feat: use model name as deployment name (#62)

Author: oleksii-donets

src/main/java/com/epam/aidial/deployment/manager/service/manifest/InferenceManifestGenerator.java

@@ -24,6 +24,8 @@
 @LogExecution
 public class InferenceManifestGenerator extends DeployableManifestGenerator {
 
+    private static final String MODEL_NAME_ARGUMENT_NAME = "--model_name";
+
     public InferenceManifestGenerator(AppProperties appconfig) {
         super(appconfig);
     }
@@ -72,6 +74,11 @@ public InferenceService serviceConfig(
             modelChain.get(InferenceMappers.MODEL_ARGS_FIELD).data().addAll(args);
         }
 
+        // Explicitly set model name to ensure the model uses the intended name.
+        // If omitted, the inference service will default to the Kubernetes service name,
+        // which may differ from the actual model name due to naming transformations.
+        modelChain.get(InferenceMappers.MODEL_ARGS_FIELD).data().addAll(List.of(MODEL_NAME_ARGUMENT_NAME, name));
+
         var envListMapper = modelChain
                 .getList(InferenceMappers.MODEL_ENV_FIELD, InferenceMappers.ENV_VAR_NAME);
         applySimpleEnvs(envListMapper, envs, Env::setValue);

src/test/resources/manifest/inference_service_with_args.json

@@ -11,7 +11,9 @@
           "--arg1",
           "value1",
           "--arg2",
-          "value2"
+          "value2",
+          "--model_name",
+          "args-inference-app"
         ],
         "env": [],
         "modelFormat": {

src/test/resources/manifest/inference_service_with_envs.json

@@ -7,6 +7,10 @@
   "spec": {
     "predictor": {
       "model": {
+        "args": [
+          "--model_name",
+          "basic-inference-app"
+        ],
         "env": [
           {
             "name": "SIMPLE_VAR",

src/test/resources/manifest/inference_service_with_port.json

@@ -7,6 +7,10 @@
   "spec": {
     "predictor": {
       "model": {
+        "args": [
+          "--model_name",
+          "custom-port-inference-app"
+        ],
         "env": [],
         "modelFormat": {
           "name": "huggingface"

src/test/resources/manifest/inference_service_with_resources.json

@@ -7,6 +7,10 @@
   "spec": {
     "predictor": {
       "model": {
+        "args": [
+          "--model_name",
+          "resource-inference-app"
+        ],
         "env": [],
         "modelFormat": {
           "name": "huggingface"

src/test/resources/manifest/inference_service_with_scaling.json

@@ -9,6 +9,10 @@
       "maxReplicas": 5,
       "minReplicas": 1,
       "model": {
+        "args": [
+          "--model_name",
+          "scaling-inference-app"
+        ],
         "env": [],
         "modelFormat": {
           "name": "huggingface"