phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-26 11:14:04 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fixing claude max token testing

Browse source
This commit is contained in:
Krrish Dholakia 2023-09-27 21:52:30 -07:00
parent dd61a5b35e
commit 54409a2a30
5 changed files with 17 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

BIN
litellm/__pycache__/main.cpython-311.pyc
View file

Binary file not shown.

BIN
litellm/__pycache__/utils.cpython-311.pyc
View file

Binary file not shown.

2
litellm/llms/anthropic.py
View file

@ -60,6 +60,8 @@ def completion(
prompt += f"{AnthropicConstants.HUMAN_PROMPT.value}{message['content']}" prompt += f"{AnthropicConstants.HUMAN_PROMPT.value}{message['content']}"
prompt += f"{AnthropicConstants.AI_PROMPT.value}" prompt += f"{AnthropicConstants.AI_PROMPT.value}"
max_tokens_to_sample = optional_params.get("max_tokens_to_sample", 256) # required anthropic param, default to 256 if user does not provide an input max_tokens_to_sample = optional_params.get("max_tokens_to_sample", 256) # required anthropic param, default to 256 if user does not provide an input
if max_tokens_to_sample != 256: # not default - print for testing
print_verbose(f"LiteLLM.Anthropic: Max Tokens Set")
data = { data = {
"model": model, "model": model,
"prompt": prompt, "prompt": prompt,

21
litellm/tests/test_completion.py
View file

@ -3,7 +3,7 @@ import traceback
from dotenv import load_dotenv from dotenv import load_dotenv
load_dotenv() load_dotenv()
import os import os, io
sys.path.insert( sys.path.insert(
0, os.path.abspath("../..") 0, os.path.abspath("../..")
@ -51,6 +51,11 @@ def test_completion_claude():
def test_completion_claude_max_tokens(): def test_completion_claude_max_tokens():
try: try:
litellm.set_verbose = True
# Redirect stdout
old_stdout = sys.stdout
sys.stdout = new_stdout = io.StringIO()
# test setting max tokens for claude-2 # test setting max tokens for claude-2
user_message = "tell me everything about YC - be verbose" user_message = "tell me everything about YC - be verbose"
messages = [{"content": user_message, "role": "user"}] messages = [{"content": user_message, "role": "user"}]
@ -58,15 +63,17 @@ def test_completion_claude_max_tokens():
response = completion( response = completion(
model="claude-instant-1", messages=messages, max_tokens=1200 model="claude-instant-1", messages=messages, max_tokens=1200
) )
# Restore stdout
sys.stdout = old_stdout
output = new_stdout.getvalue().strip()
# Add any assertions here to check the response # Add any assertions here to check the response
print(response) if "LiteLLM.Anthropic: Max Tokens Set" not in output:
text_response = response['choices'][0]['message']['content'] raise Exception("Required log message not found!")
print(len(text_response)) print(f"response: {response}")
assert(len(text_response) > 2000) litellm.set_verbose = False
print(response.response_ms)
except Exception as e: except Exception as e:
pytest.fail(f"Error occurred: {e}") pytest.fail(f"Error occurred: {e}")
# test_completion_claude_max_tokens() test_completion_claude_max_tokens()
# def test_completion_oobabooga(): # def test_completion_oobabooga():
# try: # try:

2
pyproject.toml
View file

@ -1,6 +1,6 @@
[tool.poetry] [tool.poetry]
name = "litellm" name = "litellm"
version = "0.1.792" version = "0.1.793"
description = "Library to easily interface with LLM API providers" description = "Library to easily interface with LLM API providers"
authors = ["BerriAI"] authors = ["BerriAI"]
license = "MIT License" license = "MIT License"