File size: 11,601 Bytes
395201c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 |
from openai import AuthenticationError, BadRequestError, RateLimitError, OpenAIError
import os
import sys
import traceback
import subprocess
sys.path.insert(
0, os.path.abspath("../..")
) # Adds the parent directory to the system path
import litellm
from litellm import (
embedding,
completion,
# AuthenticationError,
ContextWindowExceededError,
# RateLimitError,
# ServiceUnavailableError,
# OpenAIError,
)
from concurrent.futures import ThreadPoolExecutor
import pytest
litellm.vertex_project = "pathrise-convert-1606954137718"
litellm.vertex_location = "us-central1"
# litellm.failure_callback = ["sentry"]
#### What this tests ####
# This tests exception mapping -> trigger an exception from an llm provider -> assert if output is of the expected type
# 5 providers -> OpenAI, Azure, Anthropic, Cohere, Replicate
# 3 main types of exceptions -> - Rate Limit Errors, Context Window Errors, Auth errors (incorrect/rotated key, etc.)
# Approach: Run each model through the test -> assert if the correct error (always the same one) is triggered
models = ["command-nightly"]
# Test 1: Context Window Errors
@pytest.mark.parametrize("model", models)
def test_context_window(model):
sample_text = "Say error 50 times" * 1000000
messages = [{"content": sample_text, "role": "user"}]
try:
litellm.set_verbose = False
response = completion(model=model, messages=messages)
print(f"response: {response}")
print("FAILED!")
pytest.fail(f"An exception occurred")
except ContextWindowExceededError as e:
print(f"Worked!")
except RateLimitError:
print("RateLimited!")
except Exception as e:
print(f"{e}")
pytest.fail(f"An error occcurred - {e}")
@pytest.mark.parametrize("model", models)
def test_context_window_with_fallbacks(model):
ctx_window_fallback_dict = {"command-nightly": "claude-2", "gpt-3.5-turbo-instruct": "gpt-3.5-turbo-16k", "azure/chatgpt-v-2": "gpt-3.5-turbo-16k"}
sample_text = "how does a court case get to the Supreme Court?" * 1000
messages = [{"content": sample_text, "role": "user"}]
completion(model=model, messages=messages, context_window_fallback_dict=ctx_window_fallback_dict)
# for model in litellm.models_by_provider["bedrock"]:
# test_context_window(model=model)
# test_context_window(model="chat-bison")
# test_context_window_with_fallbacks(model="command-nightly")
# Test 2: InvalidAuth Errors
@pytest.mark.parametrize("model", models)
def invalid_auth(model): # set the model key to an invalid key, depending on the model
messages = [{"content": "Hello, how are you?", "role": "user"}]
temporary_key = None
try:
if model == "gpt-3.5-turbo" or model == "gpt-3.5-turbo-instruct":
temporary_key = os.environ["OPENAI_API_KEY"]
os.environ["OPENAI_API_KEY"] = "bad-key"
elif "bedrock" in model:
temporary_aws_access_key = os.environ["AWS_ACCESS_KEY_ID"]
os.environ["AWS_ACCESS_KEY_ID"] = "bad-key"
temporary_aws_region_name = os.environ["AWS_REGION_NAME"]
os.environ["AWS_REGION_NAME"] = "bad-key"
temporary_secret_key = os.environ["AWS_SECRET_ACCESS_KEY"]
os.environ["AWS_SECRET_ACCESS_KEY"] = "bad-key"
elif model == "azure/chatgpt-v-2":
temporary_key = os.environ["AZURE_API_KEY"]
os.environ["AZURE_API_KEY"] = "bad-key"
elif model == "claude-instant-1":
temporary_key = os.environ["ANTHROPIC_API_KEY"]
os.environ["ANTHROPIC_API_KEY"] = "bad-key"
elif model == "command-nightly":
temporary_key = os.environ["COHERE_API_KEY"]
os.environ["COHERE_API_KEY"] = "bad-key"
elif "j2" in model:
temporary_key = os.environ["AI21_API_KEY"]
os.environ["AI21_API_KEY"] = "bad-key"
elif "togethercomputer" in model:
temporary_key = os.environ["TOGETHERAI_API_KEY"]
os.environ["TOGETHERAI_API_KEY"] = "84060c79880fc49df126d3e87b53f8a463ff6e1c6d27fe64207cde25cdfcd1f24a"
elif model in litellm.openrouter_models:
temporary_key = os.environ["OPENROUTER_API_KEY"]
os.environ["OPENROUTER_API_KEY"] = "bad-key"
elif model in litellm.aleph_alpha_models:
temporary_key = os.environ["ALEPH_ALPHA_API_KEY"]
os.environ["ALEPH_ALPHA_API_KEY"] = "bad-key"
elif model in litellm.nlp_cloud_models:
temporary_key = os.environ["NLP_CLOUD_API_KEY"]
os.environ["NLP_CLOUD_API_KEY"] = "bad-key"
elif (
model
== "replicate/llama-2-70b-chat:2c1608e18606fad2812020dc541930f2d0495ce32eee50074220b87300bc16e1"
):
temporary_key = os.environ["REPLICATE_API_KEY"]
os.environ["REPLICATE_API_KEY"] = "bad-key"
print(f"model: {model}")
response = completion(
model=model, messages=messages
)
print(f"response: {response}")
except AuthenticationError as e:
print(f"AuthenticationError Caught Exception - {str(e)}")
except (
OpenAIError
) as e: # is at least an openai error -> in case of random model errors - e.g. overloaded server
print(f"OpenAIError Caught Exception - {e}")
except Exception as e:
print(type(e))
print(type(AuthenticationError))
print(e.__class__.__name__)
print(f"Uncaught Exception - {e}")
pytest.fail(f"Error occurred: {e}")
