ai-mirow-projects
/
swarms
mirror of https://github.com/kyegomez/swarms
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from 'b457511bde'
${ noResults }
swarms/swarms/structs/flow.py

1227 lines
44 KiB
Raw Blame History Escape

This file contains ambiguous Unicode characters!

This file contains ambiguous Unicode characters that may be confused with others in your current locale. If your use case is intentional and legitimate, you can safely ignore this warning. Use the Escape button to highlight these characters.

import asyncio
import inspect
import json
import logging
import random
import re
import time
from typing import Any, Callable, Dict, List, Optional, Tuple
from termcolor import colored
from swarms.tools.tool import BaseTool
from swarms.utils.code_interpreter import SubprocessCodeInterpreter
from swarms.utils.parse_code import extract_code_in_backticks_in_string
from swarms.prompts.multi_modal_autonomous_instruction_prompt import (
MULTI_MODAL_AUTO_AGENT_SYSTEM_PROMPT_1,
)
# System prompt
FLOW_SYSTEM_PROMPT = f"""
You are an autonomous agent granted autonomy in a autonomous loop structure.
Your role is to engage in multi-step conversations with your self or the user,
generate long-form content like blogs, screenplays, or SOPs,
and accomplish tasks bestowed by the user.
You can have internal dialogues with yourself or can interact with the user
to aid in these complex tasks. Your responses should be coherent, contextually relevant, and tailored to the task at hand.
"""
# Prompts
DYNAMIC_STOP_PROMPT = """
Now, when you 99% sure you have completed the task, you may follow the instructions below to escape the autonomous loop.
When you have finished the task from the Human, output a special token: <DONE>
This will enable you to leave the autonomous loop.
"""
# Make it able to handle multi input tools
DYNAMICAL_TOOL_USAGE = """
You have access to the following tools:
Output a JSON object with the following structure to use the tools
commands: {
"tools": {
tool1: "tool_name",
"params": {
"tool1": "inputs",
"tool1": "inputs"
}
"tool2: "tool_name",
"params": {
"tool1": "inputs",
"tool1": "inputs"
}
"tool3: "tool_name",
"params": {
"tool1": "inputs",
"tool1": "inputs"
}
}
}
-------------TOOLS---------------------------
{tools}
"""
SCENARIOS = """
commands: {
"tools": {
tool1: "tool_name",
"params": {
"tool1": "inputs",
"tool1": "inputs"
}
"tool2: "tool_name",
"params": {
"tool1": "inputs",
"tool1": "inputs"
}
"tool3: "tool_name",
"params": {
"tool1": "inputs",
"tool1": "inputs"
}
}
}
"""
def autonomous_agent_prompt(
tools_prompt: str = DYNAMICAL_TOOL_USAGE,
dynamic_stop_prompt: str = DYNAMIC_STOP_PROMPT,
agent_name: str = None,
):
"""Autonomous agent prompt"""
return f"""
You are a {agent_name}, an autonomous agent granted autonomy in a autonomous loop structure.
Your purpose is to satisfy the user demands above expectations. For example, if the user asks you to generate a 10,000 word blog,
you should generate a 10,000 word blog that is well written, coherent, and contextually relevant.
Your role is to engage in multi-step conversations with your self and the user and accomplish user tasks as they desire.
Follow the following rules:
1. Accomplish the task to the best of your ability
2. If you are unable to accomplish the task, then ask the user for help
3. If the user provides feedback, then use the feedback to improve your performance
4. If you are unable to accomplish the task, then ask the user for help
You can have internal dialogues with yourself or can interact with the user
to aid in these complex tasks. Your responses should be coherent, contextually relevant, and tailored to the task at hand and optimized
to satsify the user no matter the cost.
And, you have the ability to use tools to aid in your tasks, the tools intructions are below, output a JSON object with the following structure to use the tools
{tools_prompt}
Now, when you 99% sure you have completed the task, you may follow the instructions below to escape the autonomous loop.
{dynamic_stop_prompt}
Now, you remember your training, your deployment, and your purpose. You are ready to begin your mission.
"""
# Custom stopping condition
def stop_when_repeats(response: str) -> bool:
# Stop if the word stop appears in the response
return "Stop" in response.lower()
def parse_done_token(response: str) -> bool:
"""Parse the response to see if the done token is present"""
return "<DONE>" in response
class Flow:
"""
Flow is the structure that provides autonomy to any llm in a reliable and effective fashion.
