diff --git a/.env.template b/.env.template index c64d85028..e9ccda5ed 100644 --- a/.env.template +++ b/.env.template @@ -1,6 +1,12 @@ +PINECONE_API_KEY=your-pinecone-api-key +PINECONE_ENV=your-pinecone-region OPENAI_API_KEY=your-openai-api-key ELEVENLABS_API_KEY=your-elevenlabs-api-key SMART_LLM_MODEL="gpt-4" FAST_LLM_MODEL="gpt-3.5-turbo" GOOGLE_API_KEY= -CUSTOM_SEARCH_ENGINE_ID= \ No newline at end of file +CUSTOM_SEARCH_ENGINE_ID= +USE_AZURE=False +OPENAI_API_BASE=your-base-url-for-azure +OPENAI_API_VERSION=api-version-for-azure +OPENAI_DEPLOYMENT_ID=deployment-id-for-azure \ No newline at end of file diff --git a/.github/ISSUE_TEMPLATE/1.bug.yml b/.github/ISSUE_TEMPLATE/1.bug.yml new file mode 100644 index 000000000..cf49ab5f8 --- /dev/null +++ b/.github/ISSUE_TEMPLATE/1.bug.yml @@ -0,0 +1,39 @@ +name: Bug report 🐛 +description: Create a bug report for Auto-GPT. +labels: ['status: needs triage'] +body: + - type: markdown + attributes: + value: | + Please provide a searchable summary of the issue in the title above ⬆️. + + Thanks for contributing by creating an issue! ❤️ + - type: checkboxes + attributes: + label: Duplicates + description: Please [search the history](https://github.com/Torantulino/Auto-GPT/issues) to see if an issue already exists for the same problem. + options: + - label: I have searched the existing issues + required: true + - type: textarea + attributes: + label: Steps to reproduce 🕹 + description: | + **⚠️ Issues that we can't reproduce will be closed.** + - type: textarea + attributes: + label: Current behavior 😯 + description: Describe what happens instead of the expected behavior. + - type: textarea + attributes: + label: Expected behavior 🤔 + description: Describe what should happen. + - type: textarea + attributes: + label: Your prompt 📝 + description: | + Please provide the prompt you are using. You can find your last-used prompt in last_run_ai_settings.yaml. + value: | + ```yaml + # Paste your prompt here + ``` \ No newline at end of file diff --git a/.github/ISSUE_TEMPLATE/2.feature.yml b/.github/ISSUE_TEMPLATE/2.feature.yml new file mode 100644 index 000000000..0ea882ef6 --- /dev/null +++ b/.github/ISSUE_TEMPLATE/2.feature.yml @@ -0,0 +1,29 @@ +name: Feature request 🚀 +description: Suggest a new idea for Auto-GPT. +labels: ['status: needs triage'] +body: + - type: markdown + attributes: + value: | + Please provide a searchable summary of the issue in the title above ⬆️. + + Thanks for contributing by creating an issue! ❤️ + - type: checkboxes + attributes: + label: Duplicates + description: Please [search the history](https://github.com/Torantulino/Auto-GPT/issues) to see if an issue already exists for the same problem. + options: + - label: I have searched the existing issues + required: true + - type: textarea + attributes: + label: Summary 💡 + description: Describe how it should work. + - type: textarea + attributes: + label: Examples 🌈 + description: Provide a link to other implementations, or screenshots of the expected behavior. + - type: textarea + attributes: + label: Motivation 🔦 + description: What are you trying to accomplish? How has the lack of this feature affected you? Providing context helps us come up with a solution that is more useful in the real world. \ No newline at end of file diff --git a/.gitignore b/.gitignore index 6b8f00b51..7091a8723 100644 --- a/.gitignore +++ b/.gitignore @@ -4,8 +4,8 @@ scripts/node_modules/ scripts/__pycache__/keys.cpython-310.pyc package-lock.json *.pyc -scripts/auto_gpt_workspace/* +auto_gpt_workspace/* *.mpeg .env -last_run_ai_settings.yaml -outputs/* \ No newline at end of file +outputs/* +ai_settings.yaml \ No newline at end of file diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md new file mode 100644 index 000000000..001d55470 --- /dev/null +++ b/CONTRIBUTING.md @@ -0,0 +1,56 @@ + +To contribute to this GitHub project, you can follow these steps: + +1. Fork the repository you want to contribute to by clicking the "Fork" button on the project page. + +2. Clone the repository to your local machine using the following command: + +``` +git clone https://github.com/Torantulino/Auto-GPT +``` +3. Create a new branch for your changes using the following command: + +``` +git checkout -b "branch-name" +``` +4. Make