fix some config issues uncovered by placing configs in the correct dir

Signed-off-by: Terry Kong <terryk@nvidia.com>

Author: terrykong

examples/configs/recipes/llm/grpo-deepscaler-1.5b-24K.yaml

@@ -45,9 +45,6 @@ policy:
       gpu_memory_utilization: 0.8
       enforce_eager: True
       max_model_len: ${policy.max_total_sequence_length}
-      # For most cases, use "dummy" to load the initial weights, since they will be overwritten during refit
-      # For Gemma models, we need to use "auto" due to a vllm bug
-      load_format: dummy
 
 cluster:
   gpus_per_node: 8

examples/configs/recipes/llm/grpo-deepscaler-1.5b-8K.yaml

@@ -103,9 +103,6 @@ policy:
       gpu_memory_utilization: 0.6
       max_model_len: ${policy.max_total_sequence_length}
       enforce_eager: True
-      # For most cases, use "dummy" to load the initial weights, since they will be overwritten during refit
-      # For Gemma models, we need to use "auto" due to a vllm bug
-      load_format: dummy
     colocated:
       # true: generation shares training GPUs
       # false: uses dedicated generation resources

tests/test_suites/nightly.txt

@@ -13,6 +13,11 @@ tests/test_suites/llm/grpo-qwen2.5-7b-instruct-4n8g-fsdp2tp4sp.v3.sh
 # Functional 32b run
 tests/test_suites/llm/grpo-qwen2.5-32b-32n8g-fsdp2tp8sp-actckpt.v3.sh
 
+# Deepscaler (short tests)
+tests/test_suites/llm/grpo-deepscaler-1.5b-16K.sh
+tests/test_suites/llm/grpo-deepscaler-1.5b-24K.sh
+tests/test_suites/llm/grpo-deepscaler-1.5b-8K.sh
+
 #######
 # SFT #
 #######

tests/unit/test_recipes_and_test_suites.py

@@ -283,6 +283,8 @@ def test_all_recipes_can_merge_configs_with_base_config(
 ):
     from omegaconf import OmegaConf
 
+    from nemo_rl.utils.config import load_config
+
     base_yaml = os.path.join(project_root, algo_base_yaml)
     base_config = OmegaConf.load(base_yaml)
     # Would result in an error if we couldn't merge our config with the recipe's config
@@ -293,7 +295,7 @@ def test_all_recipes_can_merge_configs_with_base_config(
             #  test_all_recipes_start_with_algo_hyphen()
             continue
         recipe_yaml_path = os.path.join(recipes_dir, recipe_yaml)
-        recipe_config = OmegaConf.load(recipe_yaml_path)
+        recipe_config = load_config(recipe_yaml_path)
         OmegaConf.set_struct(recipe_config, True)
         # This will raise a error if the config can't be merged
         print(f"Merging {recipe_yaml} with {base_yaml}")