Implemented suggestions from PR

.vscode/settings.json

@@ -2,5 +2,10 @@
     "[python]": {
         "editor.defaultFormatter": "ms-python.black-formatter"
     },
-    "python.formatting.provider": "none"
+    "python.formatting.provider": "none",
+    "python.testing.pytestArgs": [
+        "aana"
+    ],
+    "python.testing.unittestEnabled": false,
+    "python.testing.pytestEnabled": true
 }

README.md

@@ -30,7 +30,7 @@ sh install.sh
 5. Run the SDK.
 
 ```bash
-CUDA_VISIBLE_DEVICES=0 poetry run aana --port 8000 --host 0.0.0.0 -- target llama2
+CUDA_VISIBLE_DEVICES=0 poetry run aana --port 8000 --host 0.0.0.0 --target llama2
 ```
 
 The target parameter specifies the set of endpoints to deploy.

aana/api/api_generation.py

@@ -284,22 +284,14 @@ async def route_func_body(body: str, files: Optional[List[UploadFile]] = None):
             return AanaJSONResponse(content=output)
 
         if file_upload_field:
-
-            async def route_func_files(
-                body: str = Form(...),
-                files: List[UploadFile] = File(
-                    None, description=file_upload_field.description
-                ),
-            ):
-                return await route_func_body(body=body, files=files)
-
-            return route_func_files
+            files = File(None, description=file_upload_field.description)
         else:
+            files = None
 
-            async def route_func(body: str = Form(...)):
-                return await route_func_body(body=body)
+        async def route_func(body: str = Form(...), files=files):
+            return await route_func_body(body=body, files=files)
 
-            return route_func
+        return route_func
 
     def register(self, app: FastAPI, pipeline: Pipeline):
         """
@@ -344,6 +336,7 @@ def add_custom_schemas_to_openapi_schema(
 
     File upload is that FastAPI doesn't support Pydantic models in multipart requests.
     There is a discussion about it on FastAPI discussion forum.
+    See https://github.com/tiangolo/fastapi/discussions/8406
     The topic starter suggests a workaround.
     The workaround is to use Forms instead of Pydantic models in the endpoint definition and
     then convert the Forms to Pydantic models in the endpoint itself

aana/tests/utils.py

@@ -1,12 +1,11 @@
-import torch
-
-
 def is_gpu_available() -> bool:
     """
     Check if a GPU is available.
 
     Returns:
         bool: True if a GPU is available, False otherwise.
     """
+    import torch
 
+    # TODO: find the way to check if GPU is available without importing torch
     return torch.cuda.is_available()

startup.sh

@@ -1,2 +1,3 @@
 #!/bin/bash
+# TODO: pass arguments to the docker to set target instead of environment variable
 poetry run aana --port 8000 --host 0.0.0.0 --target $TARGET