v1.0.0 Migration Guide

[rattrayalex]

We have released a new major version of our SDK, and we recommend upgrading promptly.

It's a total rewrite of the library, so many things have changed, but we've made upgrading easy with a code migration script and detailed docs below. It was extensively beta tested prior to release.

Getting started

pip install --upgrade openai

What's changed

• Auto-retry with backoff if there's an error
• Proper types (for mypy/pyright/editors)
• You can now instantiate a client, instead of using a global default.
• Switch to explicit client instantiation
• Weights and Biases CLI will now be included in their own package

Migration guide

For Azure OpenAI users, see Microsoft's Azure-specific migration guide.

Automatic migration with grit

You can automatically migrate your codebase using grit, either online or with the following CLI command on Mac or Linux:

openai migrate

The grit binary executes entirely locally with AST-based transforms.

Be sure to audit its changes: we suggest ensuring you have a clean working tree beforehand, and running git add --patch afterwards. Note that grit.io also offers opt-in automatic fixes powered by AI.

Automatic migration with grit on Windows

To use grit to migrate your code on Windows, you will need to use Windows Subsystem for Linux (WSL). Installing WSL is quick and easy, and you do not need to keep using Linux once the command is done.

Here's a step-by-step guide for setting up and using WSL for this purpose:

1. Open a PowerShell or Command Prompt as an administrator and run wsl --install.
2. Restart your computer.
3. Open the WSL application.
4. In the WSL terminal, cd into the appropriate directory (e.g., cd /mnt/c/Users/Myself/my/code/) and then run the following commands:

   curl -fsSL https://docs.grit.io/install | bash
   grit install
   grit apply openai

Then, you can close WSL and go back to using Windows.

Automatic migration with grit in Jupyter Notebooks

If your Jupyter notebooks are not in source control, they will be more difficult to migrate. You may want to copy each cell into grit's web interface, and paste the output back in.

If you need to migrate in a way that preserves use of the module-level client instead of instantiated clients, you can use the openai_global grit migration instead.

Initialization

# old
import openai

openai.api_key = os.environ['OPENAI_API_KEY']

# new
from openai import OpenAI

client = OpenAI(
  api_key=os.environ['OPENAI_API_KEY'],  # this is also the default, it can be omitted
)

Responses

Response objects are now pydantic models and no longer conform to the dictionary shape. However you can easily convert them to a dictionary with model.model_dump().

# before
import json
import openai

completion = openai.Completion.create(model='curie')
print(completion['choices'][0]['text'])
print(completion.get('usage'))
print(json.dumps(completion, indent=2))

# after
from openai import OpenAI

client = OpenAI()

completion = client.completions.create(model='curie')
print(completion.choices[0].text)
print(dict(completion).get('usage'))
print(completion.model_dump_json(indent=2))

Async client

We do not support calling asynchronous methods in the module-level client, instead you will have to instantiate an async client.

The rest of the API is exactly the same as the synchronous client.

# old
import openai

completion = openai.ChatCompletion.acreate(model="gpt-3.5-turbo", messages=[{"role": "user", "content": "Hello world"}])

# new
from openai import AsyncOpenAI

client = AsyncOpenAI()
completion = await client.chat.completions.create(model="gpt-3.5-turbo", messages=[{"role": "user", "content": "Hello world"}])

Module client

Important

We highly recommend instantiating client instances instead of relying on the global client.

We also expose a global client instance that is accessible in a similar fashion to versions prior to v1.

import openai

# optional; defaults to `os.environ['OPENAI_API_KEY']`
openai.api_key = '...'

# all client options can be configured just like the `OpenAI` instantiation counterpart
openai.base_url = "https://..."
openai.default_headers = {"x-foo": "true"}

completion = openai.chat.completions.create(
    model="gpt-4",
    messages=[
        {
            "role": "user",
            "content": "How do I output all files in a directory using Python?",
        },
    ],
)
print(completion.choices[0].message.content)

The API is the exact same as the standard client instance based API.

This is intended to be used within REPLs or notebooks for faster iteration, not in application code.

We recommend that you always instantiate a client (e.g., with client = OpenAI()) in application code because:

• It can be difficult to reason about where client options are configured
• It's not possible to change certain client options without potentially causing race conditions
• It's harder to mock for testing purposes
• It's not possible to control cleanup of network connections

Pagination

All list() methods that support pagination in the API now support automatic iteration, for example:

from openai import OpenAI

client = OpenAI()

for job in client.fine_tuning.jobs.list(limit=1):
    print(job)

Previously you would have to explicitly call a .auto_paging_iter() method instead.
See the README for more details.

Azure OpenAI

To use this library with Azure OpenAI, use the AzureOpenAI class instead of the OpenAI class.

A more comprehensive Azure-specific migration guide is available on the Microsoft website.

Important

The Azure API shape differs from the core API shape which means that the static types for responses / params
won't always be correct.

from openai import AzureOpenAI

# gets the API Key from environment variable AZURE_OPENAI_API_KEY
client = AzureOpenAI(
    # https://learn.microsoft.com/en-us/azure/ai-services/openai/reference#rest-api-versioning
    api_version="2023-07-01-preview",
    # https://learn.microsoft.com/en-us/azure/cognitive-services/openai/how-to/create-resource?pivots=web-portal#create-a-resource
    azure_endpoint="https://example-endpoint.openai.azure.com",
)

completion = client.chat.completions.create(
    model="deployment-name",  # e.g. gpt-35-instant
    messages=[
        {
            "role": "user",
            "content": "How do I output all files in a directory using Python?",
        },
    ],
)
print(completion.model_dump_json(indent=2))

In addition to the options provided in the base OpenAI client, the following options are provided:

• azure_endpoint
• azure_deployment
• api_version
• azure_ad_token
• azure_ad_token_provider

An example of using the client with Azure Active Directory can be found here.

All name changes

Note: all a* methods have been removed; the async client must be used instead.

• openai.api_base -> openai.base_url
• openai.proxy -> openai.proxies (docs)
• openai.InvalidRequestError -> openai.BadRequestError
• openai.Audio.transcribe() -> client.audio.transcriptions.create()
• openai.Audio.translate() -> client.audio.translations.create()
• openai.ChatCompletion.create() -> client.chat.completions.create()
• openai.Completion.create() -> client.completions.create()
• openai.Edit.create() -> client.edits.create()
• openai.Embedding.create() -> client.embeddings.create()
• openai.File.create() -> client.files.create()
• openai.File.list() -> client.files.list()
• openai.File.retrieve() -> client.files.retrieve()
• openai.File.download() -> client.files.retrieve_content()
• openai.FineTune.cancel() -> client.fine_tunes.cancel()
• openai.FineTune.list() -> client.fine_tunes.list()
• openai.FineTune.list_events() -> client.fine_tunes.list_events()
• openai.FineTune.stream_events() -> client.fine_tunes.list_events(stream=True)
• openai.FineTune.retrieve() -> client.fine_tunes.retrieve()
• openai.FineTune.delete() -> client.fine_tunes.delete()
• openai.FineTune.create() -> client.fine_tunes.create()
• openai.FineTuningJob.create() -> client.fine_tuning.jobs.create()
• openai.FineTuningJob.cancel() -> client.fine_tuning.jobs.cancel()
• openai.FineTuningJob.delete() -> client.fine_tuning.jobs.create()
• openai.FineTuningJob.retrieve() -> client.fine_tuning.jobs.retrieve()
• openai.FineTuningJob.list() -> client.fine_tuning.jobs.list()
• openai.FineTuningJob.list_events() -> client.fine_tuning.jobs.list_events()
• openai.Image.create() -> client.images.generate()
• openai.Image.create_variation() -> client.images.create_variation()
• openai.Image.create_edit() -> client.images.edit()
• openai.Model.list() -> client.models.list()
• openai.Model.delete() -> client.models.delete()
• openai.Model.retrieve() -> client.models.retrieve()
• openai.Moderation.create() -> client.moderations.create()
• openai.api_resources -> openai.resources

Removed

• openai.api_key_path
• openai.app_info
• openai.debug
• openai.log
• openai.OpenAIError
• openai.Audio.transcribe_raw()
• openai.Audio.translate_raw()
• openai.ErrorObject
• openai.Customer
• openai.api_version
• openai.verify_ssl_certs
• openai.api_type
• openai.enable_telemetry
• openai.ca_bundle_path
• openai.requestssession (we now use httpx)
• openai.aiosession (we now use httpx)
• openai.Deployment (only used for Azure) – please use the azure-mgmt-cognitiveservices client library instead (here's how to list deployments, for example).
• openai.Engine
• openai.File.find_matching_files()
• openai.embeddings_utils (now in the cookbook)

[jakobdylanc]

Is the highlighted line a typo? (image removed because it was huge)

[RobertCraigie]

Thanks, fixed!

[TSSFL]

Which code is the new equivalent to the old one here https://tssfl.com/viewtopic.php?t=6651 ?

[hammad93]

Needs a comma after defining api_version in the snippet,

client = AzureOpenAI(
    # https://learn.microsoft.com/en-us/azure/ai-services/openai/reference#rest-api-versioning
    api_version="2023-07-01-preview"
    # https://learn.microsoft.com/en-us/azure/cognitive-services/openai/how-to/create-resource?pivots=web-portal#create-a-resource
    azure_endpoint="https://example-endpoint.openai.azure.com",
)

[RobertCraigie]

Thanks, fixed!

[giopoi]

Incomprensibile ma chi le scrive queste cose?

[RobertCraigie]

Which parts did you find difficult to understand? Do you have any feedback on where we could improve the guide?

[skyl]

I followed the author because I thought it was such a good effort 🤷‍♀️

[giopoi]

You can run openai migrate to automatically upgrade your codebase to use the 1.0.0 interface.
in windows non funziona

[AveryTang11]

now there is another problem

[morgante]

Hi @TeddyTang11, the correct command is grit apply openai.

[AveryTang11]

OK, i'll try it. Thanks very much!

[shivv-001]

i'm facing thefollowing issue could you please help me with that

raise APIRemovedInV1(symbol=self._symbol)
openai.lib._old_api.APIRemovedInV1:

You tried to access openai.Completion, but this is no longer supported in openai>=1.0.0 - see the README at https://github.com/openai/openai-python for the API.

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

A detailed migration guide is available here: #742

[PierrunoYT]

Use this https://platform.openai.com/docs/api-reference/images

[stevechappell2000]

What happened to the functions array that was in chatCompletions? Seems to no longer function in the chat.completions?? Sample code??

[RobertCraigie]

It's still there, what's not working for you?

openai-python/src/openai/resources/chat/completions.py

Line 63 in 8e4e5d4

functions: List[completion_create_params.Function] | NotGiven = NOT_GIVEN,

[stevechappell2000]

I see some comment about functions has been replaced by tools? do you have a working sample?

[phongvants123]

did you figure it out, i have the same problem

[giopoi]

Adesso sono impegnato poi mi spieghero' Il gio 9 nov 2023, 15:39 Robert Craigie ***@***.***> ha scritto:

…

It's still there, what's not working for you? https://github.com/openai/openai-python/blob/8e4e5d490629a5abfbdf4b6a2ce20c7eee848a2e/src/openai/resources/chat/completions.py#L63 — Reply to this email directly, view it on GitHub <#742 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAMXYQRJPKWOUMVS7OXZSVTYDTTKRAVCNFSM6AAAAAA7DOKUK2VHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM3TKMRTGI2TC> . You are receiving this because you commented.Message ID: ***@***.***>

[ppravdin]

It would be great if you could update ChatGPT's knowledge as well. Currently, it provides OpenAI-related code of an old version, even when I provide examples and ask for updates to the new version.

[RobertCraigie]

@ppravdin you should be able to still pin the version you're installing in google colab, pip install openai==0.28

[golean1]

Are there any solutions to getting chatgpt to strictly reply with the new API update code? I seem to be having the same issue therer and in the assistant playground. Even when you attach or send the api documention to it, it seems to only hold that context for a few responses at most before regressing back to the old code. It's been a bummer to not be able to get as much help from chatgpt on the assistant api for example compared to anything else from april and prior.

[kenhuangus]

The best way is not to introduce break changes in SDK. Instead fix the old SDK to redirect calls via API to new SDK. So, people use old SDK prior to version 1.00 can still use the old SDK unless they want to leverage new functionality in new SDK. This should really be a responsibility of SDK provider or developer to make sure the burden of break fix in on their side especially if there are many users of SDK. OpenAI can be a pioneer in this and promise no break changes in the future. Technically it is doable, you just need to put more development efforts into it. But, if you train your LLM to change your old SDK to redirect calls to new SDK your efforts is minimum.

[rattrayalex]

Breaking changes between major versions, especially pre-1.0 versions, is standard practice in software development. Therefore, OpenAI chose not to prioritize the development effort required for full backwards compatibility with v0.28.x, which had many shortcomings.

@kenhuangus if you'd like to fork the v0.28.1 branch to rely on the v1.0 library, you're certainly welcome to, and I'm sure you'd find users for it.

[FlyingFathead]

Should anyone need a snippet of working chatbot code for the 1.6.1 version of openai pip package; https://github.com/FlyingFathead/TelegramBot-OpenAI-API/blob/b5c315e8dc46109279659a3adf6a997791cdbe35

Just a forewarning -- that code is WIP, been building it the past week or so. But, it works pretty well, although using httpx.AsyncClient() seemed to have been the only properly working method I could come up with in terms of getting the payload just right and everything to work without issues.

If you need GPT-4 to assist you in figuring out the changes, I could suggest giving a snippet like that to it and asking to figure out what has changed in terms of the request and how it's formed with the newer openai library version.

Good luck!

[jtoh89]

Was there a heads-up on this? We haven't deployed our tools utilizing openai api into production yet, but good thing we didn't. We had no idea that there would be a new version. Maybe next time there could be extended support for legacy?

[richlysakowski]

Thoughtless Slippery Sam just threw out a truckloads of banana peels for the developers everywhere to slip and fall.

Really thoughtless. 1000s of developers have to rewrite their code with no thought to train the LLMs themselves to help the developers ahead of the release of so many breaking changes.

[endolith]

And they included a "migrate" tool that only migrates half of the code and leaves the rest broken. 😒

[rattrayalex]

@endolith can you please share a gist with examples of code that was not fully migrated, so we can fix the migration tool? (cc @morgante)

[endolith]

@rattrayalex #742 (reply in thread)

[rattrayalex]

Thank you, sorry we missed that!

[gojira]

@rattrayalex @RobertCraigie - hey there - Keiji from Azure OpenAI here!

We have written a more extended Azure OpenAI migration guide in our documentation: https://aka.ms/oai/v1-python-migration

Would it be possible to add it to the main body of this Guide?

[rattrayalex]

Awesome, thank you @gojira! added.

[gojira]

Excellent thank you!

[rattrayalex]

By the way, we're working with the team at Grit to add auto-migration for Azure.

[gojira]

Oh sweet! That's epic! We'd be happy to test it.

[rattrayalex]

Sorry if our docs were unclear! I know there are a lot of technical terms in there. You might try pasting them into ChatGPT and asking it to clarify in Italian.

[giopoi]

II don't want to use WSL Il giorno gio 9 nov 2023 alle ore 18:09 Alex Rattray < ***@***.***> ha scritto:

…

You need to use Windows Subsystem for Linux (WSL) for this. https://learn.microsoft.com/en-us/windows/wsl/install — Reply to this email directly, view it on GitHub <#742 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAMXYQQCMXDSAJLB5HT2WYTYDUE25AVCNFSM6AAAAAA7DOKUK2VHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM3TKMRVGAZTI> . You are receiving this because you commented.Message ID: ***@***.***>

-- *Giovanni Poidomani*

[rattrayalex]

If you aren't able to use WSL to install grit to help migrate your code automatically, you can migrate your code by hand by following the guide above, or continue to use the older SDK with pip install openai==0.28.0 if you don't need to upgrade.

[exanni]

still don't understand how to use grit

[rattrayalex]

What is the difficulty you're running into? The docs are here: https://docs.grit.io/

Note that if you're on Windows, you'll need to use WSL (Windows Subsystem for Linux).

cc @morgante in case there are follow-up questions

[abi]

Did openai.Timeout get changed to openai.APITimeoutError? Don't see that in the list of name changes.

[zekesarosi]

Also it seems like you can't set a timeout parameter in a Chat Completions request anymore which is very annoying

[rattrayalex]

@zekesarosi what gives you that impression? It's client.chat.completions.create(model=…, messages=…, timeout=30)

[rattrayalex]

@abi can you link to the old openai.Timeout?

[zekesarosi]

apologies @rattrayalex
you are correct the timeout param still works, for some reason an app that I was running in the past all of a sudden started throwing errors about timeout being an invalid parameter. It was an error on my end and I didn't investigate enough before commenting.

