black fmt

Author: novaturient95

sdk/python/foundation-models/system/reinforcement-learning/reinforcement-learning.ipynb

@@ -325,33 +325,32 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "eval_job, status = (\n",
-    "    run_evaluation_pipeline(  # Function which invokes the model evaluation pipeline.\n",
-    "        ml_client=ml_client,\n",
-    "        registry_ml_client=registry_ml_client,\n",
-    "        compute_cluster=\"k8s-a100-compute\",\n",
-    "        grpo_model_dir=grpo_registered_model.path,  # Output from GPRO RL provided as data asset created from earlier step.\n",
-    "        rlpp_model_dir=rlpp_registered_model.path,  # Output from Reinforce_plus_plus RL provided as data asset created from earlier step.\n",
-    "        validation_dataset_path=test_data_path,  # Path to test dataset\n",
-    "        run_config={\n",
-    "            \"num_nodes\": 1,  # Number of nodes to be used for evaluation run.\n",
-    "            \"number_of_gpu_to_use\": 8,  # Number of GPUs in a node to be used for evaluation run.\n",
-    "            \"base_path_1_label\": \"GRPO\",  # Label to identify GRPO model outputs.\n",
-    "            \"base_path_2_label\": \"RLPP\",  # Label to identify RLPP model outputs.\n",
-    "            \"explore_pattern_1\": \"global_step_{checkpoint}/actor/lora_adapter/\",\n",
-    "            \"explore_pattern_2\": \"global_step_{checkpoint}/actor/lora_adapter/\",\n",
-    "            \"checkpoint_values_1\": \"12\",\n",
-    "            \"checkpoint_values_2\": \"12\",\n",
-    "            \"use_lora_adapters_1\": True,\n",
-    "            \"use_lora_adapters_2\": True,\n",
-    "            \"evaluate_base_model\": True,  # Set to True to evaluate base model along with RL finetuned models.\n",
-    "            \"hf_model_id\": \"deepseek-ai/DeepSeek-R1-Distill-Qwen-7B\",  # Huggingface ID of the base model\n",
-    "            \"max_prompt_length\": 8196,\n",
-    "            \"max_response_length\": 1024,\n",
-    "            \"dtype\": \"bfloat16\",\n",
-    "            \"tensor_parallel_size\": 4,\n",
-    "        },  # Configuration parameters for evaluation run.\n",
-    "    )\n",
+    "# Function which invokes the model evaluation pipeline.\n",
+    "eval_job, status = run_evaluation_pipeline(\n",
+    "    ml_client=ml_client,\n",
+    "    registry_ml_client=registry_ml_client,\n",
+    "    compute_cluster=\"k8s-a100-compute\",\n",
+    "    grpo_model_dir=grpo_registered_model.path,  # Output from GPRO RL provided as data asset created from earlier step.\n",
+    "    rlpp_model_dir=rlpp_registered_model.path,  # Output from Reinforce_plus_plus RL provided as data asset created from earlier step.\n",
+    "    validation_dataset_path=test_data_path,  # Path to test dataset\n",
+    "    run_config={\n",
+    "        \"num_nodes\": 1,  # Number of nodes to be used for evaluation run.\n",
+    "        \"number_of_gpu_to_use\": 8,  # Number of GPUs in a node to be used for evaluation run.\n",
+    "        \"base_path_1_label\": \"GRPO\",  # Label to identify GRPO model outputs.\n",
+    "        \"base_path_2_label\": \"RLPP\",  # Label to identify RLPP model outputs.\n",
+    "        \"explore_pattern_1\": \"global_step_{checkpoint}/actor/lora_adapter/\",\n",
+    "        \"explore_pattern_2\": \"global_step_{checkpoint}/actor/lora_adapter/\",\n",
+    "        \"checkpoint_values_1\": \"12\",\n",
+    "        \"checkpoint_values_2\": \"12\",\n",
+    "        \"use_lora_adapters_1\": True,\n",
+    "        \"use_lora_adapters_2\": True,\n",
+    "        \"evaluate_base_model\": True,  # Set to True to evaluate base model along with RL finetuned models.\n",
+    "        \"hf_model_id\": \"deepseek-ai/DeepSeek-R1-Distill-Qwen-7B\",  # Huggingface ID of the base model\n",
+    "        \"max_prompt_length\": 8196,\n",
+    "        \"max_response_length\": 1024,\n",
+    "        \"dtype\": \"bfloat16\",\n",
+    "        \"tensor_parallel_size\": 4,\n",
+    "    },  # Configuration parameters for evaluation run.\n",
     ")"
    ]
   },