phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-25 02:34:29 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix(utils.py): fix parallel tool calling when streaming

Browse source
This commit is contained in:
Krrish Dholakia 2023-11-29 10:56:21 -08:00
parent 9024a47dc2
commit b6bc75e27a
3 changed files with 74 additions and 51 deletions
Download patch file Download diff file Expand all files Collapse all files

15
litellm/utils.py
View file

@ -5247,11 +5247,14 @@ class CustomStreamWrapper:
original_chunk = response_obj.get("original_chunk", None)
model_response.id = original_chunk.id
if len(original_chunk.choices) > 0:
try:
delta = dict(original_chunk.choices[0].delta)
model_response.choices[0].delta = Delta(**delta)
except Exception as e:
model_response.choices[0].delta = Delta()
if original_chunk.choices[0].delta.function_call is not None or original_chunk.choices[0].delta.tool_calls is not None:
try:
delta = dict(original_chunk.choices[0].delta)
model_response.choices[0].delta = Delta(**delta)
except Exception as e:
model_response.choices[0].delta = Delta()
else:
return
else:
return
model_response.system_fingerprint = original_chunk.system_fingerprint
@ -5284,7 +5287,7 @@ class CustomStreamWrapper:
chunk = self.completion_stream
else:
chunk = next(self.completion_stream)
if chunk is not None and chunk != b'':
response = self.chunk_creator(chunk=chunk)
if response is not None: