phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-27 11:43:54 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix(utils.py): round max tokens to be int always

Browse source
This commit is contained in:
Krrish Dholakia 2024-02-06 13:10:29 -08:00
parent 647dbb9331
commit eee5353e77
2 changed files with 6 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

5
litellm/tests/test_completion.py
View file

@ -544,13 +544,13 @@ def hf_test_completion_tgi():
def test_completion_openai(): def test_completion_openai():
try: try:
litellm.set_verbose = True litellm.set_verbose = True
litellm.drop_params = True
print(f"api key: {os.environ['OPENAI_API_KEY']}") print(f"api key: {os.environ['OPENAI_API_KEY']}")
litellm.api_key = os.environ["OPENAI_API_KEY"] litellm.api_key = os.environ["OPENAI_API_KEY"]
response = completion( response = completion(
model="gpt-3.5-turbo", model="gpt-3.5-turbo",
messages=messages, messages=[{"role": "user", "content": "Hey"}],
max_tokens=10, max_tokens=10,
request_timeout=1,
metadata={"hi": "bye"}, metadata={"hi": "bye"},
) )
print("This is the response object\n", response) print("This is the response object\n", response)
@ -565,6 +565,7 @@ def test_completion_openai():
assert len(response_str) > 1 assert len(response_str) > 1
litellm.api_key = None litellm.api_key = None
raise Exception("it works!")
except Timeout as e: except Timeout as e:
pass pass
except Exception as e: except Exception as e:

4
litellm/utils.py
View file

@ -2348,7 +2348,9 @@ def client(original_function):
elif user_max_tokens + input_tokens > max_output_tokens: elif user_max_tokens + input_tokens > max_output_tokens:
user_max_tokens = max_output_tokens - input_tokens user_max_tokens = max_output_tokens - input_tokens
print_verbose(f"user_max_tokens: {user_max_tokens}") print_verbose(f"user_max_tokens: {user_max_tokens}")
kwargs["max_tokens"] = user_max_tokens kwargs["max_tokens"] = int(
round(user_max_tokens)
) # make sure max tokens is always an int
except Exception as e: except Exception as e:
print_verbose(f"Error while checking max token limit: {str(e)}") print_verbose(f"Error while checking max token limit: {str(e)}")
# MODEL CALL # MODEL CALL