Merged GPT4-V into GPT4-Turbo to accommodate OpenAI's new API change.

GaiZhenbiao · Apr 17, 2024 · 7231bea · 7231bea
1 parent 5881220
commit 7231bea
Show file tree

Hide file tree

Showing 3 changed files with 15 additions and 9 deletions.
diff --git a/modules/models/OpenAIVision.py b/modules/models/OpenAIVision.py
@@ -133,7 +133,9 @@ def _get_gpt4v_style_history(self):
                         content.append(
                             {
                                 "type": "image_url",
-                                "image_url": f"data:image/{self.get_image_type(image)};base64,{self.get_base64_image(image)}"
+                                "image_url": {
+                                    "url": f"data:image/{self.get_image_type(image)};base64,{self.get_base64_image(image)}",
+                                }
                             },
                         )
                 if content:

diff --git a/modules/models/base_model.py b/modules/models/base_model.py
@@ -161,12 +161,16 @@ def get_type(cls, model_name: str):
         model_type = None
         model_name_lower = model_name.lower()
         if "gpt" in model_name_lower:
-            if "instruct" in model_name_lower:
-                model_type = ModelType.OpenAIInstruct
-            elif "vision" in model_name_lower:
+            try:
+                assert MODEL_METADATA[model_name]["multimodal"] == True
                 model_type = ModelType.OpenAIVision
-            else:
-                model_type = ModelType.OpenAI
+            except:
+                if "instruct" in model_name_lower:
+                    model_type = ModelType.OpenAIInstruct
+                elif "vision" in model_name_lower:
+                    model_type = ModelType.OpenAIVision
+                else:
+                    model_type = ModelType.OpenAI
         elif "chatglm" in model_name_lower:
             model_type = ModelType.ChatGLM
         elif "ollama" in model_name_lower:

diff --git a/modules/presets.py b/modules/presets.py
@@ -62,7 +62,6 @@
     "GPT4",
     "GPT4 32K",
     "GPT4 Turbo",
-    "GPT4 Vision",
     "Claude 3 Haiku",
     "Claude 3 Sonnet",
     "Claude 3 Opus",
@@ -155,11 +154,12 @@
         "token_limit": 32768,
     },
     "GPT4 Turbo": {
-        "model_name": "gpt-4-turbo-preview",
+        "model_name": "gpt-4-turbo",
         "token_limit": 128000,
+        "multimodal": True
     },
     "GPT4 Vision": {
-        "model_name": "gpt-4-vision-preview",
+        "model_name": "gpt-4-turbo",
         "token_limit": 128000,
         "multimodal": True
     },