if temporary_key != None: # reset the key
if model == "gpt-3.5-turbo":
os.environ["OPENAI_API_KEY"] = temporary_key
elif model == "chatgpt-test":
os.environ["AZURE_API_KEY"] = temporary_key
azure = True
elif model == "claude-instant-1":
os.environ["ANTHROPIC_API_KEY"] = temporary_key
elif model == "command-nightly":
os.environ["COHERE_API_KEY"] = temporary_key
elif (
model
== "replicate/llama-2-70b-chat:2c1608e18606fad2812020dc541930f2d0495ce32eee50074220b87300bc16e1"
):
os.environ["REPLICATE_API_KEY"] = temporary_key
elif "j2" in model:
os.environ["AI21_API_KEY"] = temporary_key
elif ("togethercomputer" in model):
os.environ["TOGETHERAI_API_KEY"] = temporary_key
elif model in litellm.aleph_alpha_models:
os.environ["ALEPH_ALPHA_API_KEY"] = temporary_key
elif model in litellm.nlp_cloud_models:
os.environ["NLP_CLOUD_API_KEY"] = temporary_key
elif "bedrock" in model:
os.environ["AWS_ACCESS_KEY_ID"] = temporary_aws_access_key
os.environ["AWS_REGION_NAME"] = temporary_aws_region_name
os.environ["AWS_SECRET_ACCESS_KEY"] = temporary_secret_key
return
# for model in litellm.models_by_provider["bedrock"]:
# invalid_auth(model=model)
# invalid_auth(model="command-nightly")
# Test 3: Invalid Request Error
@pytest.mark.parametrize("model", models)
def test_invalid_request_error(model):
messages = [{"content": "hey, how's it going?", "role": "user"}]
with pytest.raises(BadRequestError):
completion(model=model, messages=messages, max_tokens="hello world")
def test_completion_azure_exception():
try:
import openai
print("azure gpt-3.5 test\n\n")
litellm.set_verbose=False
## Test azure call
old_azure_key = os.environ["AZURE_API_KEY"]
os.environ["AZURE_API_KEY"] = "good morning"
response = completion(
model="azure/chatgpt-v-2",
messages=[
{
"role": "user",
"content": "hello"
}
],
)
print(f"response: {response}")
print(response)
except openai.AuthenticationError as e:
os.environ["AZURE_API_KEY"] = old_azure_key
print("good job got the correct error for azure when key not set")
except Exception as e:
pytest.fail(f"Error occurred: {e}")
test_completion_azure_exception()
async def asynctest_completion_azure_exception():
try:
import openai
import litellm
print("azure gpt-3.5 test\n\n")
litellm.set_verbose=False
## Test azure call
old_azure_key = os.environ["AZURE_API_KEY"]
os.environ["AZURE_API_KEY"] = "good morning"
response = await litellm.acompletion(
model="azure/chatgpt-v-2",
messages=[
{
"role": "user",
"content": "hello"
}
],
)
print(f"response: {response}")
print(response)
except openai.AuthenticationError as e:
os.environ["AZURE_API_KEY"] = old_azure_key
print("good job got the correct error for azure when key not set")
print(e)
except Exception as e:
print("Got wrong exception")
print("exception", e)
pytest.fail(f"Error occurred: {e}")
# import asyncio
# asyncio.run(
# asynctest_completion_azure_exception()
# )
def test_completion_openai_exception():
# test if openai:gpt raises openai.AuthenticationError
try:
import openai
print("openai gpt-3.5 test\n\n")
litellm.set_verbose=False
## Test azure call
old_azure_key = os.environ["OPENAI_API_KEY"]
os.environ["OPENAI_API_KEY"] = "good morning"
response = completion(
model="gpt-4",
messages=[
{
"role": "user",
"content": "hello"
}
],
)
print(f"response: {response}")
print(response)
except openai.AuthenticationError as e:
os.environ["OPENAI_API_KEY"] = old_azure_key
print("good job got the correct error for openai when key not set")
except Exception as e:
pytest.fail(f"Error occurred: {e}")
# test_completion_openai_exception()
# # test_invalid_request_error(model="command-nightly")
# # Test 3: Rate Limit Errors
# def test_model_call(model):
# try:
# sample_text = "how does a court case get to the Supreme Court?"
# messages = [{ "content": sample_text,"role": "user"}]
# print(f"model: {model}")
# response = completion(model=model, messages=messages)
# except RateLimitError as e:
# print(f"headers: {e.response.headers}")
# return True
# # except OpenAIError: # is at least an openai error -> in case of random model errors - e.g. overloaded server
# # return True
# except Exception as e:
# print(f"Uncaught Exception {model}: {type(e).__name__} - {e}")
# traceback.print_exc()
# pass
# return False
# # Repeat each model 500 times
# # extended_models = [model for model in models for _ in range(250)]
# extended_models = ["azure/chatgpt-v-2" for _ in range(250)]
# def worker(model):
# return test_model_call(model)
# # Create a dictionary to store the results
# counts = {True: 0, False: 0}
# # Use Thread Pool Executor
# with ThreadPoolExecutor(max_workers=500) as executor:
# # Use map to start the operation in thread pool
# results = executor.map(worker, extended_models)
# # Iterate over results and count True/False
# for result in results:
# counts[result] += 1
# accuracy_score = counts[True]/(counts[True] + counts[False])
# print(f"accuracy_score: {accuracy_score}")
|