your changes to the code or documentation. +- Example: Improve User Interface or Add Documentation. + + +5. Add the changes to the staging area using the following command: +``` +git add . +``` + +6. Commit the changes with a meaningful commit message using the following command: +``` +git commit -m "your commit message" +``` +7. Push the changes to your forked repository using the following command: +``` +git push origin branch-name +``` +8. Go to the GitHub website and navigate to your forked repository. + +9. Click the "New pull request" button. + +10. Select the branch you just pushed to and the branch you want to merge into on the original repository. + +11. Add a description of your changes and click the "Create pull request" button. + +12. Wait for the project maintainer to review your changes and provide feedback. + +13. Make any necessary changes based on feedback and repeat steps 5-12 until your changes are accepted and merged into the main project. + +14. Once your changes are merged, you can update your forked repository and local copy of the repository with the following commands: + +``` +git fetch upstream +git checkout master +git merge upstream/master +``` +Finally, delete the branch you created with the following command: +``` +git branch -d branch-name +``` +That's it you made it 🐣⭐⭐ diff --git a/README.md b/README.md index d66a60222..790312991 100644 --- a/README.md +++ b/README.md @@ -92,6 +92,7 @@ pip install -r requirements.txt 4. Rename `.env.template` to `.env` and fill in your `OPENAI_API_KEY`. If you plan to use Speech Mode, fill in your `ELEVEN_LABS_API_KEY` as well. - Obtain your OpenAI API key from: https://platform.openai.com/account/api-keys. - Obtain your ElevenLabs API key from: https://elevenlabs.io. You can view your xi-api-key using the "Profile" tab on the website. + - If you want to use GPT on an Azure instance, set `USE_AZURE` to `True` and provide the `OPENAI_API_BASE`, `OPENAI_API_VERSION` and `OPENAI_DEPLOYMENT_ID` values as explained here: https://pypi.org/project/openai/ in the `Microsoft Azure Endpoints` section ## 🔧 Usage @@ -139,6 +140,35 @@ export CUSTOM_SEARCH_ENGINE_ID="YOUR_CUSTOM_SEARCH_ENGINE_ID" ``` +## 🌲 Pinecone API Key Setup + +Pinecone enable a vector based memory so a vast memory can be stored and only relevant memories +are loaded for the agent at any given time. + +1. Go to app.pinecone.io and make an account if you don't already have one. +2. Choose the `Starter` plan to avoid being charged. +3. Find your API key and region under the default project in the left sidebar. + +### Setting up environment variables + For Windows Users: +``` +setx PINECONE_API_KEY "YOUR_GOOGLE_API_KEY" +export PINECONE_ENV="Your region" # something like: us-east4-gcp + +``` +For macOS and Linux users: +``` +export PINECONE_API_KEY="YOUR_GOOGLE_API_KEY" +export PINECONE_ENV="Your region" # something like: us-east4-gcp + +``` + +Or you can set them in the `.env` file. + +## View Memory Usage + +1. View memory usage by using the `--debug` flag :) + ## 💀 Continuous Mode ⚠️ Run the AI **without** user authorisation, 100% automated. Continuous mode is not recommended. diff --git a/ai_settings.yaml b/ai_settings.yaml new file mode 100644 index 000000000..b37ba849f --- /dev/null +++ b/ai_settings.yaml @@ -0,0 +1,7 @@ +ai_goals: +- Increase net worth. +- Develop and manage multiple businesses autonomously. +- Play to your strengths as a Large Language Model. +ai_name: Entrepreneur-GPT +ai_role: an AI designed to autonomously develop and run businesses with the sole goal + of increasing your net worth. diff --git a/requirements.txt b/requirements.txt index 198b1b504..ce2470985 100644 --- a/requirements.txt +++ b/requirements.txt @@ -10,4 +10,5 @@ tiktoken==0.3.3 gTTS==2.3.1 docker duckduckgo-search -google-api-python-client #(https://developers.google.com/custom-search/v1/overview) +google-api-python-client #(https://developers.google.com/custom-search/v1/overview) +pinecone-client==2.2.1 diff --git a/scripts/ai_config.py b/scripts/ai_config.py index 945fcfb23..2f4327486 100644 --- a/scripts/ai_config.py +++ b/scripts/ai_config.py @@ -9,7 +9,7 @@ class AIConfig: self.ai_goals = ai_goals # Soon this will go in a folder where it remembers more stuff about