Organize all the llm stuff into a subpackage (Significant-Gravitas#3436)

autogpt/agent/agent.py

@@ -1,11 +1,10 @@
 from colorama import Fore, Style
 
 from autogpt.app import execute_command, get_command
-from autogpt.chat import chat_with_ai, create_chat_message
 from autogpt.config import Config
 from autogpt.json_utils.json_fix_llm import fix_json_using_multiple_techniques
 from autogpt.json_utils.utilities import validate_json
-from autogpt.llm_utils import create_chat_completion
+from autogpt.llm import chat_with_ai, create_chat_completion, create_chat_message
 from autogpt.logs import logger, print_assistant_thoughts
 from autogpt.speech import say_text
 from autogpt.spinner import Spinner

autogpt/agent/agent_manager.py

@@ -4,7 +4,7 @@
 from typing import List
 
 from autogpt.config.config import Config
-from autogpt.llm_utils import create_chat_completion
+from autogpt.llm import create_chat_completion
 from autogpt.singleton import Singleton
 from autogpt.types.openai import Message
 

autogpt/commands/analyze_code.py

@@ -2,7 +2,7 @@
 from __future__ import annotations
 
 from autogpt.commands.command import command
-from autogpt.llm_utils import call_ai_function
+from autogpt.llm import call_ai_function
 
 
 @command(

autogpt/commands/improve_code.py

@@ -3,7 +3,7 @@
 import json
 
 from autogpt.commands.command import command
-from autogpt.llm_utils import call_ai_function
+from autogpt.llm import call_ai_function
 
 
 @command(

autogpt/commands/write_tests.py

@@ -4,7 +4,7 @@
 import json
 
 from autogpt.commands.command import command
-from autogpt.llm_utils import call_ai_function
+from autogpt.llm import call_ai_function
 
 
 @command(

autogpt/json_utils/json_fix_llm.py

@@ -11,7 +11,7 @@
 
 from autogpt.config import Config
 from autogpt.json_utils.json_fix_general import correct_json
-from autogpt.llm_utils import call_ai_function
+from autogpt.llm import call_ai_function
 from autogpt.logs import logger
 from autogpt.speech import say_text
 

autogpt/llm/__init__.py

@@ -0,0 +1,22 @@
+from autogpt.llm.api_manager import ApiManager
+from autogpt.llm.chat import chat_with_ai, create_chat_message, generate_context
+from autogpt.llm.llm_utils import (
+    call_ai_function,
+    create_chat_completion,
+    get_ada_embedding,
+)
+from autogpt.llm.modelsinfo import COSTS
+from autogpt.llm.token_counter import count_message_tokens, count_string_tokens
+
+__all__ = [
+    "ApiManager",
+    "create_chat_message",
+    "generate_context",
+    "chat_with_ai",
+    "call_ai_function",
+    "create_chat_completion",
+    "get_ada_embedding",
+    "COSTS",
+    "count_message_tokens",
+    "count_string_tokens",
+]

autogpt/api_manager.py → autogpt/llm/api_manager.py

@@ -3,8 +3,8 @@
 import openai
 
 from autogpt.config import Config
+from autogpt.llm.modelsinfo import COSTS
 from autogpt.logs import logger
-from autogpt.modelsinfo import COSTS
 from autogpt.singleton import Singleton
 
 

autogpt/chat.py → autogpt/llm/chat.py

@@ -3,10 +3,10 @@
 
 from openai.error import RateLimitError
 
-from autogpt import token_counter
-from autogpt.api_manager import ApiManager
 from autogpt.config import Config
-from autogpt.llm_utils import create_chat_completion
+from autogpt.llm.api_manager import ApiManager
+from autogpt.llm.llm_utils import create_chat_completion
+from autogpt.llm.token_counter import count_message_tokens
 from autogpt.logs import logger
 from autogpt.types.openai import Message
 
@@ -43,7 +43,7 @@ def generate_context(prompt, relevant_memory, full_message_history, model):
     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)
+    current_tokens_used = count_message_tokens(current_context, model)
     return (
         next_message_to_add_index,
         current_tokens_used,
@@ -114,17 +114,15 @@ def chat_with_ai(
                     prompt, relevant_memory, full_message_history, model
                 )
 
-            current_tokens_used += token_counter.count_message_tokens(
+            current_tokens_used += count_message_tokens(
                 [create_chat_message("user", user_input)], model
             )  # Account for user input (appended later)
 
             while next_message_to_add_index >= 0:
                 # print (f"CURRENT TOKENS USED: {current_tokens_used}")
                 message_to_add = full_message_history[next_message_to_add_index]
 
