Migrate to the latest openAI version (#397)

* run migration * bump up versions * minor fixes and migrations * reset token count * minor fix * correct timeout error * fix prompt parser test
neulab · Apr 16, 2024 · 04debc9 · 04debc9
1 parent 25e0a96
commit 04debc9
Show file tree

Hide file tree

Showing 3 changed files with 19 additions and 20 deletions.
diff --git a/prompt2model/utils/api_tools.py b/prompt2model/utils/api_tools.py
@@ -10,7 +10,6 @@
 import aiolimiter
 import litellm.utils
 import openai
-import openai.error
 import tiktoken
 from aiohttp import ClientSession
 from litellm import acompletion, completion
@@ -19,22 +18,22 @@
 # Note that litellm converts all API errors into openai errors,
 # so openai errors are valid even when using other services.
 API_ERRORS = (
-    openai.error.APIError,
-    openai.error.Timeout,
-    openai.error.RateLimitError,
-    openai.error.ServiceUnavailableError,
-    openai.error.InvalidRequestError,
+    openai.APIError,
+    openai.APITimeoutError,
+    openai.RateLimitError,
+    openai.BadRequestError,
+    openai.APIStatusError,
     json.decoder.JSONDecodeError,
     AssertionError,
 )
 
 ERROR_ERRORS_TO_MESSAGES = {
-    openai.error.InvalidRequestError: "API Invalid Request: Prompt was filtered",
-    openai.error.RateLimitError: "API rate limit exceeded. Sleeping for 10 seconds.",
-    openai.error.APIConnectionError: "Error Communicating with API",
-    openai.error.Timeout: "API Timeout Error: API Timeout",
-    openai.error.ServiceUnavailableError: "API service unavailable error: {e}",
-    openai.error.APIError: "API error: {e}",
+    openai.BadRequestError: "API Invalid Request: Prompt was filtered",
+    openai.RateLimitError: "API rate limit exceeded. Sleeping for 10 seconds.",
+    openai.APIConnectionError: "Error Communicating with API",
+    openai.APITimeoutError: "API Timeout Error: API Timeout",
+    openai.APIStatusError: "API service unavailable error: {e}",
+    openai.APIError: "API error: {e}",
 }
 
 
@@ -170,14 +169,14 @@ async def _throttled_completion_acreate(
                         if isinstance(
                             e,
                             (
-                                openai.error.ServiceUnavailableError,
-                                openai.error.APIError,
+                                openai.APIStatusError,
+                                openai.APIError,
                             ),
                         ):
                             logging.warning(
                                 ERROR_ERRORS_TO_MESSAGES[type(e)].format(e=e)
                             )
-                        elif isinstance(e, openai.error.InvalidRequestError):
+                        elif isinstance(e, openai.BadRequestError):
                             logging.warning(ERROR_ERRORS_TO_MESSAGES[type(e)])
                             return {
                                 "choices": [
@@ -231,7 +230,7 @@ def handle_api_error(e) -> None:
         raise e
     if isinstance(
         e,
-        (openai.error.APIError, openai.error.Timeout, openai.error.RateLimitError),
+        (openai.APIError, openai.APITimeoutError, openai.RateLimitError),
     ):
         # For these errors, OpenAI recommends waiting before retrying.
         time.sleep(1)

diff --git a/pyproject.toml b/pyproject.toml
@@ -28,7 +28,7 @@ dependencies = [
     "gradio==3.38.0",
     "torch",
     "pytest",
-    "openai==0.27.10",
+    "openai",
     "sentencepiece",
     "bert_score",
     "sacrebleu",
@@ -45,7 +45,7 @@ dependencies = [
     "psutil",
     "protobuf",
     "nest-asyncio",
-    "litellm==0.1.583",
+    "litellm",
     "peft"
 ]
 

diff --git a/tests/prompt_parser_test.py b/tests/prompt_parser_test.py
@@ -136,7 +136,7 @@ def test_instruction_parser_with_invalid_json(mocked_parsing_method):
 @patch("time.sleep")
 @patch(
     "prompt2model.utils.APIAgent.generate_one_completion",
-    side_effect=openai.error.Timeout("timeout"),
+    side_effect=openai.APITimeoutError("timeout"),
 )
 def test_instruction_parser_with_timeout(mocked_parsing_method, mocked_sleep_method):
     """Verify that we wait and retry (a set number of times) if the API times out.
@@ -165,7 +165,7 @@ def test_instruction_parser_with_timeout(mocked_parsing_method, mocked_sleep_met
     assert isinstance(exc_info.value, RuntimeError)
     # Check if the original exception (e) is present as the cause
     original_exception = exc_info.value.__cause__
-    assert isinstance(original_exception, openai.error.Timeout)
+    assert isinstance(original_exception, openai.APITimeoutError)
     gc.collect()