LangChainLLMs cache_obj update (#508)

* [refactor]: Explicity input for cache_obj for langchain_models allowing for passing in custom cache_obj once rather than passing again and again during inference * [tests]: Updated tests for langchain_models
zilliztech · Aug 7, 2023 · fdf1630 · fdf1630
1 parent 4a63925
commit fdf1630
Show file tree

Hide file tree

Showing 2 changed files with 15 additions and 9 deletions.
diff --git a/gptcache/adapter/langchain_models.py b/gptcache/adapter/langchain_models.py
@@ -4,6 +4,7 @@
 from gptcache.manager.scalar_data.base import Answer, DataType
 from gptcache.session import Session
 from gptcache.utils import import_pydantic, import_langchain
+from gptcache.core import Cache,cache
 
 import_pydantic()
 import_langchain()
@@ -50,6 +51,7 @@ class LangChainLLMs(LLM, BaseModel):
 
     llm: Any
     session: Session = None
+    cache_obj: Cache = cache
     tmp_args: Any = None
 
     @property
@@ -80,6 +82,7 @@ def _call(
             _update_cache_callback,
             prompt=prompt,
             stop=stop,
+            cache_obj=self.cache_obj,
             session=session,
             **self.tmp_args,
         )
@@ -151,6 +154,7 @@ def _llm_type(self) -> str:
 
     chat: Any
     session: Session = None
+    cache_obj: Cache = cache
     tmp_args: Any = None
 
     def _generate(
@@ -170,6 +174,7 @@ def _generate(
             _update_cache_msg_callback,
             messages=messages,
             stop=stop,
+            cache_obj=self.cache_obj,
             session=session,
             run_manager=run_manager,
             **self.tmp_args,
@@ -193,6 +198,7 @@ async def _agenerate(
             _update_cache_msg_callback,
             messages=messages,
             stop=stop,
+            cache_obj=self.cache_obj,
             session=session,
             run_manager=run_manager,
             **self.tmp_args,

diff --git a/tests/unit_tests/adapter/test_langchain_models.py b/tests/unit_tests/adapter/test_langchain_models.py
@@ -30,7 +30,7 @@ def test_langchain_llms():
 
     os.environ["OPENAI_API_KEY"] = "API"
     langchain_openai = OpenAI(model_name="text-ada-001")
-    llm = LangChainLLMs(llm=langchain_openai)
+    llm = LangChainLLMs(llm=langchain_openai,cache_obj=llm_cache)
     assert str(langchain_openai) == str(llm)
 
     with patch("openai.Completion.create") as mock_create:
@@ -53,10 +53,10 @@ def test_langchain_llms():
               }
             }
 
-        answer = llm(prompt=question, cache_obj=llm_cache)
+        answer = llm(prompt=question)
         assert expect_answer == answer
 
-    answer = llm(prompt=question, cache_obj=llm_cache)
+    answer = llm(prompt=question)
     assert expect_answer == answer
 
 
@@ -77,7 +77,7 @@ def test_langchain_chats():
 
     os.environ["OPENAI_API_KEY"] = "API"
     langchain_openai = ChatOpenAI(temperature=0)
-    chat = LangChainChat(chat=langchain_openai)
+    chat = LangChainChat(chat=langchain_openai,cache_obj=llm_cache)
 
     assert chat.get_num_tokens("hello") == langchain_openai.get_num_tokens("hello")
     assert chat.get_num_tokens_from_messages(messages=[HumanMessage(content="test_langchain_chats")]) \
@@ -104,7 +104,7 @@ def test_langchain_chats():
             }
         }
 
-        answer = chat(messages=question, cache_obj=llm_cache)
+        answer = chat(messages=question)
         assert answer == _cache_msg_data_convert(msg).generations[0].message
 
     with patch("openai.ChatCompletion.acreate") as mock_create:
@@ -128,16 +128,16 @@ def test_langchain_chats():
             }
         }
 
-        answer = asyncio.run(chat.agenerate([question2], cache_obj=llm_cache))
+        answer = asyncio.run(chat.agenerate([question2]))
         assert answer.generations[0][0].text == _cache_msg_data_convert(msg).generations[0].text
 
-    answer = chat(messages=question, cache_obj=llm_cache)
+    answer = chat(messages=question)
     assert answer == _cache_msg_data_convert(msg).generations[0].message
 
-    answer = asyncio.run(chat.agenerate([question], cache_obj=llm_cache))
+    answer = asyncio.run(chat.agenerate([question]))
     assert answer.generations[0][0].text == _cache_msg_data_convert(msg).generations[0].text
 
-    answer = asyncio.run(chat.agenerate([question2], cache_obj=llm_cache))
+    answer = asyncio.run(chat.agenerate([question2]))
     assert answer.generations[0][0].text == _cache_msg_data_convert(msg).generations[0].text