-                tokens_to_add = token_counter.count_message_tokens(
-                    [message_to_add], model
-                )
+                tokens_to_add = count_message_tokens([message_to_add], model)
                 if current_tokens_used + tokens_to_add > send_token_limit:
                     break
 
@@ -175,7 +173,7 @@ def chat_with_ai(
                 )
                 if not plugin_response or plugin_response == "":
                     continue
-                tokens_to_add = token_counter.count_message_tokens(
+                tokens_to_add = count_message_tokens(
                     [create_chat_message("system", plugin_response)], model
                 )
                 if current_tokens_used + tokens_to_add > send_token_limit:

autogpt/llm_utils.py → autogpt/llm/llm_utils.py

@@ -8,8 +8,8 @@
 from colorama import Fore, Style
 from openai.error import APIError, RateLimitError, Timeout
 
-from autogpt.api_manager import ApiManager
 from autogpt.config import Config
+from autogpt.llm.api_manager import ApiManager
 from autogpt.logs import logger
 from autogpt.types.openai import Message
 

autogpt/logs.py

@@ -1,11 +1,9 @@
 """Logging module for Auto-GPT."""
-import json
 import logging
 import os
 import random
 import re
 import time
-import traceback
 from logging import LogRecord
 
 from colorama import Fore, Style

autogpt/memory/local.py

@@ -7,7 +7,7 @@
 import numpy as np
 import orjson
 
-from autogpt.llm_utils import get_ada_embedding
+from autogpt.llm import get_ada_embedding
 from autogpt.memory.base import MemoryProviderSingleton
 
 EMBED_DIM = 1536

autogpt/memory/milvus.py

@@ -4,7 +4,7 @@
 from pymilvus import Collection, CollectionSchema, DataType, FieldSchema, connections
 
 from autogpt.config import Config
-from autogpt.llm_utils import get_ada_embedding
+from autogpt.llm import get_ada_embedding
 from autogpt.memory.base import MemoryProviderSingleton
 
 

autogpt/memory/pinecone.py

@@ -1,7 +1,7 @@
 import pinecone
 from colorama import Fore, Style
 
-from autogpt.llm_utils import get_ada_embedding
+from autogpt.llm import get_ada_embedding
 from autogpt.logs import logger
 from autogpt.memory.base import MemoryProviderSingleton
 

autogpt/memory/redismem.py

@@ -10,7 +10,7 @@
 from redis.commands.search.indexDefinition import IndexDefinition, IndexType
 from redis.commands.search.query import Query
 
-from autogpt.llm_utils import get_ada_embedding
+from autogpt.llm import get_ada_embedding
 from autogpt.logs import logger
 from autogpt.memory.base import MemoryProviderSingleton
 

autogpt/memory/weaviate.py

@@ -3,7 +3,7 @@
 from weaviate.embedded import EmbeddedOptions
 from weaviate.util import generate_uuid5
 
-from autogpt.llm_utils import get_ada_embedding
+from autogpt.llm import get_ada_embedding
 from autogpt.memory.base import MemoryProviderSingleton
 
 

autogpt/processing/text.py

@@ -4,9 +4,8 @@
 import spacy
 from selenium.webdriver.remote.webdriver import WebDriver
 
-from autogpt import token_counter
 from autogpt.config import Config
-from autogpt.llm_utils import create_chat_completion
+from autogpt.llm import count_message_tokens, create_chat_completion
 from autogpt.memory import get_memory
 
 CFG = Config()
@@ -44,7 +43,7 @@ def split_text(
         ]
 
         expected_token_usage = (
-            token_usage_of_chunk(messages=message_with_additional_sentence, model=model)
+            count_message_tokens(messages=message_with_additional_sentence, model=model)
             + 1
         )
         if expected_token_usage <= max_length:
@@ -56,7 +55,7 @@ def split_text(
                 create_message(" ".join(current_chunk), question)
             ]
             expected_token_usage = (
-                token_usage_of_chunk(messages=message_this_sentence_only, model=model)
+                count_message_tokens(messages=message_this_sentence_only, model=model)
                 + 1
             )
             if expected_token_usage > max_length:
@@ -68,10 +67,6 @@ def split_text(
         yield " ".join(current_chunk)
 
