Dev/v0.2

.github/workflows/build.yml

@@ -40,7 +40,7 @@ jobs:
           python -m pip install --upgrade pip wheel
           pip install -e .
           python -c "import autogen"
-          pip install -e.[mathchat,retrievechat,test] datasets pytest
+          pip install -e. pytest
           pip uninstall -y openai
       - name: Test with pytest
         if: matrix.python-version != '3.10'
@@ -49,7 +49,8 @@ jobs:
       - name: Coverage
         if: matrix.python-version == '3.10'
         run: |
-          pip install coverage
+          pip install -e.[mathchat,test]
+          pip uninstall -y openai
           coverage run -a -m pytest test
           coverage xml
       - name: Upload coverage to Codecov

.github/workflows/openai.yml

@@ -4,16 +4,13 @@
 name: OpenAI
 
 on:
-  pull_request_target:
+  pull_request:
     branches: ['main']
     paths:
       - 'autogen/**'
       - 'test/**'
       - 'notebook/agentchat_auto_feedback_from_code_execution.ipynb'
       - 'notebook/agentchat_function_call.ipynb'
-      - 'notebook/agentchat_MathChat.ipynb'
-      - 'notebook/oai_completion.ipynb'
-      - 'notebook/oai_chatgpt_gpt4.ipynb'
       - '.github/workflows/openai.yml'
 
 jobs:
@@ -23,7 +20,7 @@ jobs:
         os: [ubuntu-latest]
         python-version: ["3.9", "3.10", "3.11"]
     runs-on: ${{ matrix.os }}
-    environment: openai
+    environment: openai1
     steps:
       # checkout to pr branch
       - name: Checkout
@@ -38,28 +35,13 @@ jobs:
         run: |
           docker --version
           python -m pip install --upgrade pip wheel
-          pip install -e.[blendsearch]
+          pip install -e.
           python -c "import autogen"
-          pip install coverage pytest-asyncio datasets
+          pip install coverage pytest-asyncio
       - name: Install packages for test when needed
         if: matrix.python-version == '3.9'
         run: |
           pip install docker
-      - name: Install packages for MathChat when needed
-        if: matrix.python-version != '3.11'
-        run: |
-          pip install -e .[mathchat]
-      - name: Install packages for RetrieveChat when needed
-        if: matrix.python-version == '3.9'
-        run: |
-          pip install -e .[retrievechat]
-      - name: Install packages for Teachable when needed
-        run: |
-          pip install -e .[teachable]
-      - name: Install packages for RetrieveChat with QDrant when needed
-        if: matrix.python-version == '3.11'
-        run: |
-          pip install -e .[retrievechat] qdrant_client[fastembed]
       - name: Coverage
         if: matrix.python-version == '3.9'
         env:
@@ -80,8 +62,7 @@ jobs:
           OAI_CONFIG_LIST: ${{ secrets.OAI_CONFIG_LIST }}
         run: |
           pip install nbconvert nbformat ipykernel
-          coverage run -a -m pytest test/agentchat/test_qdrant_retrievechat.py
-          coverage run -a -m pytest test/test_with_openai.py
+          coverage run -a -m pytest test/agentchat/test_function_call_groupchat.py
           coverage run -a -m pytest test/test_notebook.py
           coverage xml
           cat "$(pwd)/test/executed_openai_notebook_output.txt"

OAI_CONFIG_LIST_sample

@@ -7,14 +7,14 @@
     {
         "model": "gpt-4",
         "api_key": "<your Azure OpenAI API key here>",
-        "api_base": "<your Azure OpenAI API base here>",
+        "base_url": "<your Azure OpenAI API base here>",
         "api_type": "azure",
         "api_version": "2023-07-01-preview"
     },
     {
         "model": "gpt-3.5-turbo",
         "api_key": "<your Azure OpenAI API key here>",
-        "api_base": "<your Azure OpenAI API base here>",
+        "base_url": "<your Azure OpenAI API base here>",
         "api_type": "azure",
         "api_version": "2023-07-01-preview"
     }