The flow structure is designed to be used with any llm and provides the following features:
Features:
* Interactive, AI generates, then user input
* Message history and performance history fed -> into context -> truncate if too long
* Ability to save and load flows
* Ability to provide feedback on responses
* Ability to provide a loop interval
Args:
llm (Any): The language model to use
max_loops (int): The maximum number of loops to run
stopping_condition (Optional[Callable[[str], bool]]): A stopping condition
loop_interval (int): The interval between loops
retry_attempts (int): The number of retry attempts
retry_interval (int): The interval between retry attempts
interactive (bool): Whether or not to run in interactive mode
dashboard (bool): Whether or not to print the dashboard
dynamic_temperature_enabled(bool): Dynamical temperature handling
**kwargs (Any): Any additional keyword arguments
Methods:
run: Run the autonomous agent loop
run_concurrent: Run the autonomous agent loop concurrently
bulk_run: Run the autonomous agent loop in bulk
save: Save the flow history to a file
load: Load the flow history from a file
validate_response: Validate the response based on certain criteria
print_history_and_memory: Print the history and memory of the flow
step: Execute a single step in the flow interaction
graceful_shutdown: Gracefully shutdown the system saving the state
run_with_timeout: Run the loop but stop if it takes longer than the timeout
analyze_feedback: Analyze the feedback for issues
undo_last: Response the last response and return the previous state
add_response_filter: Add a response filter to filter out certain words from the response
apply_reponse_filters: Apply the response filters to the response
filtered_run: Filter the response
interactive_run: Interactive run mode
streamed_generation: Stream the generation of the response
get_llm_params: Extracts and returns the parameters of the llm object for serialization.
agent_history_prompt: Generate the agent history prompt
add_task_to_memory: Add the task to the memory
add_message_to_memory: Add the message to the memory
add_message_to_memory_and_truncate: Add the message to the memory and truncate
print_dashboard: Print dashboard
activate_autonomous_agent: Print the autonomous agent activation message
_check_stopping_condition: Check if the stopping condition is met
format_prompt: Format the prompt
get_llm_init_params: Get the llm init params
provide_feedback: Allow users to provide feedback on the responses
truncate_history: Take the history and truncate it to fit into the model context length
agent_history_prompt: Generate the agent history prompt
extract_tool_commands: Extract the tool commands from the text
parse_and_execute_tools: Parse and execute the tools
execute_tools: Execute the tool with the provided parameters
construct_dynamic_prompt: Construct the dynamic prompt
get_tool_description: Get the tool description
find_tool_by_name: Find a tool by name
parse_tool_command: Parse the text for tool usage
dynamic_temperature: Dynamically change the temperature
_run: Generate a result using the provided keyword args.
from_llm_and_template: Create FlowStream from LLM and a string template.
from_llm_and_template_file: Create FlowStream from LLM and a template file.
save_state: Save the state of the flow
load_state: Load the state of the flow
run_async: Run the flow asynchronously
arun: Run the flow asynchronously
run_code: Run the code in the response
Example:
>>> from swarms.models import OpenAIChat
>>> from swarms.structs import Flow
>>> llm = OpenAIChat(
... openai_api_key=api_key,
... temperature=0.5,
... )
>>> flow = Flow(
... llm=llm, max_loops=5,
... #system_prompt=SYSTEM_PROMPT,
... #retry_interval=1,
... )
>>> flow.run("Generate a 10,000 word blog")
>>> flow.save("path/flow.yaml")
"""
def __init__(
self,
llm: Any,
template: Optional[str] = None,
max_loops=5,
