phoenix/litellm
1
0
Fork 0
forked from phoenix/litellm-mirror
Code Pull requests Activity

fix(utils.py): fix stream chunk builder for sync/async success

Browse source
This commit is contained in:
Krrish Dholakia 2023-12-13 07:52:51 -08:00
parent a64bd2ca1e
commit d1aef59fbc
2 changed files with 6 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

6
litellm/utils.py
View file

@ -830,9 +830,9 @@ class Logging:
complete_streaming_response = None
if self.stream == True and self.model_call_details.get("litellm_params", {}).get("acompletion", False) == True:
# if it's acompletion == True, chunks are built/appended in async_success_handler
self.streaming_chunks.append(result)
if result.choices[0].finish_reason is not None: # if it's the last chunk
complete_streaming_response = litellm.stream_chunk_builder(self.streaming_chunks, messages=self.model_call_details.get("messages", None))
streaming_chunks = self.streaming_chunks + [result]
complete_streaming_response = litellm.stream_chunk_builder(streaming_chunks, messages=self.model_call_details.get("messages", None))
else:
# this is a completion() call
if self.stream == True:
@ -1053,7 +1053,7 @@ class Logging:
if self.stream:
if result.choices[0].finish_reason is not None: # if it's the last chunk
self.streaming_chunks.append(result)
print_verbose(f"final set of received chunks: {self.streaming_chunks}")
# print_verbose(f"final set of received chunks: {self.streaming_chunks}")
try:
complete_streaming_response = litellm.stream_chunk_builder(self.streaming_chunks, messages=self.model_call_details.get("messages", None))
except: