Set model_garden_source_model_name for hexllm deployments.

PiperOrigin-RevId: 720341876

notebooks/community/model_garden/model_garden_codegemma_deployment_on_vertex.ipynb

@@ -320,6 +320,8 @@
         "def deploy_model_hexllm(\n",
         "    model_name: str,\n",
         "    model_id: str,\n",
+        "    publisher: str,\n",
+        "    publisher_model_id: str,\n",
         "    service_account: str,\n",
         "    base_model_id: str = None,\n",
         "    data_parallel_size: int = 1,\n",
@@ -399,6 +401,9 @@
         "        serving_container_shared_memory_size_mb=(16 * 1024),  # 16 GB\n",
         "        serving_container_deployment_timeout=7200,\n",
         "        location=TPU_DEPLOYMENT_REGION,\n",
+        "        model_garden_source_model_name=(\n",
+        "            f\"publishers/{publisher}/models/{publisher_model_id}\"\n",
+        "        ),\n",
         "    )\n",
         "\n",
         "    model.deploy(\n",
@@ -419,6 +424,8 @@
         "models[\"hexllm_tpu\"], endpoints[\"hexllm_tpu\"] = deploy_model_hexllm(\n",
         "    model_name=common_util.get_job_name_with_datetime(prefix=MODEL_ID),\n",
         "    model_id=model_id,\n",
+        "    publisher=\"google\",\n",
+        "    publisher_model_id=\"codegemma\",\n",
         "    base_model_id=hf_model_id,\n",
         "    service_account=SERVICE_ACCOUNT,\n",
         "    machine_type=machine_type,\n",

notebooks/community/model_garden/model_garden_gemma2_deployment_on_vertex.ipynb

@@ -283,6 +283,8 @@
         "def deploy_model_hexllm(\n",
         "    model_name: str,\n",
         "    model_id: str,\n",
+        "    publisher: str,\n",
+        "    publisher_model_id: str,\n",
         "    service_account: str,\n",
         "    base_model_id: str = None,\n",
         "    data_parallel_size: int = 1,\n",
@@ -362,6 +364,9 @@
         "        serving_container_shared_memory_size_mb=(16 * 1024),  # 16 GB\n",
         "        serving_container_deployment_timeout=7200,\n",
         "        location=TPU_DEPLOYMENT_REGION,\n",
+        "        model_garden_source_model_name=(\n",
+        "            f\"publishers/{publisher}/models/{publisher_model_id}\"\n",
+        "        ),\n",
         "    )\n",
         "\n",
         "    model.deploy(\n",
@@ -382,6 +387,8 @@
         "models[\"hexllm_tpu\"], endpoints[\"hexllm_tpu\"] = deploy_model_hexllm(\n",
         "    model_name=common_util.get_job_name_with_datetime(prefix=MODEL_ID),\n",
         "    model_id=model_id,\n",
+        "    publisher=\"google\",\n",
+        "    publisher_model_id=\"gemma2\",\n",
         "    service_account=SERVICE_ACCOUNT,\n",
         "    machine_type=machine_type,\n",
         "    tensor_parallel_size=tensor_parallel_size,\n",

notebooks/community/model_garden/model_garden_gemma_deployment_on_vertex.ipynb

@@ -308,6 +308,8 @@
         "def deploy_model_hexllm(\n",
         "    model_name: str,\n",
         "    model_id: str,\n",
+        "    publisher: str,\n",
+        "    publisher_model_id: str,\n",
         "    service_account: str,\n",
         "    base_model_id: str = None,\n",
         "    data_parallel_size: int = 1,\n",
@@ -387,6 +389,9 @@
         "        serving_container_shared_memory_size_mb=(16 * 1024),  # 16 GB\n",
         "        serving_container_deployment_timeout=7200,\n",
         "        location=TPU_DEPLOYMENT_REGION,\n",
+        "        model_garden_source_model_name=(\n",
+        "            f\"publishers/{publisher}/models/{publisher_model_id}\"\n",
+        "        ),\n",
         "    )\n",
         "\n",
         "    model.deploy(\n",
@@ -407,6 +412,8 @@
         "models[\"hexllm_tpu\"], endpoints[\"hexllm_tpu\"] = deploy_model_hexllm(\n",
         "    model_name=common_util.get_job_name_with_datetime(prefix=MODEL_ID),\n",
         "    model_id=model_id,\n",
+        "    publisher=\"google\",\n",
+        "    publisher_model_id=\"gemma\",\n",
         "    base_model_id=hf_model_id,\n",
         "    service_account=SERVICE_ACCOUNT,\n",
         "    machine_type=machine_type,\n",