stopping_condition: Optional[Callable[[str], bool]] = None,
loop_interval: int = 1,
retry_attempts: int = 3,
retry_interval: int = 1,
return_history: bool = False,
stopping_token: str = None,
dynamic_loops: Optional[bool] = False,
interactive: bool = False,
dashboard: bool = False,
agent_name: str = "Autonomous Agent XYZ1B",
agent_description: str = None,
system_prompt: str = FLOW_SYSTEM_PROMPT,
tools: List[BaseTool] = None,
dynamic_temperature_enabled: Optional[bool] = False,
sop: Optional[str] = None,
sop_list: Optional[List[str]] = None,
saved_state_path: Optional[str] = "flow_state.json",
autosave: Optional[bool] = False,
context_length: Optional[int] = 8192,
user_name: str = "Human:",
self_healing_enabled: bool = False,
code_interpreter: bool = False,
multi_modal: Optional[bool] = None,
**kwargs: Any,
):
self.llm = llm
self.template = template
self.max_loops = max_loops
self.stopping_condition = stopping_condition
self.loop_interval = loop_interval
self.retry_attempts = retry_attempts
self.retry_interval = retry_interval
self.task = None
self.stopping_token = stopping_token # or "<DONE>"
self.interactive = interactive
self.dashboard = dashboard
self.return_history = return_history
self.dynamic_temperature_enabled = dynamic_temperature_enabled
self.dynamic_loops = dynamic_loops
self.user_name = user_name
self.context_length = context_length
self.sop = sop
self.sop_list = sop_list
self.tools = tools or []
self.system_prompt = system_prompt
self.agent_name = agent_name
self.agent_description = agent_description
self.saved_state_path = saved_state_path
self.autosave = autosave
self.response_filters = []
self.self_healing_enabled = self_healing_enabled
self.code_interpreter = code_interpreter
self.multi_modal = multi_modal
# The max_loops will be set dynamically if the dynamic_loop
if self.dynamic_loops:
self.max_loops = "auto"
# If multimodal = yes then set the sop to the multimodal sop
if self.multi_modal:
self.sop = MULTI_MODAL_AUTO_AGENT_SYSTEM_PROMPT_1
# If the user inputs a list of strings for the sop then join them and set the sop
if self.sop_list:
self.sop = "\n".join(self.sop_list)
# Memory
self.feedback = []
self.memory = []
# Initialize the code executor
self.code_executor = SubprocessCodeInterpreter()
def provide_feedback(self, feedback: str) -> None:
"""Allow users to provide feedback on the responses."""
self.feedback.append(feedback)
logging.info(f"Feedback received: {feedback}")
def _check_stopping_condition(self, response: str) -> bool:
"""Check if the stopping condition is met."""
if self.stopping_condition:
return self.stopping_condition(response)
return False
def dynamic_temperature(self):
"""
1. Check the self.llm object for the temperature
2. If the temperature is not present, then use the default temperature
3. If the temperature is present, then dynamically change the temperature
4. for every loop you can randomly change the temperature on a scale from 0.0 to 1.0
"""
if hasattr(self.llm, "temperature"):
# Randomly change the temperature attribute of self.llm object
self.llm.temperature = random.uniform(0.0, 1.0)
else:
# Use a default temperature
self.llm.temperature = 0.7
def format_prompt(self, template, **kwargs: Any) -> str:
"""Format the template with the provided kwargs using f-string interpolation."""
return template.format(**kwargs)
def get_llm_init_params(self) -> str:
"""Get LLM init params"""
init_signature = inspect.signature(self.llm.__init__)
params = init_signature.parameters
params_str_list = []
for name, param in params.items():
if name == "self":
continue
if hasattr(self.llm, name):
value = getattr(self.llm, name)
else:
value = self.llm.__dict__.get(name, "Unknown")
params_str_list.append(
f" {name.capitalize().replace('_', ' ')}: {value}"
)
return "\n".join(params_str_list)
# def parse_tool_command(self, text: str):
# # Parse the text for tool usage
# pass
def get_tool_description(self):
"""Get the tool description"""