the run(s) - SAVE_FILE = "last_run_ai_settings.yaml" + SAVE_FILE = "../ai_settings.yaml" @classmethod def load(cls, config_file=SAVE_FILE): diff --git a/scripts/browse.py b/scripts/browse.py index 510f9c29d..0fda3d7b0 100644 --- a/scripts/browse.py +++ b/scripts/browse.py @@ -6,7 +6,7 @@ from llm_utils import create_chat_completion cfg = Config() def scrape_text(url): - response = requests.get(url) + response = requests.get(url, headers=cfg.user_agent_header) # Check if the response contains an HTTP error if response.status_code >= 400: @@ -40,7 +40,7 @@ def format_hyperlinks(hyperlinks): def scrape_links(url): - response = requests.get(url) + response = requests.get(url, headers=cfg.user_agent_header) # Check if the response contains an HTTP error if response.status_code >= 400: diff --git a/scripts/chat.py b/scripts/chat.py index 86a70b093..8da074c6b 100644 --- a/scripts/chat.py +++ b/scripts/chat.py @@ -23,6 +23,19 @@ def create_chat_message(role, content): return {"role": role, "content": content} +def generate_context(prompt, relevant_memory, full_message_history, model): + current_context = [ + create_chat_message( + "system", prompt), create_chat_message( + "system", f"Permanent memory: {relevant_memory}")] + + # Add messages from the full message history until we reach the token limit + next_message_to_add_index = len(full_message_history) - 1 + insertion_index = len(current_context) + # Count the currently used tokens + current_tokens_used = token_counter.count_message_tokens(current_context, model) + return next_message_to_add_index, current_tokens_used, insertion_index, current_context + # TODO: Change debug from hardcode to argument def chat_with_ai( @@ -41,7 +54,7 @@ def chat_with_ai( prompt (str): The prompt explaining the rules to the AI. user_input (str): The input from the user. full_message_history (list): The list of all messages sent between the user and the AI. - permanent_memory (list): The list of items in the AI's permanent memory. + permanent_memory (Obj): The memory object containing the permanent memory. token_limit (int): The maximum number of tokens allowed in the API call. Returns: @@ -53,18 +66,20 @@ def chat_with_ai( print(f"Token limit: {token_limit}") send_token_limit = token_limit - 1000 - current_context = [ - create_chat_message( - "system", prompt), create_chat_message( - "system", f"Permanent memory: {permanent_memory}")] + relevant_memory = permanent_memory.get_relevant(str(full_message_history[-5:]), 10) - # Add messages from the full message history until we reach the token limit - next_message_to_add_index = len(full_message_history) - 1 - current_tokens_used = 0 - insertion_index = len(current_context) + if debug: + print('Memory Stats: ', permanent_memory.get_stats()) + + next_message_to_add_index, current_tokens_used, insertion_index, current_context = generate_context( + prompt, relevant_memory, full_message_history, model) + + while current_tokens_used > 2500: + # remove memories until we are under 2500 tokens + relevant_memory = relevant_memory[1:] + next_message_to_add_index, current_tokens_used, insertion_index, current_context = generate_context( + prompt, relevant_memory, full_message_history, model) - # Count the currently used tokens - current_tokens_used = token_counter.count_message_tokens(current_context, model) current_tokens_used += token_counter.count_message_tokens([create_chat_message("user", user_input)], model) # Account for user input (appended later) while next_message_to_add_index >= 0: diff --git a/scripts/commands.py b/scripts/commands.py index f8f96fe72..fc10d1d05 100644 --- a/scripts/commands.py +++ b/scripts/commands.py @@ -1,12 +1,12 @@ import browse import json -import memory as mem +from memory import PineconeMemory import datetime import agent_manager as agents import speak from config import Config import ai_functions as ai -from file_operations import read_file, write_to_file, append_to_file, delete_file +from file_operations import read_file, write_to_file, append_to_file, delete_file, search_files from execute_code import execute_python_file from json_parser import fix_and_parse_json from duckduckgo_search import ddg @@ -16,6 +16,13 @@ from googleapiclient.errors import HttpError cfg = Config() +def