Spaces:
Sleeping
Sleeping
File size: 8,503 Bytes
7db0ae4 |
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 |
#### What this tests ####
# This tests the router's ability to identify the least busy deployment
import sys, os, asyncio, time, random
import traceback
from dotenv import load_dotenv
load_dotenv()
import os
sys.path.insert(
0, os.path.abspath("../..")
) # Adds the parent directory to the system path
import pytest
from litellm import Router
import litellm
from litellm.router_strategy.least_busy import LeastBusyLoggingHandler
from litellm.caching import DualCache
### UNIT TESTS FOR LEAST BUSY LOGGING ###
def test_model_added():
test_cache = DualCache()
least_busy_logger = LeastBusyLoggingHandler(router_cache=test_cache, model_list=[])
kwargs = {
"litellm_params": {
"metadata": {
"model_group": "gpt-3.5-turbo",
"deployment": "azure/chatgpt-v-2",
},
"model_info": {"id": "1234"},
}
}
least_busy_logger.log_pre_api_call(model="test", messages=[], kwargs=kwargs)
request_count_api_key = f"gpt-3.5-turbo_request_count"
assert test_cache.get_cache(key=request_count_api_key) is not None
def test_get_available_deployments():
test_cache = DualCache()
least_busy_logger = LeastBusyLoggingHandler(router_cache=test_cache, model_list=[])
model_group = "gpt-3.5-turbo"
deployment = "azure/chatgpt-v-2"
kwargs = {
"litellm_params": {
"metadata": {
"model_group": model_group,
"deployment": deployment,
},
"model_info": {"id": "1234"},
}
}
least_busy_logger.log_pre_api_call(model="test", messages=[], kwargs=kwargs)
request_count_api_key = f"{model_group}_request_count"
assert test_cache.get_cache(key=request_count_api_key) is not None
# test_get_available_deployments()
def test_router_get_available_deployments():
"""
Tests if 'get_available_deployments' returns the least busy deployment
"""
model_list = [
{
"model_name": "azure-model",
"litellm_params": {
"model": "azure/gpt-turbo",
"api_key": "os.environ/AZURE_FRANCE_API_KEY",
"api_base": "https://openai-france-1234.openai.azure.com",
"rpm": 1440,
},
"model_info": {"id": 1},
},
{
"model_name": "azure-model",
"litellm_params": {
"model": "azure/gpt-35-turbo",
"api_key": "os.environ/AZURE_EUROPE_API_KEY",
"api_base": "https://my-endpoint-europe-berri-992.openai.azure.com",
"rpm": 6,
},
"model_info": {"id": 2},
},
{
"model_name": "azure-model",
"litellm_params": {
"model": "azure/gpt-35-turbo",
"api_key": "os.environ/AZURE_CANADA_API_KEY",
"api_base": "https://my-endpoint-canada-berri992.openai.azure.com",
"rpm": 6,
},
"model_info": {"id": 3},
},
]
router = Router(
model_list=model_list,
routing_strategy="least-busy",
set_verbose=False,
num_retries=3,
) # type: ignore
router.leastbusy_logger.test_flag = True
model_group = "azure-model"
deployment = "azure/chatgpt-v-2"
request_count_dict = {1: 10, 2: 54, 3: 100}
cache_key = f"{model_group}_request_count"
router.cache.set_cache(key=cache_key, value=request_count_dict)
deployment = router.get_available_deployment(model=model_group, messages=None)
print(f"deployment: {deployment}")
assert deployment["model_info"]["id"] == 1
## run router completion - assert completion event, no change in 'busy'ness once calls are complete
router.completion(
model=model_group,
messages=[{"role": "user", "content": "Hey, how's it going?"}],
)
return_dict = router.cache.get_cache(key=cache_key)
assert router.leastbusy_logger.logged_success == 1
assert return_dict[1] == 10
assert return_dict[2] == 54
assert return_dict[3] == 100
## Test with Real calls ##
@pytest.mark.asyncio
async def test_router_atext_completion_streaming():
prompt = "Hello, can you generate a 500 words poem?"