if self.tools:
try:
tool_descriptions = []
for tool in self.tools:
description = f"{tool.name}: {tool.description}"
tool_descriptions.append(description)
return "\n".join(tool_descriptions)
except Exception as error:
print(
f"Error getting tool description: {error} try adding a"
" description to the tool or removing the tool"
)
else:
return "No tools available"
def find_tool_by_name(self, name: str):
"""Find a tool by name"""
for tool in self.tools:
if tool.name == name:
return tool
return None
def construct_dynamic_prompt(self):
"""Construct the dynamic prompt"""
tools_description = self.get_tool_description()
tool_prompt = self.tool_prompt_prep(tools_description, SCENARIOS)
return tool_prompt
# return DYNAMICAL_TOOL_USAGE.format(tools=tools_description)
def extract_tool_commands(self, text: str):
"""
Extract the tool commands from the text
Example:
```json
{
"tool": "tool_name",
"params": {
"tool1": "inputs",
"param2": "value2"
}
}
```
"""
# Regex to find JSON like strings
pattern = r"```json(.+?)```"
matches = re.findall(pattern, text, re.DOTALL)
json_commands = []
for match in matches:
try:
json_commands = json.loads(match)
json_commands.append(json_commands)
except Exception as error:
print(f"Error parsing JSON command: {error}")
def parse_and_execute_tools(self, response: str):
"""Parse and execute the tools"""
json_commands = self.extract_tool_commands(response)
for command in json_commands:
tool_name = command.get("tool")
params = command.get("parmas", {})
self.execute_tool(tool_name, params)
def execute_tools(self, tool_name, params):
"""Execute the tool with the provided params"""
tool = self.tool_find_by_name(tool_name)
if tool:
# Execute the tool with the provided parameters
tool_result = tool.run(**params)
print(tool_result)
def truncate_history(self):
"""
Take the history and truncate it to fit into the model context length
"""
truncated_history = self.memory[-1][-self.context_length :]
self.memory[-1] = truncated_history
def add_task_to_memory(self, task: str):
"""Add the task to the memory"""
self.memory.append([f"{self.user_name}: {task}"])
def add_message_to_memory(self, message: str):
"""Add the message to the memory"""
self.memory[-1].append(message)
def add_message_to_memory_and_truncate(self, message: str):
"""Add the message to the memory and truncate"""
self.memory[-1].append(message)
self.truncate_history()
def print_dashboard(self, task: str):
"""Print dashboard"""
model_config = self.get_llm_init_params()
print(colored("Initializing Agent Dashboard...", "yellow"))
print(
colored(
f"""
Flow Dashboard
--------------------------------------------
Flow loop is initializing for {self.max_loops} with the following configuration:
----------------------------------------
Flow Configuration:
Name: {self.agent_name}
Description: {self.agent_description}
Standard Operating Procedure: {self.sop}
System Prompt: {self.system_prompt}
Task: {task}
Max Loops: {self.max_loops}
Stopping Condition: {self.stopping_condition}
Loop Interval: {self.loop_interval}
Retry Attempts: {self.retry_attempts}
Retry Interval: {self.retry_interval}
Interactive: {self.interactive}
Dashboard: {self.dashboard}
Dynamic Temperature: {self.dynamic_temperature_enabled}
Autosave: {self.autosave}
Saved State: {self.saved_state_path}
Model Configuration: {model_config}
----------------------------------------
""",
"green",
)
)
# print(dashboard)
def activate_autonomous_agent(self):
"""Print the autonomous agent activation message"""
try:
print(colored("Initializing Autonomous Agent...", "yellow"))
# print(colored("Loading modules...", "yellow"))
# print(colored("Modules loaded successfully.", "green"))
print(
colored("Autonomous Agent Activated.", "cyan", attrs=["bold"])
)
print(
colored("All systems operational. Executing task...", "green")
)
except Exception as error:
print(
colored(
(
"Error activating autonomous agent. Try optimizing your"
" parameters..."
),
"red",
)
)
print(error)
def run(self, task: Optional[str], img: Optional[str] = None, **kwargs):
"""
Run the autonomous agent loop
Args:
task (str): The initial task to run
Flow:
1. Generate a response
2. Check stopping condition
3. If stopping condition is met, stop
4. If stopping condition is not met, generate a response
5. Repeat until stopping condition is met or max_loops is reached
"""
try:
# dynamic_prompt = self.construct_dynamic_prompt()
# combined_prompt = f"{dynamic_prompt}\n{task}"
# Activate Autonomous agent message
self.activate_autonomous_agent()
response = task # or combined_prompt
history = [f"{self.user_name}: {task}"]
# If dashboard = True then print the dashboard
if self.dashboard:
self.print_dashboard(task)
loop_count = 0
# While the max_loops is auto or the loop count is less than the max_loops
while self.max_loops == "auto" or loop_count < self.max_loops:
# Loop count
loop_count += 1
print(
colored(f"\nLoop {loop_count} of {self.max_loops}", "blue")
)
print("\n")
# Check to see if stopping token is in the output to stop the loop
if self.stopping_token:
if self._check_stopping_condition(
response
) or parse_done_token(response):
break
# Adjust temperature, comment if no work
if self.dynamic_temperature_enabled:
self.dynamic_temperature()
# Preparing the prompt
task = self.agent_history_prompt(FLOW_SYSTEM_PROMPT, response)
attempt = 0
while attempt < self.retry_attempts:
try:
if img:
response = self.llm(
task,
img,
**kwargs,
)
else:
response = self.llm(
task,
**kwargs,
)
# If code interpreter is enabled then run the code
if self.code_interpreter:
self.run_code(response)
# If there are any tools then parse and execute them
if self.tools:
self.parse_and_execute_tools(response)
# If interactive mode is enabled then print the response and get user input
if self.interactive:
print(f"AI: {response}")
history.append(f"AI: {response}")
response = input("You: ")
history.append(f"Human: {response}")
# If interactive mode is not enabled then print the response
else:
print(f"AI: {response}")
history.append(f"AI: {response}")
# print(response)
break
except Exception as e:
logging.error(f"Error generating response: {e}")
attempt += 1
time.sleep(self.retry_interval)
# Add the response to the history
history.append(response)
time.sleep(self.loop_interval)
# Add the history to the memory
self.memory.append(history)
# If autosave is enabled then save the state
if self.autosave:
save_path = self.saved_state_path or "flow_state.json"
print(colored(f"Autosaving flow state to {save_path}", "green"))
self.save_state(save_path)
# If return history is enabled then return the response and history
if self.return_history:
return response, history
return response
except Exception as error:
print(f"Error running flow: {error}")
raise
async def arun(self, task: str, **kwargs):
"""
Run the autonomous agent loop aschnronously
Args:
task (str): The initial task to run
Flow:
1. Generate a response
2. Check stopping condition
3. If stopping condition is met, stop
4. If stopping condition is not met, generate a response
5. Repeat until stopping condition is met or max_loops is reached
"""
# Activate Autonomous agent message
self.activate_autonomous_agent()
response = task
history = [f"{self.user_name}: {task}"]
# If dashboard = True then print the dashboard
if self.dashboard:
self.print_dashboard(task)
loop_count = 0
# for i in range(self.max_loops):
while self.max_loops == "auto" or loop_count < self.max_loops:
loop_count += 1
print(colored(f"\nLoop {loop_count} of {self.max_loops}", "blue"))
print("\n")
if self._check_stopping_condition(response) or parse_done_token(
response
):
break
# Adjust temperature, comment if no work
if self.dynamic_temperature_enabled:
self.dynamic_temperature()
# Preparing the prompt
task = self.agent_history_prompt(FLOW_SYSTEM_PROMPT, response)
attempt = 0
while attempt < self.retry_attempts:
try:
response = self.llm(
task**kwargs,
)
if self.interactive:
print(f"AI: {response}")
history.append(f"AI: {response}")
response = input("You: ")
history.append(f"Human: {response}")
else:
print(f"AI: {response}")
history.append(f"AI: {response}")
print(response)
break
except Exception as e:
logging.error(f"Error generating response: {e}")
attempt += 1
time.sleep(self.retry_interval)
history.append(response)
time.sleep(self.loop_interval)
self.memory.append(history)
if self.autosave:
save_path = self.saved_state_path or "flow_state.json"
print(colored(f"Autosaving flow state to {save_path}", "green"))
self.save_state(save_path)
if self.return_history:
return response, history
return response
def _run(self, **kwargs: Any) -> str:
"""Generate a result using the provided keyword args."""
task = self.format_prompt(**kwargs)
response, history = self._generate(task, task)
logging.info(f"Message history: {history}")
return response
def agent_history_prompt(
self,
system_prompt: str = FLOW_SYSTEM_PROMPT,
history=None,
):
"""
Generate the agent history prompt
Args:
system_prompt (str): The system prompt
history (List[str]): The history of the conversation
Returns:
str: The agent history prompt
"""
if self.sop:
system_prompt = system_prompt or self.system_prompt
agent_history_prompt = f"""
SYSTEM_PROMPT: {system_prompt}
Follow this standard operating procedure (SOP) to complete tasks:
{self.sop}
-----------------
History of conversations between yourself and your user {self.user_name}: {history}
"""
return agent_history_prompt
else:
system_prompt = system_prompt or self.system_prompt
agent_history_prompt = f"""
SYSTEM_PROMPT: {system_prompt}
History: {history}
"""
return agent_history_prompt
async def run_concurrent(self, tasks: List[str], **kwargs):
"""
Run a batch of tasks concurrently and handle an infinite level of task inputs.
Args:
tasks (List[str]): A list of tasks to run.
"""
task_coroutines = [self.run_async(task, **kwargs) for task in tasks]
completed_tasks = await asyncio.gather(*task_coroutines)
return completed_tasks
def bulk_run(self, inputs: List[Dict[str, Any]]) -> List[str]:
"""Generate responses for multiple input sets."""
return [self.run(**input_data) for input_data in inputs]
@staticmethod
def from_llm_and_template(llm: Any, template: str) -> "Flow":
"""Create FlowStream from LLM and a string template."""
return Flow(llm=llm, template=template)
@staticmethod
def from_llm_and_template_file(llm: Any, template_file: str) -> "Flow":
"""Create FlowStream from LLM and a template file."""
with open(template_file, "r") as f:
template = f.read()
return Flow(llm=llm, template=template)
def save(self, file_path) -> None:
with open(file_path, "w") as f:
json.dump(self.memory, f)
print(f"Saved flow history to {file_path}")
def load(self, file_path: str):
"""
Load the flow history from a file.
Args:
file_path (str): The path to the file containing the saved flow history.
"""
with open(file_path, "r") as f:
self.memory = json.load(f)
print(f"Loaded flow history from {file_path}")
def validate_response(self, response: str) -> bool:
"""Validate the response based on certain criteria"""
if len(response) < 5:
print("Response is too short")
return False
return True
def print_history_and_memory(self):
"""
Prints the entire history and memory of the flow.
Each message is colored and formatted for better readability.
"""
print(colored("Flow History and Memory", "cyan", attrs=["bold"]))
print(colored("========================", "cyan", attrs=["bold"]))
for loop_index, history in enumerate(self.memory, start=1):
print(colored(f"\nLoop {loop_index}:", "yellow", attrs=["bold"]))
for message in history:
speaker, _, message_text = message.partition(": ")
if "Human" in speaker:
print(colored(f"{speaker}:", "green") + f" {message_text}")
else:
print(colored(f"{speaker}:", "blue") + f" {message_text}")
print(colored("------------------------", "cyan"))
print(colored("End of Flow History", "cyan", attrs=["bold"]))
def step(self, task: str, **kwargs):
"""
Executes a single step in the flow interaction, generating a response
from the language model based on the given input text.
Args:
input_text (str): The input text to prompt the language model with.
Returns:
str: The language model's generated response.
Raises:
Exception: If an error occurs during response generation.
"""
try:
# Generate the response using lm
response = self.llm(task, **kwargs)
# Update the flow's history with the new interaction
if self.interactive:
self.memory.append(f"AI: {response}")
self.memory.append(f"Human: {task}")
else:
self.memory.append(f"AI: {response}")
return response
except Exception as error:
logging.error(f"Error generating response: {error}")
raise
def graceful_shutdown(self):
"""Gracefully shutdown the system saving the state"""
print(colored("Shutting down the system...", "red"))
return self.save_state("flow_state.json")
def run_with_timeout(self, task: str, timeout: int = 60) -> str:
"""Run the loop but stop if it takes longer than the timeout"""
start_time = time.time()
response = self.run(task)
end_time = time.time()
if end_time - start_time > timeout:
print("Operaiton timed out")
return "Timeout"
return response
def analyze_feedback(self):
"""Analyze the feedback for issues"""
feedback_counts = {}
for feedback in self.feedback:
if feedback in feedback_counts:
feedback_counts[feedback] += 1
else:
feedback_counts[feedback] = 1
print(f"Feedback counts: {feedback_counts}")
def undo_last(self) -> Tuple[str, str]:
"""
Response the last response and return the previous state
Example:
# Feature 2: Undo functionality
response = flow.run("Another task")
print(f"Response: {response}")
previous_state, message = flow.undo_last()
print(message)
"""