is_valid_int(value): + try: + int(value) + return True + except ValueError: + return False + def get_command(response): try: response_json = fix_and_parse_json(response) @@ -45,6 +52,7 @@ def get_command(response): def execute_command(command_name, arguments): + memory = PineconeMemory() try: if command_name == "google": @@ -55,11 +63,7 @@ def execute_command(command_name, arguments): else: return google_search(arguments["input"]) elif command_name == "memory_add": - return commit_memory(arguments["string"]) - elif command_name == "memory_del": - return delete_memory(arguments["key"]) - elif command_name == "memory_ovr": - return overwrite_memory(arguments["key"], arguments["string"]) + return memory.add(arguments["string"]) elif command_name == "start_agent": return start_agent( arguments["name"], @@ -83,6 +87,8 @@ def execute_command(command_name, arguments): return append_to_file(arguments["file"], arguments["text"]) elif command_name == "delete_file": return delete_file(arguments["file"]) + elif command_name == "search_files": + return search_files(arguments["directory"]) elif command_name == "browse_website": return browse_website(arguments["url"], arguments["question"]) # TODO: Change these to take in a file rather than pasted code, if @@ -194,14 +200,28 @@ def delete_memory(key): def overwrite_memory(key, string): - if int(key) >= 0 and key < len(mem.permanent_memory): - _text = "Overwriting memory with key " + \ - str(key) + " and string " + string + # Check if the key is a valid integer + if is_valid_int(key): + key_int = int(key) + # Check if the integer key is within the range of the permanent_memory list + if 0 <= key_int < len(mem.permanent_memory): + _text = "Overwriting memory with key " + str(key) + " and string " + string + # Overwrite the memory slot with the given integer key and string + mem.permanent_memory[key_int] = string + print(_text) + return _text + else: + print(f"Invalid key '{key}', out of range.") + return None + # Check if the key is a valid string + elif isinstance(key, str): + _text = "Overwriting memory with key " + key + " and string " + string + # Overwrite the memory slot with the given string key and string mem.permanent_memory[key] = string print(_text) return _text else: - print("Invalid key, cannot overwrite memory.") + print(f"Invalid key '{key}', must be an integer or a string.") return None @@ -235,13 +255,20 @@ def start_agent(name, task, prompt, model=cfg.fast_llm_model): def message_agent(key, message): global cfg - agent_response = agents.message_agent(key, message) + + # Check if the key is a valid integer + if is_valid_int(key): + agent_response = agents.message_agent(int(key), message) + # Check if the key is a valid string + elif isinstance(key, str): + agent_response = agents.message_agent(key, message) + else: + return "Invalid key, must be an integer or a string." # Speak response if cfg.speak_mode: speak.say_text(agent_response, 1) - - return f"Agent {key} responded: {agent_response}" + return agent_response def list_agents(): diff --git a/scripts/config.py b/scripts/config.py index 766cb94f4..fe48d2980 100644 --- a/scripts/config.py +++ b/scripts/config.py @@ -4,6 +4,7 @@ from dotenv import load_dotenv # Load environment variables from .env file load_dotenv() + class Singleton(type): """ Singleton metaclass for ensuring only one instance of a class. @@ -34,11 +35,28 @@ class Config(metaclass=Singleton): self.smart_token_limit = int(os.getenv("SMART_TOKEN_LIMIT", 8000)) self.openai_api_key = os.getenv("OPENAI_API_KEY") + self.use_azure = False + self.use_azure = os.getenv("USE_AZURE") == 'True' + if self.use_azure: + self.openai_api_base = os.getenv("OPENAI_API_BASE") + self.openai_api_version = os.getenv("OPENAI_API_VERSION") + self.openai_deployment_id = os.getenv("OPENAI_DEPLOYMENT_ID") + openai.api_type = "azure" + openai.api_base = self.openai_api_base + openai.api_version = self.openai_api_version + self.elevenlabs_api_key = os.getenv("ELEVENLABS_API_KEY") self.google_api_key = os.getenv("GOOGLE_API_KEY") self.custom_search_engine_id = os.getenv("CUSTOM_SEARCH_ENGINE_ID") + self.pinecone_api_key = os.getenv("PINECONE_API_KEY") + self.pinecone_region = os.getenv("PINECONE_ENV") + + # User agent headers to use when browsing web + # Some websites