 
-def token_usage_of_chunk(messages, model):
-    return token_counter.count_message_tokens(messages, model)
-
-
 def summarize_text(
     url: str, text: str, question: str, driver: Optional[WebDriver] = None
 ) -> str:
@@ -112,7 +107,7 @@ def summarize_text(
         memory.add(memory_to_add)
 
         messages = [create_message(chunk, question)]
-        tokens_for_chunk = token_counter.count_message_tokens(messages, model)
+        tokens_for_chunk = count_message_tokens(messages, model)
         print(
             f"Summarizing chunk {i + 1} / {len(chunks)} of length {len(chunk)} characters, or {tokens_for_chunk} tokens"
         )

autogpt/prompts/prompt.py

@@ -1,8 +1,8 @@
 from colorama import Fore
 
-from autogpt.api_manager import ApiManager
 from autogpt.config.ai_config import AIConfig
 from autogpt.config.config import Config
+from autogpt.llm import ApiManager
 from autogpt.logs import logger
 from autogpt.prompts.generator import PromptGenerator
 from autogpt.setup import prompt_user

autogpt/setup.py

@@ -6,7 +6,7 @@
 from autogpt import utils
 from autogpt.config import Config
 from autogpt.config.ai_config import AIConfig
-from autogpt.llm_utils import create_chat_completion
+from autogpt.llm import create_chat_completion
 from autogpt.logs import logger
 
 CFG = Config()

tests/conftest.py

@@ -3,8 +3,8 @@
 import pytest
 from pytest_mock import MockerFixture
 
-from autogpt.api_manager import ApiManager
 from autogpt.config import Config
+from autogpt.llm import ApiManager
 from autogpt.workspace import Workspace
 
 pytest_plugins = ["tests.integration.agent_factory"]

tests/integration/test_llm_utils.py

@@ -3,7 +3,7 @@
 import pytest
 from numpy.random import RandomState
 
-from autogpt.llm_utils import get_ada_embedding
+from autogpt.llm.llm_utils import get_ada_embedding
 from tests.utils import requires_api_key
 
 

tests/integration/weaviate_memory_tests.py

@@ -5,7 +5,7 @@
 from weaviate.util import get_valid_uuid
 
 from autogpt.config import Config
-from autogpt.llm_utils import get_ada_embedding
+from autogpt.llm import get_ada_embedding
 from autogpt.memory.weaviate import WeaviateMemory
 
 

tests/test_agent.py

@@ -3,10 +3,7 @@
 import pytest
 
 from autogpt.agent import Agent
-from autogpt.chat import chat_with_ai
 from autogpt.config import Config
-from autogpt.speech import say_text
-from autogpt.utils import clean_input
 
 
 @pytest.fixture

tests/test_agent_manager.py

@@ -1,7 +1,7 @@
 import pytest
 
 from autogpt.agent.agent_manager import AgentManager
-from autogpt.llm_utils import create_chat_completion
+from autogpt.llm import create_chat_completion
 
 
 @pytest.fixture

tests/test_api_manager.py

@@ -2,7 +2,7 @@
 
 import pytest
 
-from autogpt.api_manager import COSTS, ApiManager
+from autogpt.llm import COSTS, ApiManager
 
 api_manager = ApiManager()
 

tests/test_llm_utils.py

@@ -1,8 +1,8 @@
 import pytest
 from openai.error import APIError, RateLimitError
 
-from autogpt.llm_utils import get_ada_embedding, retry_openai_api
-from autogpt.modelsinfo import COSTS
+from autogpt.llm import COSTS, get_ada_embedding
+from autogpt.llm.llm_utils import retry_openai_api
 
 
 @pytest.fixture(params=[RateLimitError, APIError])
@@ -19,7 +19,7 @@ def mock_create_embedding(mocker):
     mock_response.usage.prompt_tokens = 5
     mock_response.__getitem__.side_effect = lambda key: [{"embedding": [0.1, 0.2, 0.3]}]
     return mocker.patch(
-        "autogpt.llm_utils.create_embedding", return_value=mock_response
+        "autogpt.llm.llm_utils.create_embedding", return_value=mock_response
     )
 
 

tests/test_token_counter.py

@@ -1,7 +1,6 @@
 import pytest
 
-import tests.context
-from autogpt.token_counter import count_message_tokens, count_string_tokens
+from autogpt.llm import count_message_tokens, count_string_tokens
 
 
 def test_count_message_tokens():

tests/unit/test_chat.py

@@ -2,7 +2,7 @@
 import time
 from unittest.mock import patch
 
-from autogpt.chat import create_chat_message, generate_context
+from autogpt.llm import create_chat_message, generate_context
 
 
 def test_happy_path_role_content():

tests/unit/test_get_self_feedback.py

@@ -1,6 +1,6 @@
 from autogpt.agent.agent import Agent
 from autogpt.config import AIConfig
-from autogpt.llm_utils import create_chat_completion
+from autogpt.llm import create_chat_completion
 
 
 def test_get_self_feedback(mocker):