Merge pull request stanfordnlp#1638 from chenmoneygithub/fix-saving-arg

Fix saving/loading retriever

dspy/primitives/module.py

@@ -8,6 +8,7 @@
 # NOTE: Note: It's important (temporary decision) to maintain named_parameters that's different in behavior from
 # named_sub_modules for the time being.
 
+
 class BaseModule:
     def __init__(self):
         pass
@@ -29,7 +30,7 @@ def add_parameter(param_name, param_value):
                     visited.add(id(param_value))
                     param_name = postprocess_parameter_name(param_name, param_value)
                     named_parameters.append((param_name, param_value))
-            
+
             elif isinstance(param_value, dspy.Module):
                 # When a sub-module is pre-compiled, keep it frozen.
                 if not getattr(param_value, "_compiled", False):
@@ -42,7 +43,7 @@ def add_parameter(param_name, param_value):
         for name, value in self.__dict__.items():
             if isinstance(value, Parameter):
                 add_parameter(name, value)
-            
+
             elif isinstance(value, dspy.Module):
                 # When a sub-module is pre-compiled, keep it frozen.
                 if not getattr(value, "_compiled", False):
@@ -153,7 +154,11 @@ def dump_state(self, save_verbose):
 
     def load_state(self, state, use_legacy_loading=False):
         for name, param in self.named_parameters():
-            param.load_state(state[name], use_legacy_loading=use_legacy_loading)
+            if isinstance(param, BaseModule):
+                param.load_state(state[name], use_legacy_loading=use_legacy_loading)
+            else:
+                # `use_legacy_loading` is only applicable for BaseModule instances.
+                param.load_state(state[name])
 
     def save(self, path, save_field_meta=False):
         with open(path, "w") as f:
@@ -168,11 +173,11 @@ def postprocess_parameter_name(name, value):
     # For ChainOfThought backward compatibility, remove ending ._predict if it's there
     if name.endswith("._predict"):
         name = name[:-9]
-    
+
     if name.endswith(".self"):
         name = name[:-5]
-    
+
     if name == "_predict":
         return "self"
-    
-    return name
+
+    return name

tests/retrieve/test_llama_index_rm.py

@@ -21,7 +21,7 @@
 @pytest.fixture()
 def rag_setup() -> dict:
     """Builds the necessary fixtures to test LI"""
-    pytest.importorskip("llamaindex")
+    pytest.importorskip("llama_index")
     dataset = HotPotQA(train_seed=1, train_size=8, eval_seed=2023, dev_size=4, test_size=0)
     trainset = [x.with_inputs("question") for x in dataset.train]
     devset = [x.with_inputs("question") for x in dataset.dev]
@@ -46,7 +46,7 @@ def rag_setup() -> dict:
 
 def test_lirm_as_rm(rag_setup):
     """Test the retriever as retriever method"""
-    pytest.importorskip("llamaindex")
+    pytest.importorskip("llama_index")
     retriever = rag_setup.get("retriever")
     test_res_li = retriever.retrieve("At My Window was released by which American singer-songwriter?")
     rm = rag_setup.get("rm")
@@ -59,3 +59,25 @@ def test_lirm_as_rm(rag_setup):
     assert isinstance(test_res_dspy, list), "Ensuring the results are a list from the DSPy retriever"
 
     assert len(test_res_li) == len(test_res_dspy), "Rough equality check of the results"
+
+
+def test_save_load_llama_index_rag(rag_setup, tmp_path):
+    pytest.importorskip("llama_index")
+
+    class RAG(dspy.Module):
+        def __init__(self):
+            super().__init__()
+            self.retriever = dspy.Retrieve(k=3)
+            self.cot = dspy.ChainOfThought("question, context -> answer")
+
+    rag = RAG()
+    rag.retriever.k = 4
+
+    file_path = tmp_path / "rag.json"
+    rag.save(file_path)
+    loaded_rag = RAG()
+    # Before loading, the retriever k should be 3.
+    assert loaded_rag.retriever.k == 3
+    # After loading, the retriever k should be 4.
+    loaded_rag.load(file_path)
+    assert loaded_rag.retriever.k == 4