notebooks/community/model_garden/model_garden_gemma_finetuning_on_vertex.ipynb

@@ -1216,6 +1216,8 @@
         "def deploy_model_hexllm(\n",
         "    model_name: str,\n",
         "    model_id: str,\n",
+        "    publisher: str,\n",
+        "    publisher_model_id: str,\n",
         "    service_account: str,\n",
         "    base_model_id: str = None,\n",
         "    data_parallel_size: int = 1,\n",
@@ -1295,6 +1297,9 @@
         "        serving_container_shared_memory_size_mb=(16 * 1024),  # 16 GB\n",
         "        serving_container_deployment_timeout=7200,\n",
         "        location=TPU_DEPLOYMENT_REGION,\n",
+        "        model_garden_source_model_name=(\n",
+        "            f\"publishers/{publisher}/models/{publisher_model_id}\"\n",
+        "        ),\n",
         "    )\n",
         "\n",
         "    model.deploy(\n",
@@ -1351,9 +1356,11 @@
         "    print(\"Using model from: \", output_folder)\n",
         "    models[\"hexllm_tpu\"], endpoints[\"hexllm_tpu\"] = deploy_model_hexllm(\n",
         "        model_name=common_util.get_job_name_with_datetime(prefix=\"gemma-serve-hexllm\"),\n",
-        "        base_model_id=f\"google/{KAGGLE_MODEL_ID}\",\n",
         "        model_id=output_folder,\n",
+        "        publisher=\"google\",\n",
+        "        publisher_model_id=\"gemma\",\n",
         "        service_account=SERVICE_ACCOUNT,\n",
+        "        base_model_id=f\"google/{KAGGLE_MODEL_ID}\",\n",
         "        machine_type=machine_type,\n",
         "        max_num_batched_tokens=max_num_batched_tokens,\n",
         "        tokens_pad_multiple=tokens_pad_multiple,\n",

notebooks/community/model_garden/model_garden_phi3_deployment.ipynb

@@ -644,6 +644,8 @@
         "def deploy_model_hexllm(\n",
         "    model_name: str,\n",
         "    model_id: str,\n",
+        "    publisher: str,\n",
+        "    publisher_model_id: str,\n",
         "    service_account: str,\n",
         "    base_model_id: str = None,\n",
         "    data_parallel_size: int = 1,\n",
@@ -723,6 +725,9 @@
         "        serving_container_shared_memory_size_mb=(16 * 1024),  # 16 GB\n",
         "        serving_container_deployment_timeout=7200,\n",
         "        location=TPU_DEPLOYMENT_REGION,\n",
+        "        model_garden_source_model_name=(\n",
+        "            f\"publishers/{publisher}/models/{publisher_model_id}\"\n",
+        "        ),\n",
         "    )\n",
         "\n",
         "    model.deploy(\n",
@@ -747,6 +752,8 @@
         "models[\"hexllm_tpu\"], endpoints[\"hexllm_tpu\"] = deploy_model_hexllm(\n",
         "    model_name=common_util.get_job_name_with_datetime(prefix=MODEL_ID),\n",
         "    model_id=model_id,\n",
+        "    publisher=\"microsoft\",\n",
+        "    publisher_model_id=\"phi3\",\n",
         "    service_account=SERVICE_ACCOUNT,\n",
         "    tensor_parallel_size=tensor_parallel_size,\n",
         "    machine_type=machine_type,\n",