README.md

@@ -113,7 +113,7 @@ Please find more [code examples](https://microsoft.github.io/autogen/docs/Exampl
 Autogen also helps maximize the utility out of the expensive LLMs such as ChatGPT and GPT-4. It offers enhanced LLM inference with powerful functionalities like tuning, caching, error handling, and templating. For example, you can optimize generations by LLM with your own tuning data, success metrics, and budgets.
 
 ```python
-# perform tuning
+# perform tuning for openai<1
 config, analysis = autogen.Completion.tune(
     data=tune_data,
     metric="success",

autogen/agentchat/assistant_agent.py

@@ -43,7 +43,7 @@ def __init__(
             system_message (str): system message for the ChatCompletion inference.
                 Please override this attribute if you want to reprogram the agent.
             llm_config (dict): llm inference configuration.
-                Please refer to [Completion.create](/docs/reference/oai/completion#create)
+                Please refer to [OpenAIWrapper.create](/docs/reference/oai/client#create)
                 for available options.
             is_termination_msg (function): a function that takes a message in the form of a dictionary
                 and returns a boolean value indicating if this received message is a termination message.

autogen/agentchat/contrib/teachable_agent.py

@@ -18,7 +18,7 @@ def colored(x, *args, **kwargs):
 
 
 class TeachableAgent(ConversableAgent):
-    """Teachable Agent, a subclass of ConversableAgent using a vector database to remember user teachings.
+    """(Experimental) Teachable Agent, a subclass of ConversableAgent using a vector database to remember user teachings.
     In this class, the term 'user' refers to any caller (human or not) sending messages to this agent.
     Not yet tested in the group-chat setting."""
 
@@ -40,7 +40,7 @@ def __init__(
             system_message (str): system message for the ChatCompletion inference.
             human_input_mode (str): This agent should NEVER prompt the human for input.
             llm_config (dict or False): llm inference configuration.
-                Please refer to [Completion.create](/docs/reference/oai/completion#create)
+                Please refer to [OpenAIWrapper.create](/docs/reference/oai/client#create)
                 for available options.
                 To disable llm-based auto reply, set to False.
             analyzer_llm_config (dict or False): llm inference configuration passed to TextAnalyzerAgent.
@@ -125,11 +125,8 @@ def _generate_teachable_assistant_reply(
                 messages = messages.copy()
                 messages[-1]["content"] = new_user_text
 
-        # Generate a response.
-        msgs = self._oai_system_message + messages
-        response = oai.ChatCompletion.create(messages=msgs, **self.llm_config)
-        response_text = oai.ChatCompletion.extract_text_or_function_call(response)[0]
-        return True, response_text
+        # Generate a response by reusing existing generate_oai_reply
+        return self.generate_oai_reply(messages, sender, config)
 
     def learn_from_user_feedback(self):
         """Reviews the user comments from the last chat, and decides what teachings to store as memos."""
@@ -265,12 +262,14 @@ def analyze(self, text_to_analyze, analysis_instructions):
             self.send(recipient=self.analyzer, message=analysis_instructions, request_reply=True)  # Request the reply.
             return self.last_message(self.analyzer)["content"]
         else:
+            # TODO: This is not an encouraged usage pattern. It breaks the conversation-centric design.
+            # consider using the arg "silent"
             # Use the analyzer's method directly, to leave analyzer message out of the printed chat.
             return self.analyzer.analyze_text(text_to_analyze, analysis_instructions)
 
 
 class MemoStore:
-    """
+    """(Experimental)
     Provides memory storage and retrieval for a TeachableAgent, using a vector database.
     Each DB entry (called a memo) is a pair of strings: an input text and an output text.
     The input text might be a question, or a task to perform.

autogen/agentchat/contrib/text_analyzer_agent.py

@@ -10,7 +10,7 @@
 
 
 class TextAnalyzerAgent(ConversableAgent):
-    """Text Analysis agent, a subclass of ConversableAgent designed to analyze text as instructed."""
+    """(Experimental) Text Analysis agent, a subclass of ConversableAgent designed to analyze text as instructed."""
 
     def __init__(
         self,
@@ -26,7 +26,7 @@ def __init__(
             system_message (str): system message for the ChatCompletion inference.
             human_input_mode (str): This agent should NEVER prompt the human for input.
             llm_config (dict or False): llm inference configuration.