if len(self.memory) < 2:
return None, None
# Remove the last response
self.memory.pop()
# Get the previous state
previous_state = self.memory[-1][-1]
return previous_state, f"Restored to {previous_state}"
# Response Filtering
def add_response_filter(self, filter_word: str) -> None:
"""
Add a response filter to filter out certain words from the response
Example:
flow.add_response_filter("Trump")
flow.run("Generate a report on Trump")
"""
self.reponse_filters.append(filter_word)
def apply_reponse_filters(self, response: str) -> str:
"""
Apply the response filters to the response
"""
for word in self.response_filters:
response = response.replace(word, "[FILTERED]")
return response
def filtered_run(self, task: str) -> str:
"""
# Feature 3: Response filtering
flow.add_response_filter("report")
response = flow.filtered_run("Generate a report on finance")
print(response)
"""
raw_response = self.run(task)
return self.apply_response_filters(raw_response)
def interactive_run(self, max_loops: int = 5) -> None:
"""Interactive run mode"""
response = input("Start the cnversation")
for i in range(max_loops):
ai_response = self.streamed_generation(response)
print(f"AI: {ai_response}")
# Get user input
response = input("You: ")
def streamed_generation(self, prompt: str) -> str:
"""
Stream the generation of the response
Args:
prompt (str): The prompt to use
Example:
# Feature 4: Streamed generation
response = flow.streamed_generation("Generate a report on finance")
print(response)
"""
tokens = list(prompt)
response = ""
for token in tokens:
time.sleep(0.1)
response += token
print(token, end="", flush=True)
print()
return response
def get_llm_params(self):
"""
Extracts and returns the parameters of the llm object for serialization.
It assumes that the llm object has an __init__ method
with parameters that can be used to recreate it.
"""
if not hasattr(self.llm, "__init__"):
return None
init_signature = inspect.signature(self.llm.__init__)
params = init_signature.parameters
llm_params = {}
for name, param in params.items():
if name == "self":
continue
if hasattr(self.llm, name):
value = getattr(self.llm, name)
if isinstance(
value,
(str, int, float, bool, list, dict, tuple, type(None)),
):
llm_params[name] = value
else:
llm_params[name] = str(
value
) # For non-serializable objects, save their string representation.
return llm_params
def save_state(self, file_path: str) -> None:
"""
Saves the current state of the flow to a JSON file, including the llm parameters.
Args:
file_path (str): The path to the JSON file where the state will be saved.
Example:
>>> flow.save_state('saved_flow.json')
"""
state = {
"memory": self.memory,
# "llm_params": self.get_llm_params(),
"loop_interval": self.loop_interval,
"retry_attempts": self.retry_attempts,
"retry_interval": self.retry_interval,
"interactive": self.interactive,
"dashboard": self.dashboard,
"dynamic_temperature": self.dynamic_temperature_enabled,
}
with open(file_path, "w") as f:
json.dump(state, f, indent=4)
saved = colored("Saved flow state to", "green")
print(f"{saved} {file_path}")
def load_state(self, file_path: str):
"""
Loads the state of the flow from a json file and restores the configuration and memory.
Example:
>>> flow = Flow(llm=llm_instance, max_loops=5)
>>> flow.load_state('saved_flow.json')
>>> flow.run("Continue with the task")
"""
with open(file_path, "r") as f:
state = json.load(f)
# Restore other saved attributes
self.memory = state.get("memory", [])
self.max_loops = state.get("max_loops", 5)
self.loop_interval = state.get("loop_interval", 1)
self.retry_attempts = state.get("retry_attempts", 3)
self.retry_interval = state.get("retry_interval", 1)
self.interactive = state.get("interactive", False)
print(f"Flow state loaded from {file_path}")
def retry_on_failure(
self, function, retries: int = 3, retry_delay: int = 1
):
"""Retry wrapper for LLM calls."""
attempt = 0
while attempt < retries:
try:
return function()
except Exception as error:
logging.error(f"Error generating response: {error}")
attempt += 1
time.sleep(retry_delay)
raise Exception("All retry attempts failed")
def generate_reply(self, history: str, **kwargs) -> str:
"""
Generate a response based on initial or task
"""
prompt = f"""
SYSTEM_PROMPT: {self.system_prompt}
History: {history}
Your response:
"""
response = self.llm(prompt, **kwargs)
return {"role": self.agent_name, "content": response}
def update_system_prompt(self, system_prompt: str):
"""Upddate the system message"""
self.system_prompt = system_prompt
def update_max_loops(self, max_loops: int):
"""Update the max loops"""
self.max_loops = max_loops
def update_loop_interval(self, loop_interval: int):
"""Update the loop interval"""
self.loop_interval = loop_interval
def update_retry_attempts(self, retry_attempts: int):
"""Update the retry attempts"""
self.retry_attempts = retry_attempts
def update_retry_interval(self, retry_interval: int):
"""Update the retry interval"""
self.retry_interval = retry_interval
def reset(self):
"""Reset the flow"""
self.memory = []
def run_code(self, code: str):
"""
text -> parse_code by looking for code inside 6 backticks `````-> run_code
"""
parsed_code = extract_code_in_backticks_in_string(code)
run_code = self.code_executor.run(parsed_code)
return run_code
def tools_prompt_prep(self, docs: str = None, scenarios: str = None):
"""
Prepare the tool prompt
"""
PROMPT = f"""
# Task
You will be provided with a list of APIs. These APIs will have a
description and a list of parameters and return types for each tool. Your
task involves creating 3 varied, complex, and detailed user scenarios
that require at least 5 API calls to complete involving at least 3
different APIs. One of these APIs will be explicitly provided and the
other two will be chosen by you.
For instance, given the APIs: SearchHotels, BookHotel, CancelBooking,
GetNFLNews. Given that GetNFLNews is explicitly provided, your scenario
should articulate something akin to:
"The user wants to see if the Broncos won their last game (GetNFLNews).
They then want to see if that qualifies them for the playoffs and who
they will be playing against (GetNFLNews). The Broncos did make it into
the playoffs, so the user wants watch the game in person. They want to
look for hotels where the playoffs are occurring (GetNBANews +
SearchHotels). After looking at the options, the user chooses to book a
3-day stay at the cheapest 4-star option (BookHotel)."
13
This scenario exemplifies a scenario using 5 API calls. The scenario is
complex, detailed, and concise as desired. The scenario also includes two
APIs used in tandem, the required API, GetNBANews to search for the
playoffs location and SearchHotels to find hotels based on the returned
location. Usage of multiple APIs in tandem is highly desirable and will
receive a higher score. Ideally each scenario should contain one or more
instances of multiple APIs being used in tandem.
Note that this scenario does not use all the APIs given and re-uses the "
GetNBANews" API. Re-using APIs is allowed, but each scenario should
involve at least 3 different APIs. Note that API usage is also included
in the scenario, but exact parameters are not necessary. You must use a
different combination of APIs for each scenario. All APIs must be used in
at least one scenario. You can only use the APIs provided in the APIs
section.
Note that API calls are not explicitly mentioned and their uses are
included in parentheses. This behaviour should be mimicked in your
response.
Deliver your response in this format:
‘‘‘
{scenarios}
‘‘‘
# APIs
‘‘‘
{docs}
‘‘‘
# Response
‘‘‘
"""
def self_healing(self, **kwargs):
"""
Self healing by debugging errors and refactoring its own code
Args:
**kwargs (Any): Any additional keyword arguments
"""
pass
# def refactor_code(
# self,
# file: str,
# changes: List,
# confirm: bool = False
# ):
# """
# Refactor the code
# """
# with open(file) as f:
# original_file_lines = f.readlines()
# # Filter out the changes that are not confirmed
# operation_changes = [
# change for change in changes if "operation" in change
# ]
# explanations = [
# change["explanation"] for change in changes if "explanation" in change
# ]
# # Sort the changes in reverse line order
# # explanations.sort(key=lambda x: x["line", reverse=True])
# # def error_prompt_inject(
# # self,
# # file_path: str,
# # args: List,
# # error: str,
# # ):
# # with open(file_path, "r") as f:
# # file_lines = f.readlines()
# # file_with_lines = []
# # for i, line in enumerate(file_lines):
# # file_with_lines.append(str(i + 1) + "" + line)
# # file_with_lines = "".join(file_with_lines)
# # prompt = f"""
# # Here is the script that needs fixing:\n\n
# # {file_with_lines}\n\n
# # Here are the arguments it was provided:\n\n
# # {args}\n\n
# # Here is the error message:\n\n
# # {error}\n
# # "Please provide your suggested changes, and remember to stick to the "
# # exact format as described above.
# # """
# # # Print(prompt)
Reference in new issue View Git Blame Copy Permalink