Spaces:
Sleeping
Sleeping
File size: 11,374 Bytes
d8d14f1 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 |
import random
from threading import Lock
from time import sleep
from typing import Callable, List, Optional
from swarms.structs.agent import Agent
from swarms.structs.base_swarm import BaseSwarm
from swarms.utils.loguru_logger import initialize_logger
logger = initialize_logger(log_folder="swarm_load_balancer")
class AgentLoadBalancer(BaseSwarm):
"""
A load balancer class that distributes tasks among a group of agents.
Args:
agents (List[Agent]): The list of agents available for task execution.
max_retries (int, optional): The maximum number of retries for a task if it fails. Defaults to 3.
max_loops (int, optional): The maximum number of loops to run a task. Defaults to 5.
cooldown_time (float, optional): The cooldown time between retries. Defaults to 0.
Attributes:
agents (List[Agent]): The list of agents available for task execution.
agent_status (Dict[str, bool]): The status of each agent, indicating whether it is available or not.
max_retries (int): The maximum number of retries for a task if it fails.
max_loops (int): The maximum number of loops to run a task.
agent_performance (Dict[str, Dict[str, int]]): The performance statistics of each agent.
lock (Lock): A lock to ensure thread safety.
cooldown_time (float): The cooldown time between retries.
Methods:
get_available_agent: Get an available agent for task execution.
set_agent_status: Set the status of an agent.
update_performance: Update the performance statistics of an agent.
log_performance: Log the performance statistics of all agents.
run_task: Run a single task using an available agent.
run_multiple_tasks: Run multiple tasks using available agents.
run_task_with_loops: Run a task multiple times using an available agent.
run_task_with_callback: Run a task with a callback function.
run_task_with_timeout: Run a task with a timeout.
"""
def __init__(
self,
agents: List[Agent],
max_retries: int = 3,
max_loops: int = 5,
cooldown_time: float = 0,
):
self.agents = agents
self.agent_status = {
agent.agent_name: True for agent in agents
}
self.max_retries = max_retries
self.max_loops = max_loops
self.agent_performance = {
agent.agent_name: {"success_count": 0, "failure_count": 0}
for agent in agents
}
self.lock = Lock()
self.cooldown_time = cooldown_time
self.swarm_initialization()
def swarm_initialization(self):
logger.info(
"Initializing AgentLoadBalancer with the following agents:"
)
# Make sure all the agents exist
assert self.agents, "No agents provided to the Load Balancer"
# Assert that all agents are of type Agent
for agent in self.agents:
assert isinstance(
agent, Agent
), "All agents should be of type Agent"
for agent in self.agents:
logger.info(f"Agent Name: {agent.agent_name}")
logger.info("Load Balancer Initialized Successfully!")
def get_available_agent(self) -> Optional[Agent]:
"""
Get an available agent for task execution.
Returns:
Optional[Agent]: An available agent, or None if no agents are available.
"""
with self.lock:
available_agents = [
agent
for agent in self.agents
if self.agent_status[agent.agent_name]
]
logger.info(
f"Available agents: {[agent.agent_name for agent in available_agents]}"
)
if not available_agents:
return None
return random.choice(available_agents)
def set_agent_status(self, agent: Agent, status: bool) -> None:
"""
Set the status of an agent.
Args:
agent (Agent): The agent whose status needs to be set.
status (bool): The status to set for the agent.
"""
with self.lock:
self.agent_status[agent.agent_name] = status
def update_performance(self, agent: Agent, success: bool) -> None:
"""
Update the performance statistics of an agent.
Args:
agent (Agent): The agent whose performance statistics need to be updated.
success (bool): Whether the task executed by the agent was successful or not.
"""
with self.lock:
if success:
self.agent_performance[agent.agent_name][
"success_count"
] += 1
else:
self.agent_performance[agent.agent_name][
"failure_count"
] += 1
def log_performance(self) -> None:
"""
Log the performance statistics of all agents.
"""
logger.info("Agent Performance:")
for agent_name, stats in self.agent_performance.items():
logger.info(f"{agent_name}: {stats}")
def run(self, task: str, *args, **kwargs) -> str:
"""
Run a single task using an available agent.
Args:
task (str): The task to be executed.
Returns:
str: The output of the task execution.