-                Please refer to [Completion.create](/docs/reference/oai/completion#create)
+                Please refer to [OpenAIWrapper.create](/docs/reference/oai/client#create)
                 for available options.
                 To disable llm-based auto reply, set to False.
             teach_config (dict or None): Additional parameters used by TeachableAgent.
@@ -74,9 +74,5 @@ def analyze_text(self, text_to_analyze, analysis_instructions):
         msg_text = "\n".join(
             [analysis_instructions, text_to_analyze, analysis_instructions]
         )  # Repeat the instructions.
-        messages = self._oai_system_message + [{"role": "user", "content": msg_text}]
-
         # Generate and return the analysis string.
-        response = oai.ChatCompletion.create(context=None, messages=messages, **self.llm_config)
-        output_text = oai.ChatCompletion.extract_text_or_function_call(response)[0]
-        return output_text
+        return self.generate_oai_reply([{"role": "user", "content": msg_text}], None, None)[1]

autogen/agentchat/conversable_agent.py

@@ -4,7 +4,7 @@
 import json
 import logging
 from typing import Any, Callable, Dict, List, Optional, Tuple, Type, Union
-from autogen import oai
+from autogen import OpenAIWrapper
 from .agent import Agent
 from autogen.code_utils import (
     DEFAULT_MODEL,
@@ -93,7 +93,7 @@ def __init__(
                 - timeout (Optional, int): The maximum execution time in seconds.
                 - last_n_messages (Experimental, Optional, int): The number of messages to look back for code execution. Default to 1.
             llm_config (dict or False): llm inference configuration.
-                Please refer to [Completion.create](/docs/reference/oai/completion#create)
+                Please refer to [OpenAIWrapper.create](/docs/reference/oai/client#create)
                 for available options.
                 To disable llm-based auto reply, set to False.
             default_auto_reply (str or dict or None): default auto reply when no code execution or llm-based reply is generated.
@@ -107,10 +107,12 @@ def __init__(
         )
         if llm_config is False:
             self.llm_config = False
+            self.client = None
         else:
             self.llm_config = self.DEFAULT_CONFIG.copy()
             if isinstance(llm_config, dict):
                 self.llm_config.update(llm_config)
+            self.client = OpenAIWrapper(**self.llm_config)
 
         self._code_execution_config = {} if code_execution_config is None else code_execution_config
         self.human_input_mode = human_input_mode
@@ -254,8 +256,10 @@ def _message_to_dict(message: Union[Dict, str]):
         """
         if isinstance(message, str):
             return {"content": message}
-        else:
+        elif isinstance(message, dict):
             return message
+        else:
+            return dict(message)
 
     def _append_oai_message(self, message: Union[Dict, str], role, conversation_id: Agent) -> bool:
         """Append a message to the ChatCompletion conversation.
@@ -285,6 +289,7 @@ def _append_oai_message(self, message: Union[Dict, str], role, conversation_id:
         oai_message["role"] = "function" if message.get("role") == "function" else role
         if "function_call" in oai_message:
             oai_message["role"] = "assistant"  # only messages with role 'assistant' can have a function call.
+            oai_message["function_call"] = dict(oai_message["function_call"])
         self._oai_messages[conversation_id].append(oai_message)
         return True
 
@@ -306,7 +311,7 @@ def send(
                 - role (str): the role of the message, any role that is not "function"
                     will be modified to "assistant".
                 - context (dict): the context of the message, which will be passed to
-                    [Completion.create](../oai/Completion#create).
+                    [OpenAIWrapper.create](../oai/client#create).
                     For example, one agent can send a message A as:
         ```python
         {
@@ -355,7 +360,7 @@ async def a_send(
                 - role (str): the role of the message, any role that is not "function"
                     will be modified to "assistant".
                 - context (dict): the context of the message, which will be passed to
-                    [Completion.create](../oai/Completion#create).
+                    [OpenAIWrapper.create](../oai/client#create).
                     For example, one agent can send a message A as:
         ```python
         {
@@ -398,18 +403,21 @@ def _print_received_message(self, message: Union[Dict, str], sender: Agent):
             content = message.get("content")
             if content is not None:
                 if "context" in message:
-                    content = oai.ChatCompletion.instantiate(
