fix dependcy + logg info (#148)

model_server/app/main.py

@@ -59,6 +59,7 @@ async def models():
 
 @app.post("/embeddings")
 async def embedding(req: EmbeddingRequest, res: Response):
+    logger.info(f"Embedding req: {req}")
     if req.model != transformers["model_name"]:
         raise HTTPException(status_code=400, detail="unknown model: " + req.model)
 
@@ -70,7 +71,7 @@ async def embedding(req: EmbeddingRequest, res: Response):
     embeddings = embeddings[0][:, 0]
     # normalize embeddings
     embeddings = torch.nn.functional.normalize(embeddings, p=2, dim=1).detach().numpy()
-    print(f"Embedding Call Complete Time: {time.time()-start}")
+    logger.info(f"Embedding Call Complete Time: {time.time()-start}")
     data = []
 
     for embedding in embeddings.tolist():

model_server/pyproject.toml

@@ -22,12 +22,13 @@ pyyaml = "6.0.2"
 accelerate = "*"
 psutil = "6.0.0"
 optimum-intel = "*"
-openvino = "*"
+openvino = "2024.4.0"
 pandas = "*"
 dateparser = "*"
 openai = "1.50.2"
 tf-keras = "*"
-onnx = "*"
+onnx = "1.17.0"
+onnxruntime = "1.19.2"
 
 [tool.poetry.scripts]
 archgw_modelserver = "app:run_server"

model_server/requirements.txt

@@ -16,4 +16,5 @@ openai==1.50.2
 pandas
 tf-keras
 onnx==1.17.0
+onnxruntime==1.19.2
 pytest