notebooks/community/model_garden/model_garden_pytorch_llama3_1_deployment.ipynb

@@ -313,6 +313,8 @@
         "def deploy_model_hexllm(\n",
         "    model_name: str,\n",
         "    model_id: str,\n",
+        "    publisher: str,\n",
+        "    publisher_model_id: str,\n",
         "    service_account: str,\n",
         "    base_model_id: str = None,\n",
         "    data_parallel_size: int = 1,\n",
@@ -392,6 +394,9 @@
         "        serving_container_shared_memory_size_mb=(16 * 1024),  # 16 GB\n",
         "        serving_container_deployment_timeout=7200,\n",
         "        location=TPU_DEPLOYMENT_REGION,\n",
+        "        model_garden_source_model_name=(\n",
+        "            f\"publishers/{publisher}/models/{publisher_model_id}\"\n",
+        "        ),\n",
         "    )\n",
         "\n",
         "    model.deploy(\n",
@@ -412,6 +417,8 @@
         "models[\"hexllm_tpu\"], endpoints[\"hexllm_tpu\"] = deploy_model_hexllm(\n",
         "    model_name=common_util.get_job_name_with_datetime(prefix=MODEL_ID),\n",
         "    model_id=model_id,\n",
+        "    publisher=\"meta\",\n",
+        "    publisher_model_id=\"llama3_1\",\n",
         "    service_account=SERVICE_ACCOUNT,\n",
         "    base_model_id=hf_model_id,\n",
         "    tensor_parallel_size=tensor_parallel_size,\n",

notebooks/community/model_garden/model_garden_pytorch_llama3_2_deployment.ipynb

@@ -304,6 +304,8 @@
         "def deploy_model_hexllm(\n",
         "    model_name: str,\n",
         "    model_id: str,\n",
+        "    publisher: str,\n",
+        "    publisher_model_id: str,\n",
         "    service_account: str,\n",
         "    base_model_id: str = None,\n",
         "    data_parallel_size: int = 1,\n",
@@ -383,6 +385,9 @@
         "        serving_container_shared_memory_size_mb=(16 * 1024),  # 16 GB\n",
         "        serving_container_deployment_timeout=7200,\n",
         "        location=TPU_DEPLOYMENT_REGION,\n",
+        "        model_garden_source_model_name=(\n",
+        "            f\"publishers/{publisher}/models/{publisher_model_id}\"\n",
+        "        ),\n",
         "    )\n",
         "\n",
         "    model.deploy(\n",
@@ -403,6 +408,8 @@
         "models[\"hexllm_tpu\"], endpoints[\"hexllm_tpu\"] = deploy_model_hexllm(\n",
         "    model_name=common_util.get_job_name_with_datetime(prefix=\"llama3_2-serve-hexllm\"),\n",
         "    model_id=model_id,\n",
+        "    publisher=\"meta\",\n",
+        "    publisher_model_id=\"llama3-2\",\n",
         "    service_account=SERVICE_ACCOUNT,\n",
         "    base_model_id=hf_model_id,\n",
         "    tensor_parallel_size=tensor_parallel_size,\n",

notebooks/community/model_garden/model_garden_pytorch_qwen2_deployment.ipynb

@@ -571,6 +571,8 @@
         "def deploy_model_hexllm(\n",
         "    model_name: str,\n",
         "    model_id: str,\n",
+        "    publisher: str,\n",
+        "    publisher_model_id: str,\n",
         "    service_account: str,\n",
         "    base_model_id: str = None,\n",
         "    data_parallel_size: int = 1,\n",
@@ -650,6 +652,9 @@
         "        serving_container_shared_memory_size_mb=(16 * 1024),  # 16 GB\n",
         "        serving_container_deployment_timeout=7200,\n",
         "        location=TPU_DEPLOYMENT_REGION,\n",
+        "        model_garden_source_model_name=(\n",
+        "            f\"publishers/{publisher}/models/{publisher_model_id}\"\n",
+        "        ),\n",
         "    )\n",
         "\n",
         "    model.deploy(\n",
@@ -670,6 +675,8 @@
         "models[\"hexllm_tpu\"], endpoints[\"hexllm_tpu\"] = deploy_model_hexllm(\n",
         "    model_name=common_util.get_job_name_with_datetime(prefix=MODEL_ID),\n",
         "    model_id=model_id,\n",
+        "    publisher=\"qwen\",\n",
+        "    publisher_model_id=\"qwen2\",\n",
         "    service_account=SERVICE_ACCOUNT,\n",
         "    tensor_parallel_size=tpu_count,\n",
         "    machine_type=machine_type,\n",