model = "azure-model"
model_list = [
{
"model_name": "azure-model",
"litellm_params": {
"model": "azure/gpt-turbo",
"api_key": "os.environ/AZURE_FRANCE_API_KEY",
"api_base": "https://openai-france-1234.openai.azure.com",
"rpm": 1440,
},
"model_info": {"id": 1},
},
{
"model_name": "azure-model",
"litellm_params": {
"model": "azure/gpt-35-turbo",
"api_key": "os.environ/AZURE_EUROPE_API_KEY",
"api_base": "https://my-endpoint-europe-berri-992.openai.azure.com",
"rpm": 6,
},
"model_info": {"id": 2},
},
{
"model_name": "azure-model",
"litellm_params": {
"model": "azure/gpt-35-turbo",
"api_key": "os.environ/AZURE_CANADA_API_KEY",
"api_base": "https://my-endpoint-canada-berri992.openai.azure.com",
"rpm": 6,
},
"model_info": {"id": 3},
},
]
router = Router(
model_list=model_list,
routing_strategy="least-busy",
set_verbose=False,
num_retries=3,
) # type: ignore
### Call the async calls in sequence, so we start 1 call before going to the next.
## CALL 1
await asyncio.sleep(random.uniform(0, 2))
await router.atext_completion(model=model, prompt=prompt, stream=True)
## CALL 2
await asyncio.sleep(random.uniform(0, 2))
await router.atext_completion(model=model, prompt=prompt, stream=True)
## CALL 3
await asyncio.sleep(random.uniform(0, 2))
await router.atext_completion(model=model, prompt=prompt, stream=True)
cache_key = f"{model}_request_count"
## check if calls equally distributed
cache_dict = router.cache.get_cache(key=cache_key)
for k, v in cache_dict.items():
assert v == 1
# asyncio.run(test_router_atext_completion_streaming())
@pytest.mark.asyncio
async def test_router_completion_streaming():
messages = [
{"role": "user", "content": "Hello, can you generate a 500 words poem?"}
]
model = "azure-model"
model_list = [
{
"model_name": "azure-model",
"litellm_params": {
"model": "azure/gpt-turbo",
"api_key": "os.environ/AZURE_FRANCE_API_KEY",
"api_base": "https://openai-france-1234.openai.azure.com",
"rpm": 1440,
},
"model_info": {"id": 1},
},
{
"model_name": "azure-model",
"litellm_params": {
"model": "azure/gpt-35-turbo",
"api_key": "os.environ/AZURE_EUROPE_API_KEY",
"api_base": "https://my-endpoint-europe-berri-992.openai.azure.com",
"rpm": 6,
},
"model_info": {"id": 2},
},
{
"model_name": "azure-model",
"litellm_params": {
"model": "azure/gpt-35-turbo",
"api_key": "os.environ/AZURE_CANADA_API_KEY",
"api_base": "https://my-endpoint-canada-berri992.openai.azure.com",
"rpm": 6,
},
"model_info": {"id": 3},
},
]
router = Router(
model_list=model_list,
routing_strategy="least-busy",
set_verbose=False,
num_retries=3,
) # type: ignore
### Call the async calls in sequence, so we start 1 call before going to the next.
## CALL 1
await asyncio.sleep(random.uniform(0, 2))
await router.acompletion(model=model, messages=messages, stream=True)
## CALL 2
await asyncio.sleep(random.uniform(0, 2))
await router.acompletion(model=model, messages=messages, stream=True)
## CALL 3
await asyncio.sleep(random.uniform(0, 2))
await router.acompletion(model=model, messages=messages, stream=True)
cache_key = f"{model}_request_count"
## check if calls equally distributed
cache_dict = router.cache.get_cache(key=cache_key)
for k, v in cache_dict.items():
assert v == 1
|