Remove openai package

memgpt/agent.py

@@ -8,7 +8,7 @@
 from memgpt.config import AgentConfig, MemGPTConfig
 from .system import get_login_event, package_function_response, package_summarize_message, get_initial_boot_messages
 from .memory import CoreMemory as Memory, summarize_messages
-from .openai_tools import completions_with_backoff as create
+from .openai_tools import completions_with_backoff as create, is_context_overflow_error
 from memgpt.openai_tools import chat_completion_with_backoff
 from .utils import get_local_time, parse_json, united_diff, printd, count_tokens, get_schema_diff
 from .constants import (
@@ -649,14 +649,14 @@ def step(self, user_message, first_message=False, first_message_retry_limit=FIRS
             printd(f"step() failed\nuser_message = {user_message}\nerror = {e}")
 
             # If we got a context alert, try trimming the messages length, then try again
-            if "maximum context length" in str(e):
+            if is_context_overflow_error(e):
                 # A separate API call to run a summarizer
                 self.summarize_messages_inplace()
 
                 # Try step again
                 return self.step(user_message, first_message=first_message)
             else:
-                printd(f"step() failed with openai.InvalidRequestError, but didn't recognize the error message: '{str(e)}'")
+                printd(f"step() failed with an unrecognized exception: '{str(e)}'")
                 raise e
 
     def summarize_messages_inplace(self, cutoff=None, preserve_last_N_messages=True):

memgpt/cli/cli.py

@@ -6,7 +6,6 @@
 import os
 from prettytable import PrettyTable
 import questionary
-import openai
 
 from llama_index import set_global_service_context
 from llama_index import VectorStoreIndex, SimpleDirectoryReader, ServiceContext

memgpt/cli/cli_config.py

@@ -1,6 +1,5 @@
 import builtins
 import questionary
-import openai
 from prettytable import PrettyTable
 import typer
 import os

memgpt/embeddings.py

@@ -105,9 +105,12 @@ def embedding_model():
         )
         return model
     elif endpoint == "azure":
+        # https://learn.microsoft.com/en-us/azure/ai-services/openai/reference#embeddings
+        model = "text-embedding-ada-002"
+        deployment = config.azure_embedding_deployment if config.azure_embedding_deployment is not None else model
         return OpenAIEmbedding(
-            model="text-embedding-ada-002",
-            deployment_name=config.azure_embedding_deployment,
+            model=model,
+            deployment_name=deployment,
             api_key=config.azure_key,
             api_base=config.azure_endpoint,
             api_type="azure",

memgpt/local_llm/chat_completion_proxy.py

@@ -4,18 +4,20 @@
 import requests
 import json
 
-from .webui.api import get_webui_completion
-from .webui.legacy_api import get_webui_completion as get_webui_completion_legacy
-from .lmstudio.api import get_lmstudio_completion
-from .llamacpp.api import get_llamacpp_completion
-from .koboldcpp.api import get_koboldcpp_completion
-from .ollama.api import get_ollama_completion
-from .vllm.api import get_vllm_completion
-from .llm_chat_completion_wrappers import airoboros, dolphin, zephyr, simple_summary_wrapper
-from .constants import DEFAULT_WRAPPER
-from .utils import DotDict, get_available_wrappers
-from ..prompts.gpt_summarize import SYSTEM as SUMMARIZE_SYSTEM_MESSAGE
-from ..errors import LocalLLMConnectionError, LocalLLMError
+from box import Box
+
+from memgpt.local_llm.webui.api import get_webui_completion
+from memgpt.local_llm.webui.legacy_api import get_webui_completion as get_webui_completion_legacy
+from memgpt.local_llm.lmstudio.api import get_lmstudio_completion
+from memgpt.local_llm.llamacpp.api import get_llamacpp_completion
+from memgpt.local_llm.koboldcpp.api import get_koboldcpp_completion
+from memgpt.local_llm.ollama.api import get_ollama_completion
+from memgpt.local_llm.vllm.api import get_vllm_completion
+from memgpt.local_llm.llm_chat_completion_wrappers import simple_summary_wrapper
+from memgpt.local_llm.constants import DEFAULT_WRAPPER
+from memgpt.local_llm.utils import get_available_wrappers
+from memgpt.prompts.gpt_summarize import SYSTEM as SUMMARIZE_SYSTEM_MESSAGE
+from memgpt.errors import LocalLLMConnectionError, LocalLLMError
 
 endpoint = os.getenv("OPENAI_API_BASE")
 endpoint_type = os.getenv("BACKEND_TYPE")  # default None == ChatCompletion
@@ -119,25 +121,21 @@ def get_chat_completion(
         raise LocalLLMError(f"Failed to parse JSON from local LLM response - error: {str(e)}")
 
     # unpack with response.choices[0].message.content
-    response = DotDict(
+    response = Box(
         {
             "model": model,
             "choices": [
-                DotDict(
-                    {
-                        "message": DotDict(chat_completion_result),
-                        "finish_reason": "stop",  # TODO vary based on backend response
-                    }
-                )
-            ],
-            "usage": DotDict(
                 {
-                    # TODO fix, actually use real info
-                    "prompt_tokens": 0,
-                    "completion_tokens": 0,
-                    "total_tokens": 0,
+                    "message": chat_completion_result,
+                    "finish_reason": "stop",  # TODO vary based on backend response
                 }
-            ),
+            ],
+            "usage": {
+                # TODO fix, actually use real info
+                "prompt_tokens": 0,
+                "completion_tokens": 0,
+                "total_tokens": 0,
+            },
         }
     )
     return response

memgpt/local_llm/utils.py

@@ -6,23 +6,6 @@
 import memgpt.local_llm.llm_chat_completion_wrappers.zephyr as zephyr
 
 
-class DotDict(dict):
-    """Allow dot access on properties similar to OpenAI response object"""
-
-    def __getattr__(self, attr):
-        return self.get(attr)
-
-    def __setattr__(self, key, value):
-        self[key] = value
-
-    # following methods necessary for pickling
-    def __getstate__(self):
-        return vars(self)
-
-    def __setstate__(self, state):
-        vars(self).update(state)
-
-
 def load_grammar_file(grammar):
     # Set grammar
     grammar_file = os.path.join(os.path.dirname(os.path.abspath(__file__)), "grammars", f"{grammar}.gbnf")

memgpt/main.py

@@ -614,6 +614,11 @@ def process_agent_step(user_message, no_verify):
                     with console.status("[bold cyan]Thinking...") as status:
                         new_messages, user_message, skip_next_user_input = process_agent_step(user_message, no_verify)
                         break
+            except KeyboardInterrupt:
+                print("User interrupt occured.")
+                retry = questionary.confirm("Retry agent.step()?").ask()
+                if not retry:
+                    break
             except Exception as e:
                 print("An exception ocurred when running agent.step(): ")
                 traceback.print_exc()