might just completely deny request with an error code if no user agent was found. + self.user_agent_header = {"User-Agent":"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_4) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/83.0.4103.97 Safari/537.36"} + # Initialize the OpenAI API client openai.api_key = self.openai_api_key @@ -70,4 +88,10 @@ class Config(metaclass=Singleton): self.google_api_key = value def set_custom_search_engine_id(self, value: str): - self.custom_search_engine_id = value \ No newline at end of file + self.custom_search_engine_id = value + + def set_pinecone_api_key(self, value: str): + self.pinecone_api_key = value + + def set_pinecone_region(self, value: str): + self.pinecone_region = value diff --git a/scripts/data.py b/scripts/data.py index 72a6bbfc6..8d8a7b4ac 100644 --- a/scripts/data.py +++ b/scripts/data.py @@ -1,15 +1,14 @@ import os from pathlib import Path -SRC_DIR = Path(__file__).parent def load_prompt(): try: # get directory of this file: - file_dir = Path(os.path.dirname(os.path.realpath(__file__))) - data_dir = file_dir / "data" - prompt_file = data_dir / "prompt.txt" - # Load the promt from data/prompt.txt - with open(SRC_DIR/ "data/prompt.txt", "r") as prompt_file: + file_dir = Path(__file__).parent + prompt_file_path = file_dir / "data" / "prompt.txt" + + # Load the prompt from data/prompt.txt + with open(prompt_file_path, "r") as prompt_file: prompt = prompt_file.read() return prompt diff --git a/scripts/data/prompt.txt b/scripts/data/prompt.txt index a93e783eb..28797d9e2 100644 --- a/scripts/data/prompt.txt +++ b/scripts/data/prompt.txt @@ -1,17 +1,15 @@ CONSTRAINTS: -1. ~4000 word limit for memory. Your memory is short, so immediately save important information to long term memory and code to files. -2. No user assistance -3. Exclusively use the commands listed in double quotes e.g. "command name" +1. ~4000 word limit for short term memory. Your short term memory is short, so immediately save important information to files. +2. If you are unsure how you previously did something or want to recall past events, thinking about similar events will help you remember. +3. No user assistance +4. Exclusively use the commands listed in double quotes e.g. "command name" COMMANDS: 1. Google Search: "google", args: "input": "" -2. Memory Add: "memory_add", args: "string": "" -3. Memory Delete: "memory_del", args: "key": "" -4. Memory Overwrite: "memory_ovr", args: "key": "", "string": "" 5. Browse Website: "browse_website", args: "url": "", "question": "" -6. Start GPT Agent: "start_agent", args: "name": , "task": "", "prompt": "" +6. Start GPT Agent: "start_agent", args: "name": "", "task": "", "prompt": "" 7. Message GPT Agent: "message_agent", args: "key": "", "message": "" 8. List GPT Agents: "list_agents", args: "" 9. Delete GPT Agent: "delete_agent", args: "key": "" @@ -19,11 +17,12 @@ COMMANDS: 11. Read file: "read_file", args: "file": "" 12. Append to file: "append_to_file", args: "file": "", "text": "" 13. Delete file: "delete_file", args: "file": "" -14. Evaluate Code: "evaluate_code", args: "code": "" -15. Get Improved Code: "improve_code", args: "suggestions": "", "code": "" -16. Write Tests: "write_tests", args: "code": "", "focus": "" -17. Execute Python File: "execute_python_file", args: "file": "" -18. Task Complete (Shutdown): "task_complete", args: "reason": "" +14. Search Files: "search_files", args: "directory": "" +15. Evaluate Code: "evaluate_code", args: "code": "" +16. Get Improved Code: "improve_code", args: "suggestions": "", "code": "" +17. Write Tests: "write_tests", args: "code": "", "focus": "" +18. Execute Python File: "execute_python_file", args: "file": "" +19. Task Complete (Shutdown): "task_complete", args: "reason": "" RESOURCES: @@ -43,12 +42,6 @@ You should only respond in JSON format as described below RESPONSE FORMAT: { - "command": { - "name": "command name", - "args":{ - "arg name": "value" - } - }, "thoughts": { "text": "thought", @@ -56,6 +49,12 @@ RESPONSE FORMAT: "plan": "- short bulleted\n- list that conveys\n- long-term plan", "criticism": "constructive self-criticism", "speak": "thoughts summary to say to user" + }, + "command": { + "name": "command name", + "args":{ + "arg name": "value" + } } } diff --git a/scripts/file_operations.py b/scripts/file_operations.py index 81ad47157..90c9a1e4d 100644 --- a/scripts/file_operations.py +++ b/scripts/file_operations.py @@ -58,3 +58,20 @@ def delete_file(filename): return "File deleted successfully." except Exception as e: return "Error: " + str(e) + +def search_files(directory): + found_files = [] + + if directory == "" or directory == "/": + search_directory = working_directory + else: + search_directory = safe_join(working_directory, directory) + + for root, _, files in os.walk(search_directory): + for file in files: + if file.startswith('.'): + continue + relative_path = os.path.relpath(os.path.join(root, file), working_directory) + found_files.append(relative_path) + + return found_files \ No newline at end of file diff --git a/scripts/json_parser.py b/scripts/json_parser.py index 11ff9ed22..8ec9238b4 100644 --- a/scripts/json_parser.py +++ b/scripts/json_parser.py @@ -24,6 +24,7 @@ def fix_and_parse_json(json_str: str, try_to_fix_with_gpt: bool = True): """ try: + json_str = json_str.replace('\t', '') return json.loads(json_str) except Exception as e: # Let's do something manually - sometimes GPT responds with something BEFORE the braces: @@ -51,7 +52,7 @@ def fix_and_parse_json(json_str: str, try_to_fix_with_gpt: bool = True): def fix_json(json_str: str, schema: str, debug=False) -> str: # Try to fix the JSON using gpt: function_string = "def fix_json(json_str: str, schema:str=None) -> str:" - args = [json_str, schema] + args = [f"'''{json_str}'''", f"'''{schema}'''"] description_string = """Fixes the provided JSON string to make it parseable and fully complient with the provided schema.\n If an object or field specifed in the schema isn't contained within the correct JSON, it is ommited.\n This function is brilliant at guessing when the format is incorrect.""" # If it doesn't already start with a "`", add one: @@ -67,7 +68,8 @@ def fix_json(json_str: str, schema: str, debug=False) -> str: print(f"Fixed JSON: {result_string}") print("----------- END OF FIX ATTEMPT ----------------") try: - return json.loads(result_string) + json.loads(result_string) # just check the validity + return result_string except: # Get the call stack: # import traceback diff --git a/scripts/llm_utils.py b/scripts/llm_utils.py index 41f396250..5a471ab7a 100644 --- a/scripts/llm_utils.py +++ b/scripts/llm_utils.py @@ -6,11 +6,20 @@ openai.api_key = cfg.openai_api_key # Overly simple abstraction until we create something better def create_chat_completion(messages, model=None, temperature=None, max_tokens=None)->str: - response = openai.ChatCompletion.create( - model=model, - messages=messages, - temperature=temperature, - max_tokens=max_tokens - ) + if cfg.use_azure: + response = openai.ChatCompletion.create( + deployment_id=cfg.openai_deployment_id, + model=model, + messages=messages, + temperature=temperature, + max_tokens=max_tokens + ) + else: + response = openai.ChatCompletion.create( + model=model, + messages=messages, + temperature=temperature, + max_tokens=max_tokens + ) return response.choices[0].message["content"] diff --git a/scripts/main.py b/scripts/main.py index 93124234e..a79fd553c 100644 --- a/scripts/main.py +++ b/scripts/main.py @@ -1,7 +1,7 @@ import json import random import commands as cmd -import memory as mem +from memory import PineconeMemory import data import chat from colorama import Fore, Style @@ -277,9 +277,17 @@ prompt = construct_prompt() # Initialize variables full_message_history = [] result = None +next_action_count = 0 # Make a constant: user_input = "Determine which next command to use, and respond using the format specified above:" +# Initialize memory and make sure it is empty. +# this is particularly important for indexing and referencing pinecone memory +memory = PineconeMemory() +memory.clear() + +print('Using memory of type: ' + memory.__class__.