llama-index-llms-openai 0.1.30__tar.gz → 0.2.0__tar.gz

{llama_index_llms_openai-0.1.30 → llama_index_llms_openai-0.2.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: llama-index-llms-openai
-Version: 0.1.30
+Version: 0.2.0
 Summary: llama-index llms openai integration
 License: MIT
 Author: llama-index
@@ -11,7 +11,7 @@ Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
-Requires-Dist: llama-index-core (>=0.10.57,<0.11.0)
+Requires-Dist: llama-index-core (>=0.11.0,<0.12.0)
 Requires-Dist: openai (>=1.40.0,<2.0.0)
 Description-Content-Type: text/markdown
 

{llama_index_llms_openai-0.1.30 → llama_index_llms_openai-0.2.0}/llama_index/llms/openai/base.py

@@ -166,7 +166,8 @@ class OpenAI(FunctionCallingLLM):
         gt=0,
     )
     logprobs: Optional[bool] = Field(
-        description="Whether to return logprobs per token."
+        description="Whether to return logprobs per token.",
+        default=None,
     )
     top_logprobs: int = Field(
         description="The number of top token log probs to return.",
@@ -187,7 +188,7 @@ class OpenAI(FunctionCallingLLM):
         description="The timeout, in seconds, for API requests.",
         gte=0,
     )
-    default_headers: Dict[str, str] = Field(
+    default_headers: Optional[Dict[str, str]] = Field(
         default=None, description="The default headers for API requests."
     )
     reuse_client: bool = Field(
@@ -390,7 +391,13 @@ class OpenAI(FunctionCallingLLM):
                 base_kwargs["top_logprobs"] = self.top_logprobs
             else:
                 base_kwargs["logprobs"] = self.top_logprobs  # int in this case
-        return {**base_kwargs, **self.additional_kwargs}
+
+        # can't send stream_options to the API when not streaming
+        all_kwargs = {**base_kwargs, **self.additional_kwargs}
+        if "stream" not in all_kwargs and "stream_options" in all_kwargs:
+            del all_kwargs["stream_options"]
+
+        return all_kwargs
 
     @llm_retry_decorator
     def _chat(self, messages: Sequence[ChatMessage], **kwargs: Any) -> ChatResponse:
@@ -486,8 +493,7 @@ class OpenAI(FunctionCallingLLM):
             is_function = False
             for response in client.chat.completions.create(
                 messages=message_dicts,
-                stream=True,
-                **self._get_model_kwargs(**kwargs),
+                **self._get_model_kwargs(stream=True, **kwargs),
             ):
                 response = cast(ChatCompletionChunk, response)
                 if len(response.choices) > 0:
@@ -561,15 +567,14 @@ class OpenAI(FunctionCallingLLM):
     @llm_retry_decorator
     def _stream_complete(self, prompt: str, **kwargs: Any) -> CompletionResponseGen:
         client = self._get_client()
-        all_kwargs = self._get_model_kwargs(**kwargs)
+        all_kwargs = self._get_model_kwargs(stream=True, **kwargs)
         self._update_max_tokens(all_kwargs, prompt)
 
         def gen() -> CompletionResponseGen:
             text = ""
             for response in client.completions.create(
                 prompt=prompt,
-                stream=True,
-                **all_kwargs,
+                **kwargs,
             ):
                 if len(response.choices) > 0:
                     delta = response.choices[0].text
@@ -728,8 +733,7 @@ class OpenAI(FunctionCallingLLM):
             first_chat_chunk = True
             async for response in await aclient.chat.completions.create(
                 messages=message_dicts,
-                stream=True,
-                **self._get_model_kwargs(**kwargs),
+                **self._get_model_kwargs(stream=True, **kwargs),
             ):
                 response = cast(ChatCompletionChunk, response)
                 if len(response.choices) > 0:
@@ -815,14 +819,13 @@ class OpenAI(FunctionCallingLLM):
         self, prompt: str, **kwargs: Any
     ) -> CompletionResponseAsyncGen:
         aclient = self._get_aclient()
-        all_kwargs = self._get_model_kwargs(**kwargs)
+        all_kwargs = self._get_model_kwargs(stream=True, **kwargs)
         self._update_max_tokens(all_kwargs, prompt)
 
         async def gen() -> CompletionResponseAsyncGen:
             text = ""
             async for response in await aclient.completions.create(
                 prompt=prompt,
-                stream=True,
                 **all_kwargs,
             ):
                 if len(response.choices) > 0:

{llama_index_llms_openai-0.1.30 → llama_index_llms_openai-0.2.0}/pyproject.toml

@@ -29,12 +29,12 @@ exclude = ["**/BUILD"]
 license = "MIT"
 name = "llama-index-llms-openai"
 readme = "README.md"
-version = "0.1.30"
+version = "0.2.0"
 
 [tool.poetry.dependencies]
 python = ">=3.8.1,<4.0"
-llama-index-core = "^0.10.57"
 openai = "^1.40.0"
+llama-index-core = "^0.11.0"
 
 [tool.poetry.group.dev.dependencies]
 ipython = "8.10.0"