chore: add end-to-end identity model

Signed-off-by: Guan Luo <[email protected]>

Author: GuanLuo

lib/bindings/python/tests/test_tensor.py

@@ -22,8 +22,8 @@ async def test_register(runtime: DistributedRuntime):
 
     model_config = {
         "name": "tensor",
-        "inputs": [{"name": "input", "data_type": "Bool", "shape": [1]}],
-        "outputs": [],
+        "inputs": [{"name": "input", "data_type": "Int32", "shape": [-1]}],
+        "outputs": [{"name": "output", "data_type": "Int32", "shape": [-1]}],
     }
     runtime_config = ModelRuntimeConfig()
     runtime_config.set_tensor_model_config(model_config)
@@ -47,7 +47,8 @@ async def test_register(runtime: DistributedRuntime):
 
 
 async def generate(request, context):
-    raise NotImplementedError
+    print(f"Received request: {request}")
+    yield {"model": request["model"], "tensors": request["tensors"]}
 
 
 if __name__ == "__main__":

lib/llm/tests/kserve_service.rs

@@ -284,15 +284,53 @@ pub mod kserve_test {
         manager
             .add_completions_model("split", split.clone())
             .unwrap();
+        manager
+            .save_model_entry("split", ModelEntry {
+            name: "split".to_string(),
+            endpoint_id: EndpointId {
+                namespace: "namespace".to_string(),
+                component: "component".to_string(),
+                name: "split".to_string(),
+            },
+            model_type: ModelType::Completions,
+            model_input: ModelInput::Text,
+            runtime_config: None,
+        });
+
         manager
             .add_chat_completions_model("failure", failure.clone())
             .unwrap();
         manager
             .add_completions_model("failure", failure.clone())
             .unwrap();
+        manager
+            .save_model_entry("failure", ModelEntry {
+            name: "failure".to_string(),
+            endpoint_id: EndpointId {
+                namespace: "namespace".to_string(),
+                component: "component".to_string(),
+                name: "failure".to_string(),
+            },
+            model_type: ModelType::Completions | ModelType::Chat,
+            model_input: ModelInput::Text,
+            runtime_config: None,
+        });
         manager
             .add_completions_model("long_running", long_running.clone())
             .unwrap();
+        manager
+            .save_model_entry("failure", ModelEntry {
+            name: "failure".to_string(),
+            endpoint_id: EndpointId {
+                namespace: "namespace".to_string(),
+                component: "component".to_string(),
+                name: "failure".to_string(),
+            },
+            model_type: ModelType::Completions,
+            model_input: ModelInput::Text,
+            runtime_config: None,
+        });
+
 
         (service, split, failure, long_running)
     }