I am using this code
response = client.chat.completions.create(
model=“gpt-3.5-turbo”, # Use the appropriate model for your needs
messages=[
{“role”: “user”, “content”: prompt}
],
max_tokens=150 # Adjust token limit if necessary
)
I have tried to change my code based on openai versions 0.28.0 and the latest one 1.40.1 and both give the same error.
The above code gives this
An error occurred: ‘ChatCompletionMessage’ object is not subscriptable
The following code
response = openai.ChatCompletion.create(
model="gpt-3.5-turbo", # Use the appropriate model for your needs
messages=[
{"role": "user", "content": prompt}
],
max_tokens=150 # Adjust token limit if necessary
)
Error
An error occurred:
You tried to access openai.ChatCompletion, but this is no longer supported in openai>=1.0.0 - see the README at
You can run openai migrate to automatically upgrade your codebase to use the 1.0.0 interface.
Alternatively, you can pin your installation to the old version, e.g. pip install openai==0.28
from openai import OpenAI
client = OpenAI()
openai_api_key = os.getenv("OPENAI_API_KEY")
if not openai_api_key:
raise ValueError("OPENAI_API_KEY is not set in the environment variables.")
client.api_key = openai_api_key
response = client.chat.completions.create(
model="gpt-3.5-turbo",
messages=[
{"role": "system", "content": "You are a helpful assistant"},
{"role": "user", "content": "Please tell me about yourself"}
],
temperature=0,
)
print response.choices[0].message.content.strip()
Again Error
Code:
from openai import OpenAI
client = OpenAI()
openai_api_key = os.getenv(“OPENAI_API_KEY”)
if not openai_api_key:
raise ValueError(“OPENAI_API_KEY is not set in the environment variables.”)
client.api_key = openai_api_key
response = client.chat.completions.create(
model="gpt-3.5-turbo", # Use the appropriate model for your needs
messages=[
{"role": "user", "content": prompt}
],
temperature=0
)
text = response.choices[0].message[‘content’].strip()
Error:
An error occurred: ‘ChatCompletionMessage’ object is not subscriptable
Your development environment will have a path setting somewhere, I don’t know your setup so I can’t give you any more advise than that.
If you’re using windows have a look in the environment variables diaglog box.
could also be set in your vscode env.
But it 100% seems like you are linking to an older version of the library somewhere.
No, not your Key, the PATH just below it, see if you have multiple places your OpenAI library could be living.
As in, have you installed the OpenAI lib in more than on dev environment, folder structure or drive? you seem to be getting the latest version back when you run pip, but it seems that is not the case for your runtime env.
Try uninstalling the OpenAI library and see if the code still runs. If it does then you know you have an old version someplace.
The code I provided, response.choices[0].message.content, might have been confusing.
The intent of this code was not to suggest that using .strip() is problematic.
The issue lies in the use of bracket notation to access the content field.
Due to a mechanism called Pydantic, which ensures type safety, the error “object is not subscriptable” occurs.
With each version update, the internal specifications of the OpenAI Python module have changed slightly. In the current version of the OpenAI Python module, this notation (bracket notation) cannot be used directly, and dot notation must be used instead.
So, as long as bracket notation is avoided, there should be no error with either
When I wrote “directly” here, I meant that you can also use bracket notation if you convert the response object to a dictionary.
By converting the response object like this:
res = response.dict()
and then applying bracket notation to the relevant parts, as shown below, you can avoid errors and retrieve the result:
res['choices'][0]['message']['content'].strip()
Additionally, the reason ChatGPT often recommends using “.strip()” to remove whitespace is that whitespace can often cause problems in subsequent processing.
However, whether or not the presence of whitespace is actually a problem depends on the specific requirements of your task.
So, instead of taking ChatGPT’s suggestions at face value, please make sure to apply them according to your own specific needs
