]] = {
+            models: Dict[Literal["whisper", "m2m100", "nllb", "mt5", "ALMA", "madlad400"], List[ModelConfig]] = {
                 key: [ModelConfig(**item) for item in value]
                 for key, value in data_models.items()
             }