Raises:
RuntimeError: If no available agents are found to handle the request.
"""
try:
retries = 0
while retries < self.max_retries:
agent = self.get_available_agent()
if not agent:
raise RuntimeError(
"No available agents to handle the request."
)
try:
self.set_agent_status(agent, False)
output = agent.run(task, *args, **kwargs)
self.update_performance(agent, True)
return output
except Exception as e:
logger.error(
f"Error with agent {agent.agent_name}: {e}"
)
self.update_performance(agent, False)
retries += 1
sleep(self.cooldown_time)
if retries >= self.max_retries:
raise e
finally:
self.set_agent_status(agent, True)
except Exception as e:
logger.error(
f"Task failed: {e} try again by optimizing the code."
)
raise RuntimeError(f"Task failed: {e}")
def run_multiple_tasks(self, tasks: List[str]) -> List[str]:
"""
Run multiple tasks using available agents.
Args:
tasks (List[str]): The list of tasks to be executed.
Returns:
List[str]: The list of outputs corresponding to each task execution.
"""
results = []
for task in tasks:
result = self.run(task)
results.append(result)
return results
def run_task_with_loops(self, task: str) -> List[str]:
"""
Run a task multiple times using an available agent.
Args:
task (str): The task to be executed.
Returns:
List[str]: The list of outputs corresponding to each task execution.
"""
results = []
for _ in range(self.max_loops):
result = self.run(task)
results.append(result)
return results
def run_task_with_callback(
self, task: str, callback: Callable[[str], None]
) -> None:
"""
Run a task with a callback function.
Args:
task (str): The task to be executed.
callback (Callable[[str], None]): The callback function to be called with the task result.
"""
try:
result = self.run(task)
callback(result)
except Exception as e:
logger.error(f"Task failed: {e}")
callback(str(e))
def run_task_with_timeout(self, task: str, timeout: float) -> str:
"""
Run a task with a timeout.
Args:
task (str): The task to be executed.
timeout (float): The maximum time (in seconds) to wait for the task to complete.
Returns:
str: The output of the task execution.
Raises:
TimeoutError: If the task execution exceeds the specified timeout.
Exception: If the task execution raises an exception.
"""
import threading
result = [None]
exception = [None]
def target():
try:
result[0] = self.run(task)
except Exception as e:
exception[0] = e
thread = threading.Thread(target=target)
thread.start()
thread.join(timeout)
if thread.is_alive():
raise TimeoutError(
f"Task timed out after {timeout} seconds."
)
if exception[0]:
raise exception[0]
return result[0]
# if __name__ == "__main__":
# from swarms import llama3Hosted()
# # User initializes the agents
# agents = [
# Agent(
# agent_name="Transcript Generator 1",
# agent_description="Generate a transcript for a youtube video on what swarms are!",
# llm=llama3Hosted(),
# max_loops="auto",
# autosave=True,
# dashboard=False,
# streaming_on=True,
# verbose=True,
# stopping_token="<DONE>",
# interactive=True,
# state_save_file_type="json",
# saved_state_path="transcript_generator_1.json",
# ),
# Agent(
# agent_name="Transcript Generator 2",
# agent_description="Generate a transcript for a youtube video on what swarms are!",
# llm=llama3Hosted(),
# max_loops="auto",
# autosave=True,
# dashboard=False,
# streaming_on=True,
# verbose=True,
# stopping_token="<DONE>",
# interactive=True,
# state_save_file_type="json",
# saved_state_path="transcript_generator_2.json",
# )
# # Add more agents as needed
# ]
# load_balancer = LoadBalancer(agents)
# try:
# result = load_balancer.run_task("Generate a transcript for a youtube video on what swarms are!")
# print(result)
# # Running multiple tasks
# tasks = [
# "Generate a transcript for a youtube video on what swarms are!",
# "Generate a transcript for a youtube video on AI advancements!"
# ]
# results = load_balancer.run_multiple_tasks(tasks)
# for res in results:
# print(res)
# # Running task with loops
# loop_results = load_balancer.run_task_with_loops("Generate a transcript for a youtube video on what swarms are!")
# for res in loop_results:
# print(res)
# except RuntimeError as e:
# print(f"Error: {e}")
# # Log performance
# load_balancer.log_performance()
|