+                    content = OpenAIWrapper.instantiate(
                         content,
                         message["context"],
                         self.llm_config and self.llm_config.get("allow_format_str_template", False),
                     )
                 print(content, flush=True)
             if "function_call" in message:
-                func_print = f"***** Suggested function Call: {message['function_call'].get('name', '(No function name found)')} *****"
+                function_call = dict(message["function_call"])
+                func_print = (
+                    f"***** Suggested function Call: {function_call.get('name', '(No function name found)')} *****"
+                )
                 print(colored(func_print, "green"), flush=True)
                 print(
                     "Arguments: \n",
-                    message["function_call"].get("arguments", "(No arguments found)"),
+                    function_call.get("arguments", "(No arguments found)"),
                     flush=True,
                     sep="",
                 )
@@ -447,7 +455,7 @@ def receive(
                     This field is only needed to distinguish between "function" or "assistant"/"user".
                 4. "name": In most cases, this field is not needed. When the role is "function", this field is needed to indicate the function name.
                 5. "context" (dict): the context of the message, which will be passed to
-                    [Completion.create](../oai/Completion#create).
+                    [OpenAIWrapper.create](../oai/client#create).
             sender: sender of an Agent instance.
             request_reply (bool or None): whether a reply is requested from the sender.
                 If None, the value is determined by `self.reply_at_receive[sender]`.
@@ -483,7 +491,7 @@ async def a_receive(
                     This field is only needed to distinguish between "function" or "assistant"/"user".
                 4. "name": In most cases, this field is not needed. When the role is "function", this field is needed to indicate the function name.
                 5. "context" (dict): the context of the message, which will be passed to
-                    [Completion.create](../oai/Completion#create).
+                    [OpenAIWrapper.create](../oai/client#create).
             sender: sender of an Agent instance.
             request_reply (bool or None): whether a reply is requested from the sender.
                 If None, the value is determined by `self.reply_at_receive[sender]`.
@@ -596,17 +604,17 @@ def generate_oai_reply(
         config: Optional[Any] = None,
     ) -> Tuple[bool, Union[str, Dict, None]]:
         """Generate a reply using autogen.oai."""
-        llm_config = self.llm_config if config is None else config
-        if llm_config is False:
+        client = self.client if config is None else config
+        if client is None:
             return False, None
         if messages is None:
             messages = self._oai_messages[sender]
 
         # TODO: #1143 handle token limit exceeded error
-        response = oai.ChatCompletion.create(
-            context=messages[-1].pop("context", None), messages=self._oai_system_message + messages, **llm_config
+        response = client.create(
+            context=messages[-1].pop("context", None), messages=self._oai_system_message + messages
         )
-        return True, oai.ChatCompletion.extract_text_or_function_call(response)[0]
+        return True, client.extract_text_or_function_call(response)[0]
 
     def generate_code_execution_reply(
         self,

autogen/agentchat/groupchat.py

@@ -10,7 +10,7 @@
 
 @dataclass
 class GroupChat:
-    """A group chat class that contains the following data fields:
+    """(In preview) A group chat class that contains the following data fields:
     - agents: a list of participating agents.
     - messages: a list of messages in the group chat.
     - max_round: the maximum number of rounds.

autogen/agentchat/user_proxy_agent.py

@@ -63,7 +63,7 @@ def __init__(
                 - last_n_messages (Experimental, Optional, int): The number of messages to look back for code execution. Default to 1.
             default_auto_reply (str or dict or None): the default auto reply message when no code execution or llm based reply is generated.
             llm_config (dict or False): llm inference configuration.
-                Please refer to [Completion.create](/docs/reference/oai/completion#create)
+                Please refer to [OpenAIWrapper.create](/docs/reference/oai/client#create)
                 for available options.
                 Default to false, which disables llm-based auto reply.
             system_message (str): system message for ChatCompletion inference.