autogen
autogen copied to clipboard
Published
20 hours ago •
microsoft
microsoft
[Bug]: `ValidationError` occurs when running branch `gemini`
Describe the bug
First case Scenario
Given
Environment: My local Windows 10 machine
Python version: Python 3.11.7 (tags/v3.11.7:fa7a6f2, Dec 4 2023, 19:24:49) [MSC v.1937 64 bit (AMD64)] on win32
autogen: version 0.2.2
branch: gemini
When
I tried to run first function from samples listed below on my local windows machine
Then
I got an error StopCandidateException ( I rarely get that exception, like floating bug)
raise generation_types.StopCandidateException(response.candidates[0])
google.generativeai.types.generation_types.StopCandidateException: index: 0
finish_reason: RECITATION
see full traceback listing below
Then when I tried to reproduce the Exception above, I got another Exception. (See step 5 from steps to reproduce)
Second Case Scenario
I tryed to reproduce the mentioned bug on Github Spaces platform.
Given:
Environment: Running on github spaces Python version:
-
Python 3.11.2 (main, Mar 13 2023, 12:18:29) [GCC 12.2.0] on linux -
Python 3.10.13 (main, Nov 29 2023, 05:20:19) [GCC 12.2.0] on linux
autogen: version 0.2.2
branch: gemini
commit #: https://github.com/microsoft/autogen/commit/c6792a8adcd5304ad209bb575cd54237c500809c
When:
I tried to run second function from samples with gemini branch and my Google Gen AI API KEY
Then:
I got ValidationError in pydantic_core package.
pydantic_core._pydantic_core.ValidationError: 1 validation error for Choice
logprobs
Field required [type=missing, input_value={'finish_reason': 'stop',...=None, tool_calls=None)}, input_type=dict]
For further information visit https://errors.pydantic.dev/2.5/v/missing
See full traceback below
Steps to reproduce
Step 1
At first I tryed to install autogen package from gemini branch with following commmands (python3.10 by default):
pip install https://github.com/microsoft/autogen/archive/gemini.zip
pip install "google-generativeai" "pydash" "pillow"
pip install "pyautogen[gemini]~=0.2.0b4"
Step 2
And I got an Exception from second case scenario.
Step 3
Then I tryed to create an isolated environment with poetry and python3.11, installed autogen with following commands:
pip install https://github.com/microsoft/autogen/archive/gemini.zip
pip install "google-generativeai" "pydash" "pillow"
pip install "pyautogen[gemini]~=0.2.0b4"
Step 4
And I got an Exception from Second case scenario in pydantic_core package.
Step 5
Then I thought maybe get rid of hard coded version for the last installed package -> 0.2.0b4 and tried on my local machine within isolated poetry environment the following commands:
pip install https://github.com/microsoft/autogen/archive/gemini.zip
pip install "google-generativeai" "pydash" "pillow"
pip install "pyautogen[gemini]"
Step 6
Run just first function within main block. Comment out second function call.
Step 7
And I got StopCandidateException (First case scenario). But It is floating bug, and I managed to get it only once.
Expected Behavior
Agents should start communicating.
Screenshots and logs
Full Traceback for First case scenario :
> $ python sample.py
user_proxy (to assistant):
Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]
--------------------------------------------------------------------------------
Traceback (most recent call last):
File "\home\PycharmProjects\autogen_gemini_test\sample.py", line 45, in <module>
first()
File "\home\PycharmProjects\autogen_gemini_test\sample.py", line 25, in first
user_proxy.initiate_chat(assistant, message="Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]")
File "\home\AppData\Local\pypoetry\Cache\virtualenvs\autogen-gemini-test-yrRBJdLh-py3.11\Lib\site-packages\autogen\agentchat\conversable_agent.py", line 562, in initiate_chat
self.send(self.generate_init_message(**context), recipient, silent=silent)
File "\home\AppData\Local\pypoetry\Cache\virtualenvs\autogen-gemini-test-yrRBJdLh-py3.11\Lib\site-packages\autogen\agentchat\conversable_agent.py", line 360, in send
recipient.receive(message, self, request_reply, silent)
File "\home\AppData\Local\pypoetry\Cache\virtualenvs\autogen-gemini-test-yrRBJdLh-py3.11\Lib\site-packages\autogen\agentchat\conversable_agent.py", line 493, in receive
reply = self.generate_reply(messages=self.chat_messages[sender], sender=sender)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "\home\AppData\Local\pypoetry\Cache\virtualenvs\autogen-gemini-test-yrRBJdLh-py3.11\Lib\site-packages\autogen\agentchat\conversable_agent.py", line 968, in generate_reply
final, reply = reply_func(self, messages=messages, sender=sender, config=reply_func_tuple["config"])
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "\home\AppData\Local\pypoetry\Cache\virtualenvs\autogen-gemini-test-yrRBJdLh-py3.11\Lib\site-packages\autogen\agentchat\conversable_agent.py", line 637, in generate_oai_reply
response = client.create(
^^^^^^^^^^^^^^
File "\home\AppData\Local\pypoetry\Cache\virtualenvs\autogen-gemini-test-yrRBJdLh-py3.11\Lib\site-packages\autogen\oai\client.py", line 274, in create
response = client.call(params)
^^^^^^^^^^^^^^^^^^^
File "\home\AppData\Local\pypoetry\Cache\virtualenvs\autogen-gemini-test-yrRBJdLh-py3.11\Lib\site-packages\autogen\oai\gemini.py", line 93, in call
response = chat.send_message(gemini_messages[-1].parts[0].text, stream=stream)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "\home\AppData\Local\pypoetry\Cache\virtualenvs\autogen-gemini-test-yrRBJdLh-py3.11\Lib\site-packages\google\generativeai\generative_models.py", line 384, in send_message
raise generation_types.StopCandidateException(response.candidates[0])
google.generativeai.types.generation_types.StopCandidateException: index: 0
finish_reason: RECITATION
Full Traceback for Second case scenario :
> $ python simple_chat.py
user_proxy (to assistant):
Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]
--------------------------------------------------------------------------------
Traceback (most recent call last):
File "/workspaces/autogen/samples/simple_chat.py", line 46, in <module>
another()
File "/workspaces/autogen/samples/simple_chat.py", line 25, in another
user_proxy.initiate_chat(assistant, message="Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]")
File "/home/vscode/.cache/pypoetry/virtualenvs/samples-czj8q62m-py3.11/lib/python3.11/site-packages/autogen/agentchat/conversable_agent.py", line 562, in initiate_chat
self.send(self.generate_init_message(**context), recipient, silent=silent)
File "/home/vscode/.cache/pypoetry/virtualenvs/samples-czj8q62m-py3.11/lib/python3.11/site-packages/autogen/agentchat/conversable_agent.py", line 360, in send
recipient.receive(message, self, request_reply, silent)
File "/home/vscode/.cache/pypoetry/virtualenvs/samples-czj8q62m-py3.11/lib/python3.11/site-packages/autogen/agentchat/conversable_agent.py", line 493, in receive
reply = self.generate_reply(messages=self.chat_messages[sender], sender=sender)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/vscode/.cache/pypoetry/virtualenvs/samples-czj8q62m-py3.11/lib/python3.11/site-packages/autogen/agentchat/conversable_agent.py", line 968, in generate_reply
final, reply = reply_func(self, messages=messages, sender=sender, config=reply_func_tuple["config"])
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/vscode/.cache/pypoetry/virtualenvs/samples-czj8q62m-py3.11/lib/python3.11/site-packages/autogen/agentchat/conversable_agent.py", line 637, in generate_oai_reply
response = client.create(
^^^^^^^^^^^^^^
File "/home/vscode/.cache/pypoetry/virtualenvs/samples-czj8q62m-py3.11/lib/python3.11/site-packages/autogen/oai/client.py", line 274, in create
response = client.call(params)
^^^^^^^^^^^^^^^^^^^
File "/home/vscode/.cache/pypoetry/virtualenvs/samples-czj8q62m-py3.11/lib/python3.11/site-packages/autogen/oai/gemini.py", line 123, in call
choices = [Choice(finish_reason="stop", index=0, message=message)]
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/vscode/.cache/pypoetry/virtualenvs/samples-czj8q62m-py3.11/lib/python3.11/site-packages/pydantic/main.py", line 164, in __init__
__pydantic_self__.__pydantic_validator__.validate_python(data, self_instance=__pydantic_self__)
pydantic_core._pydantic_core.ValidationError: 1 validation error for Choice
logprobs
Field required [type=missing, input_value={'finish_reason': 'stop',...=None, tool_calls=None)}, input_type=dict]
For further information visit https://errors.pydantic.dev/2.5/v/missing
Additional Information
Code samples I tried to run
from autogen import UserProxyAgent, ConversableAgent, config_list_from_json, AssistantAgent
config_list_gemini = [{
"model": "gemini-pro",
"api_key": "AIza-my-api-key",
"api_type": "google"
}]
config_list_gemini_vision = [{
"model": "gemini-pro-vision",
"api_key": "AIza-my-api-key",
"api_type": "google"
}]
def first():
assistant = AssistantAgent("assistant",
llm_config={"config_list": config_list_gemini, "seed": 42},
max_consecutive_auto_reply=13)
user_proxy = UserProxyAgent("user_proxy",
code_execution_config={"work_dir": "coding", "use_docker": False},
human_input_mode="NEVER",
is_termination_msg = lambda x: content_str(x.get("content")).find("TERMINATE") >= 0)
user_proxy.initiate_chat(assistant, message="Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]")
def second():
assistant = ConversableAgent("agent", llm_config={"config_list": config_list_gemini})
user_proxy = UserProxyAgent("user", code_execution_config=True)
assistant.initiate_chat(user_proxy, message="How can I help you today?")
if __name__ == "__main__":
first()
second()
Packages installed for the first case:
$ pip freeze
- annotated-types==0.6.0
- anyio==4.2.0
- cachetools==5.3.2
- certifi==2023.11.17
- charset-normalizer==3.3.2
- colorama==0.4.6
- diskcache==5.6.3
- distro==1.9.0
- FLAML==2.1.1
- google-ai-generativelanguage==0.4.0
- google-api-core==2.15.0
- google-auth==2.26.1
- google-generativeai==0.3.2
- googleapis-common-protos==1.62.0
- grpcio==1.60.0
- grpcio-status==1.60.0
- h11==0.14.0
- httpcore==1.0.2
- httpx==0.26.0
- idna==3.6
- numpy==1.26.3
- openai==1.6.1
- pillow==10.2.0
- proto-plus==1.23.0
- protobuf==4.25.1
- pyasn1==0.5.1
- pyasn1-modules==0.3.0
- pyautogen @ https://github.com/microsoft/autogen/archive/gemini.zip#sha256=a7ecbc81aa9279dde95be5ef7e33a8cd1733d90db3622c0f29ca8a53b4de511c
- pydantic==2.5.3
- pydantic_core==2.14.6
- pydash==7.0.6
- python-dotenv==1.0.0
- regex==2023.12.25
- requests==2.31.0
- rsa==4.9
- sniffio==1.3.0
- termcolor==2.4.0
- tiktoken==0.5.2
- tqdm==4.66.1
- typing_extensions==4.9.0
- urllib3==2.1.0
Packages installed for the second case:
$ pip freeze
- annotated-types==0.6.0
- anyio==4.2.0
- cachetools==5.3.2
- certifi==2023.11.17
- charset-normalizer==3.3.2
- diskcache==5.6.3
- distro==1.9.0
- FLAML==2.1.1
- google-ai-generativelanguage==0.4.0
- google-api-core==2.15.0
- google-auth==2.26.1
- google-generativeai==0.3.2
- googleapis-common-protos==1.62.0
- grpcio==1.60.0
- grpcio-status==1.60.0
- h11==0.14.0
- httpcore==1.0.2
- httpx==0.26.0
- idna==3.6
- numpy==1.26.3
- openai==1.6.1
- pillow==10.2.0
- proto-plus==1.23.0
- protobuf==4.25.1
- pyasn1==0.5.1
- pyasn1-modules==0.3.0
- pyautogen @ https://github.com/microsoft/autogen/archive/gemini.zip#sha256=a7ecbc81aa9279dde95be5ef7e33a8cd1733d90db3622c0f29ca8a53b4de511c
- pydantic==2.5.3
- pydantic_core==2.14.6
- pydash==7.0.6
- python-dotenv==1.0.0
- regex==2023.12.25
- requests==2.31.0
- rsa==4.9
- sniffio==1.3.0
- termcolor==2.4.0
- tiktoken==0.5.2
- tqdm==4.66.1
- typing_extensions==4.9.0
- urllib3==2.1.0
Thanks for the details! It seems you are using a different pydantic version. Can you try:
pip install "google-generativeai" "pydash" "pillow" "pydantic==1.10.13"
Thanks for the details! It seems you are using a different pydantic version. Can you try: pip install "google-generativeai" "pydash" "pillow" "pydantic==1.10.13"
Looks like it helps with Exception in pydantic_core package, but here is two more bugs in gemini branch:
First case: exceeds MAX_TOKENS raises StopCandidateException:
Installation phase
pip install https://github.com/microsoft/autogen/archive/gemini.zip
pip install "google-generativeai" "pydash" "pillow" "pydantic==1.10.13"
pip install "pyautogen[gemini]"
The code I try to run:
config_list_gemini = [{
"model": "gemini-pro",
"api_key": "AIza-my-api-key",
"api_type": "google"
}]
assistant = AssistantAgent("assistant",
llm_config={"config_list": config_list_gemini, "seed": 42},
max_consecutive_auto_reply=13)
user_proxy = UserProxyAgent("user_proxy",
code_execution_config={"work_dir": "coding", "use_docker": False},
human_input_mode="NEVER",
is_termination_msg = lambda x: content_str(x.get("content")).find("TERMINATE") >= 0)
user_proxy.initiate_chat(assistant, message="Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]")
Full Traceback listing:
agent (to user):
How can I help you today?
--------------------------------------------------------------------------------
Provide feedback to agent. Press enter to skip and use auto-reply, or type 'exit' to end the conversation: Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]
user (to agent):
Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]
--------------------------------------------------------------------------------
>>>>>>>> USING AUTO REPLY...
---------------------------------------------------------------------------
StopCandidateException Traceback (most recent call last)
<ipython-input-14-e58c8e72340a> in <cell line: 11>()
9 is_termination_msg = lambda x: content_str(x.get("content")).find("TERMINATE") >= 0)
10
---> 11 user_proxy.initiate_chat(assistant, message="Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]")
12
7 frames
/usr/local/lib/python3.10/dist-packages/autogen/agentchat/conversable_agent.py in initiate_chat(self, recipient, clear_history, silent, **context)
560 """
561 self._prepare_chat(recipient, clear_history)
--> 562 self.send(self.generate_init_message(**context), recipient, silent=silent)
563
564 async def a_initiate_chat(
/usr/local/lib/python3.10/dist-packages/autogen/agentchat/conversable_agent.py in send(self, message, recipient, request_reply, silent)
358 valid = self._append_oai_message(message, "assistant", recipient)
359 if valid:
--> 360 recipient.receive(message, self, request_reply, silent)
361 else:
362 raise ValueError(
/usr/local/lib/python3.10/dist-packages/autogen/agentchat/conversable_agent.py in receive(self, message, sender, request_reply, silent)
491 if request_reply is False or request_reply is None and self.reply_at_receive[sender] is False:
492 return
--> 493 reply = self.generate_reply(messages=self.chat_messages[sender], sender=sender)
494 if reply is not None:
495 self.send(reply, sender, silent=silent)
/usr/local/lib/python3.10/dist-packages/autogen/agentchat/conversable_agent.py in generate_reply(self, messages, sender, exclude)
966 continue
967 if self._match_trigger(reply_func_tuple["trigger"], sender):
--> 968 final, reply = reply_func(self, messages=messages, sender=sender, config=reply_func_tuple["config"])
969 if final:
970 return reply
/usr/local/lib/python3.10/dist-packages/autogen/agentchat/conversable_agent.py in generate_oai_reply(self, messages, sender, config)
635
636 # TODO: #1143 handle token limit exceeded error
--> 637 response = client.create(
638 context=messages[-1].pop("context", None), messages=self._oai_system_message + messages
639 )
/usr/local/lib/python3.10/dist-packages/autogen/oai/client.py in create(self, **config)
272 try:
273 if isinstance(client, GeminiClient):
--> 274 response = client.call(params)
275 else:
276 response = self._completions_create(client, params)
/usr/local/lib/python3.10/dist-packages/autogen/oai/gemini.py in call(self, params)
91 chat = model.start_chat(history=gemini_messages[:-1])
92 try:
---> 93 response = chat.send_message(gemini_messages[-1].parts[0].text, stream=stream)
94 except InternalServerError as e:
95 print(e)
/usr/local/lib/python3.10/dist-packages/google/generativeai/generative_models.py in send_message(self, content, generation_config, safety_settings, stream, **kwargs)
382 glm.Candidate.FinishReason.MAX_TOKENS,
383 ):
--> 384 raise generation_types.StopCandidateException(response.candidates[0])
385
386 self._last_sent = content
StopCandidateException: finish_reason: RECITATION
index: 0
Second case: raises ValueError when I asked for feedback and send empty string:
Instalation phase is the same
Expected behaviour:
Common sense tells me that when I asked Press enter to skip that means 'Provide an empty String to python's input() function', but it should be handled behaviour and the program shouldn't crash when I send empty string back, but it does.
The sample I try to run
from autogen import UserProxyAgent, ConversableAgent
config_list_gemini = [{
"model": "gemini-pro",
"api_key": "AIza-my-api-key",
"api_type": "google"
}]
def main():
assistant = ConversableAgent("agent", llm_config={"config_list": config_list_gemini})
user_proxy = UserProxyAgent("user", code_execution_config=False)
assistant.initiate_chat(user_proxy, message="How can I help you today?")
if __name__ == "__main__":
main()
Full Traceback listing:
agent (to user):
How can I help you today?
--------------------------------------------------------------------------------
Provide feedback to agent. Press enter to skip and use auto-reply, or type 'exit' to end the conversation: Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]
user (to agent):
Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]
--------------------------------------------------------------------------------
>>>>>>>> USING AUTO REPLY...
INFO:autogen.token_count_utils:Gemini is not supported in tiktoken. Returning num tokens assuming gpt-4-0613.
WARNING:autogen.token_count_utils:Model gemini-pro not found. Using cl100k_base encoding.
Warning: model not found. Using cl100k_base encoding.
agent (to user):
```python
def bubble_sort(array):
"""
Sorts the given array using the Bubble Sort algorithm.
Parameters:
array: The array to be sorted.
Returns:
The sorted array.
"""
# Iterate over the array multiple times
for i in range(len(array) - 1):
# In each iteration, compare adjacent elements and swap them if they are in the wrong order
for j in range(len(array) - 1 - i):
if array[j] > array[j + 1]:
array[j], array[j + 1] = array[j + 1], array[j]
return array
if __name__ == "__main__":
array = [4, 1, 3, 2]
print(f"Unsorted array: {array}")
sorted_array = bubble_sort(array)
print(f"Sorted array: {sorted_array}")
Output:
Unsorted array: [4, 1, 3, 2]
Sorted array: [1, 2, 3, 4]
--------------------------------------------------------------------------------
Provide feedback to agent. Press enter to skip and use auto-reply, or type 'exit' to end the conversation:
>>>>>>>> NO HUMAN INPUT RECEIVED.
>>>>>>>> USING AUTO REPLY...
user (to agent):
--------------------------------------------------------------------------------
>>>>>>>> USING AUTO REPLY...
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-7-9c7ec639fed5> in <cell line: 20>()
19
20 if __name__ == "__main__":
---> 21 main()
15 frames
/usr/local/lib/python3.10/dist-packages/google/generativeai/types/content_types.py in to_content(content)
192 def to_content(content: ContentType):
193 if not content:
--> 194 raise ValueError("content must not be empty")
195
196 if isinstance(content, Mapping):
ValueError: content must not be empty
Looking at the block below, it seems like you have the "human_input_mode" parameter wrong.
Provide feedback to agent. Press enter to skip and use auto-reply, or type 'exit' to end the conversation: Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2] user (to agent):
Also, the code and the log you provided did not match. For instance, the agents' names do not match from the code to the output log.
One possibility is that the code is changed after caching the same question, and you can try to run rm -rf .cache in the same folder before rerunning.
Please fix the code. Still getting RECITATION Exception:
Installation (This time running in Google Colab)
!pip install https://github.com/microsoft/autogen/archive/gemini.zip
!pip install "google-generativeai" "pydash" "pillow"
!pip install git+https://github.com/microsoft/autogen.git@gemini
from google.colab import userdata
Source code
from autogen import UserProxyAgent, ConversableAgent, config_list_from_json, AssistantAgent
config_list_gemini = [{
"model": "gemini-pro",
"api_key": userdata.get('GOOGLE_AI_API_KEY'),
"api_type": "google"
}]
def first():
assistant = AssistantAgent("assistant",
llm_config={"config_list": config_list_gemini, "seed": 42},
max_consecutive_auto_reply=13)
user_proxy = UserProxyAgent("user_proxy",
code_execution_config={"work_dir": "coding", "use_docker": False},
human_input_mode="NEVER",
is_termination_msg = lambda x: content_str(x.get("content")).find("TERMINATE") >= 0)
user_proxy.initiate_chat(assistant, message="Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]")
if __name__ == "__main__":
first()
Traceback
user_proxy (to assistant):
Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]
--------------------------------------------------------------------------------
---------------------------------------------------------------------------
StopCandidateException Traceback (most recent call last)
[<ipython-input-2-fd2ce89ef12a>](https://localhost:8080/#) in <cell line: 37>()
36
37 if __name__ == "__main__":
---> 38 first()
39 second()
8 frames
[/usr/local/lib/python3.10/dist-packages/google/generativeai/generative_models.py](https://localhost:8080/#) in send_message(self, content, generation_config, safety_settings, stream, **kwargs)
382 glm.Candidate.FinishReason.MAX_TOKENS,
383 ):
--> 384 raise generation_types.StopCandidateException(response.candidates[0])
385
386 self._last_sent = content
StopCandidateException: finish_reason: RECITATION
index: 0
The same setting and caught different ExceptionValueError
To troubleshoot this issue, let's try a few things:
1. Make sure that you are providing input to the program correctly. You can do this by running the program from the command line and providing the input directly. For example, you could run the program like this:
python complete_graph.py <<EOF
4
1 2
1 3
3 4
5
1 2
1 3
2 3
4 1
4 2
EOF
This will provide the following input to the program:
4
1 2
1 3
3 4
5
1 2
1 3
2 3
4 1
4 2
2. If you are providing input to the program correctly, then the issue may be with how the input is being read. You can try to debug the program by printing the value of the `line` variable before trying to convert it to an integer. This will help you see what value is being read from the input.
3. If the `line` variable is not empty, then the issue may be with the input itself. Make sure that the input is in the correct format and that it does not contain any invalid characters.
Once you have identified the source of the issue, you can take steps to fix it. For example, if you are not providing input to the program correctly, you can simply provide the input directly when you run the program. If there is an issue with how the input is being read, you can try to debug the program to find the source of the issue. And if there is an issue with the input itself, you can correct the input and then run the program again.
--------------------------------------------------------------------------------
>>>>>>>> EXECUTING CODE BLOCK 0 (inferred language is sh)...
>>>>>>>> EXECUTING CODE BLOCK 1 (inferred language is unknown)...
user_proxy (to assistant):
exitcode: 1 (execution failed)
Code output:
4
unknown language unknown
--------------------------------------------------------------------------------
INFO:autogen.token_count_utils:Gemini is not supported in tiktoken. Returning num tokens assuming gpt-4-0613.
WARNING:autogen.token_count_utils:Model gemini-pro not found. Using cl100k_base encoding.
Warning: model not found. Using cl100k_base encoding.
assistant (to user_proxy):
It seems that the program is now reading the input correctly, but it is encountering an error when trying to convert the input to an integer. The error message "unknown language unknown" suggests that the program is trying to convert the input to an integer using a language that is not supported.
To fix this issue, we need to make sure that the program is using the correct language to convert the input to an integer. In Python, we can use the `int()` function to convert a string to an integer. The `int()` function takes two arguments: the string to be converted and the base of the integer. The base of the integer is the number of digits that are used to represent the integer. For example, the base of the decimal system is 10, which means that we use 10 digits (0, 1, 2, 3, 4, 5, 6, 7, 8, 9) to represent numbers.
In our case, we want to convert the input string to an integer using the decimal system. Therefore, we need to call the `int()` function with the input string as the first argument and the base 10 as the second argument.
Here is the corrected code:
# filename: complete_graph.py
import sys
# Read the number of vertices from the first line of the input.
line = sys.stdin.readline()
if not line:
print("Error: Empty input.")
exit(1)
n = int(line, 10) # Convert the input string to an integer using base 10
# ... (the rest of the code)
Please try running the program again and let me know if you encounter any further issues.
--------------------------------------------------------------------------------
>>>>>>>> EXECUTING CODE BLOCK 0 (inferred language is python)...
user_proxy (to assistant):
exitcode: 1 (execution failed)
Code output:
--------------------------------------------------------------------------------
INFO:autogen.token_count_utils:Gemini is not supported in tiktoken. Returning num tokens assuming gpt-4-0613.
WARNING:autogen.token_count_utils:Model gemini-pro not found. Using cl100k_base encoding.
Warning: model not found. Using cl100k_base encoding.
assistant (to user_proxy):
If the program is still exiting with an error code of 1 and not producing any output, it means that there is still an issue with the program.
One possible reason for this is that the input is still not being read correctly. To troubleshoot this issue, you can try printing the value of the `line` variable before trying to convert it to an integer. This will help you see what value is being read from the input.
Another possible reason for the issue is that there is a bug in the program itself. You can try to debug the program to find the source of the bug. You can do this by adding print statements to the program to see what values are being stored in the variables at different points in the program.
Once you have identified the source of the issue, you can take steps to fix it. For example, if the input is not being read correctly, you can try to modify the program to read the input in a different way. If there is a bug in the program, you can try to fix the bug by modifying the code.
Here are some additional things you can try to troubleshoot the issue:
* Make sure that the program is being run with the correct Python interpreter.
* Make sure that the program is being run from the correct directory.
* Make sure that the input file is in the correct format and that it is located in the correct directory.
If you are still having trouble getting the program to run correctly, please provide me with the input that you are using and the full error message that is being displayed. This will help me to better understand the issue and provide you with a more accurate solution.
--------------------------------------------------------------------------------
user_proxy (to assistant):
--------------------------------------------------------------------------------
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
[<ipython-input-7-d9609b5d4d8b>](https://localhost:8080/#) in <cell line: 24>()
23
24 if __name__ == "__main__":
---> 25 first()
33 frames
[/usr/local/lib/python3.10/dist-packages/google/generativeai/types/content_types.py](https://localhost:8080/#) in to_content(content)
192 def to_content(content: ContentType):
193 if not content:
--> 194 raise ValueError("content must not be empty")
195
196 if isinstance(content, Mapping):
ValueError: content must not be empty
@OTR Thanks for contributing these examples and logs! Code updated. Please try again.
You hit the RECITATION problem reported here and here
The problem is triggered by the fact that "Write a program in python that Sort the array with Bubble Sort: [4, 1, 3, 2]" is a really common problem. Whatever is the answer it will generate, there is a similar data in its training data. Gemini will then refuse to give it to you. Gemini does not want to plagiate.
@OTR @rakotomandimby @naourass Thanks for all your interest using AutoGen for Gemini. Do you have any suggestions regarding the "RECITATION" problem for Gemini? As pointed out by @rakotomandimby, there are already lots of complaints for the Gemini API (not from AutoGen, but from other usages).
If a try... catch... exception could not resolve this issue elegantly, how to proceed and kick Gemini to give us a response (from prompt engineering perspective)?
@BeibinLi Could you please provide a working example of using autogen at gemini branch?
Your last commit points to not existing page https://github.com/microsoft/autogen/commit/9bc4d4e3b9f888b3bba550222c2321b051af705d
Such notebook doesn't exist: https://github.com/microsoft/autogen/blob/main/notebook/agentchat_gemini.ipynb
When I am attempting to launch my sample code, which previously launched without errors, I am encountering a compilation error:
!pip install https://github.com/microsoft/autogen/archive/gemini.zip
!pip install "google-generativeai" "pydash" "pillow"
!pip install git+https://github.com/microsoft/autogen.git@gemini
Traceback:
https://gist.github.com/OTR/2c175ef404955dfca68bce57d5727e0e
@OTR Thanks for pointing out. The notebook is at: https://github.com/microsoft/autogen/blob/gemini/notebook/agentchat_gemini.ipynb
For the installation bug, it seems like a conversable-agent.jpg file is in the tmp folder, which caused the issue.
Can you try to run the following code before installing?
rm -rf /tmp/pip-req-build-spmrq54h
Thanks!!!
Closing due to inactivity. Gemini is now officially supported by AutoGen, and you can check our roadmap for gemini at: https://github.com/microsoft/autogen/issues/2387