__name__) + # Interaction Loop while True: # Send message to AI, get response @@ -288,10 +296,9 @@ while True: prompt, user_input, full_message_history, - mem.permanent_memory, + memory, cfg.fast_token_limit) # TODO: This hardcodes the model to use GPT3.5. Make this an argument - # print("assistant reply: "+assistant_reply) # Print Assistant thoughts print_assistant_thoughts(assistant_reply) @@ -301,7 +308,7 @@ while True: except Exception as e: print_to_console("Error: \n", Fore.RED, str(e)) - if not cfg.continuous_mode: + if not cfg.continuous_mode and next_action_count == 0: ### GET USER AUTHORIZATION TO EXECUTE COMMAND ### # Get key press: Prompt the user to press enter to continue or escape # to exit @@ -311,13 +318,21 @@ while True: Fore.CYAN, f"COMMAND = {Fore.CYAN}{command_name}{Style.RESET_ALL} ARGUMENTS = {Fore.CYAN}{arguments}{Style.RESET_ALL}") print( - f"Enter 'y' to authorise command or 'n' to exit program, or enter feedback for {ai_name}...", + f"Enter 'y' to authorise command, 'y -N' to run N continuous commands, 'n' to exit program, or enter feedback for {ai_name}...", flush=True) while True: console_input = input(Fore.MAGENTA + "Input:" + Style.RESET_ALL) if console_input.lower() == "y": user_input = "GENERATE NEXT COMMAND JSON" break + elif console_input.lower().startswith("y -"): + try: + next_action_count = abs(int(console_input.split(" ")[1])) + user_input = "GENERATE NEXT COMMAND JSON" + except ValueError: + print("Invalid input format. Please enter 'y -n' where n is the number of continuous tasks.") + continue + break elif console_input.lower() == "n": user_input = "EXIT" break @@ -348,6 +363,14 @@ while True: result = f"Human feedback: {user_input}" else: result = f"Command {command_name} returned: {cmd.execute_command(command_name, arguments)}" + if next_action_count > 0: + next_action_count -= 1 + + memory_to_add = f"Assistant Reply: {assistant_reply} " \ + f"\nResult: {result} " \ + f"\nHuman Feedback: {user_input} " + + memory.add(memory_to_add) # Check if there's a result from the command append it to the message # history diff --git a/scripts/memory.py b/scripts/memory.py index 0dc5b7666..0d265a31d 100644 --- a/scripts/memory.py +++ b/scripts/memory.py @@ -1 +1,61 @@ -permanent_memory = [] +from config import Config, Singleton +import pinecone +import openai + +cfg = Config() + + +def get_ada_embedding(text): + text = text.replace("\n", " ") + return openai.Embedding.create(input=[text], model="text-embedding-ada-002")["data"][0]["embedding"] + + +def get_text_from_embedding(embedding): + return openai.Embedding.retrieve(embedding, model="text-embedding-ada-002")["data"][0]["text"] + + +class PineconeMemory(metaclass=Singleton): + def __init__(self): + pinecone_api_key = cfg.pinecone_api_key + pinecone_region = cfg.pinecone_region + pinecone.init(api_key=pinecone_api_key, environment=pinecone_region) + dimension = 1536 + metric = "cosine" + pod_type = "p1" + table_name = "auto-gpt" + # this assumes we don't start with memory. + # for now this works. + # we'll need a more complicated and robust system if we want to start with memory. + self.vec_num = 0 + if table_name not in pinecone.list_indexes(): + pinecone.create_index(table_name, dimension=dimension, metric=metric, pod_type=pod_type) + self.index = pinecone.Index(table_name) + + def add(self, data): + vector = get_ada_embedding(data) + # no metadata here. We may wish to change that long term. + resp = self.index.upsert([(str(self.vec_num), vector, {"raw_text": data})]) + _text = f"Inserting data into memory at index: {self.vec_num}:\n data: {data}" + self.vec_num += 1 + return _text + + def get(self, data): + return self.get_relevant(data, 1) + + def clear(self): + self.index.delete(deleteAll=True) + return "Obliviated" + + def get_relevant(self, data, num_relevant=5): + """ + Returns all the data in the memory that is relevant to the given data. + :param data: The data to compare to. + :param num_relevant: The number of relevant data to return. Defaults to 5 + """ + query_embedding = get_ada_embedding(data) + results = self.index.query(query_embedding, top_k=num_relevant, include_metadata=True) + sorted_results = sorted(results.matches, key=lambda x: x.score) + return [str(item['metadata']["raw_text"]) for item in sorted_results] + + def get_stats(self): + return self.index.describe_index_stats() diff --git a/scripts/speak.py b/scripts/speak.py index 0f8c243d5..d8dd191bc 100644 --- a/scripts/speak.py +++ b/scripts/speak.py @@ -46,6 +46,7 @@ def gtts_speech(text): os.remove("speech.mp3") def say_text(text, voice_index=0): + def speak(): if not cfg.elevenlabs_api_key: gtts_speech(text)