[abi]

@rattrayalex See Timeout here: https://platform.openai.com/docs/guides/error-codes/python-library-error-types

[TheSoldiersDream]

I am stuck with a bunch of bots that no longer work

[NguyenGPhuc]

Having the exact same problem. I was able to connect and get a response before with the old codes. This newer one makes no sense to me. Where exactly I'm suppose to place my api key?

client = OpenAI(
api_key=os.environ['OPENAI_API_KEY'], # this is also the default, it can be omitted
)

[earosenfeld]

is there a working example of this new code?

[rattrayalex]

Working examples demoing v1 are available here: https://github.com/openai/openai-python/tree/main/examples

[skyl]

Putting api keys in environment variables is good practice compared to hardcoding them into your scripts. Maybe ask ChatGPT about it ;)

[FlyingFathead]

Just in the past days, I've been working on my Telegram bot code that uses the openai package version 1.6.1. I noticed how much the API request has changed, here's a piece of my WIP code, and although disorganized, it's been working pretty well: https://github.com/FlyingFathead/TelegramBot-OpenAI-API/blob/b5c315e8dc46109279659a3adf6a997791cdbe35/main.py#L411

As you can see from the code, I had to more or less disassemble the payload and rebuild the request to use an async httpx client. Although I agree that the update caused a massive extra hassle, at least it's now been working without issues so far and with an up-to-date openai library. You can take a peek at the link to see if it's of any use in your particular implementation.

Like someone else commented, GPT-4 is not of much help with these API changes at the moment due to it not having been updated on any of this. However, you can i.e. try to have GPT-4 take a peek at the new implementation and/or A/B compare it with yours if it can come up with something that suits your needs.

[RockChinQ]

Hi there. While I use openai.proxies, I found that there's no proxies attr in openai module, but such code to set forward proxies is valid

openai.proxies = {
    'http': 'http://127.0.0.1:7890',
    'https': 'http://127.0.0.1:7890',
}

Should proxies attr be added to the module or add it to stub files to make it more intuitive?

[rattrayalex]

Proxies will not be added to the module client; please instantiate a client for that: https://github.com/openai/openai-python#configuring-the-http-client

[Jerry-hyl]

Hello! Have you solve the problem? I still have no idea how to to set proxies for this new version of api, could you give me a demo?

[endolith]

openai migrate breaks code because it doesn't make enough changes:

TypeError: 'ChatCompletion' object is not subscriptable

You need to make additional changes by hand like:

-output = completion['choices'][0]['message']['content']
-prom = completion['usage']['prompt_tokens']
-comp = completion['usage']['completion_tokens']
+output = completion.choices[0].message.content
+prom = completion.usage.prompt_tokens
+comp = completion.usage.completion_tokens

[morgante]

Yes, we can fix this.

[morgante]

@endolith Can you try the latest migration command? We added a fix and test case based on your example.

[endolith]

I had already fixed it manually

[phongvants123]

how did you fix it

[PierrunoYT]

https://platform.openai.com/docs/api-reference/chat

[MrPereir4]

I'm encountering issues when trying to use the OpenAI API with AWS Lambda following an update that involves pydantic. It's reporting that there is no module named pydantic_core. I am using Python 3.9 for both the Lambda function code and the OpenAI API layer.

[ERROR] Runtime.ImportModuleError: Unable to import module 'lambda_function': No module named 'pydantic_core._pydantic_core' Traceback (most recent call last):

I'm receiving the error message above from the CloudWatch error log.

[andrehenriquemendes]

Did you solve it?

[rattrayalex]

A separate GitHub issue would have been more appropriate for this. However, if any of the tips in pydantic/pydantic#6557 fix your issue, you may not need to file one.

[ResearcherSara]

Hello and thanks, I have a question and really thankful if you can help me out. I have installed openai(1.7.1), my python version is 3.11.7, in the first step of my code which is import openai in visual studio on windows I got "dynamic module does not define module export function (PyInit__pydantic_core)" error. I checked my pydantic(2.5.3) and pydantic_core(2.15.0), they seem fine. This sounds a bit weird for me because I am unable to import openai in any circumstances.

[EdricSouza]

What can I do about 'OpenAIError'? because I need the exception:

i = 0
embeddings = []
for text in df['text']:
     time.sleep(2)
     print(i)
     try:
         embedding = openai.Embedding.create(input=text, engine='text-embedding-ada-002')['data'][0]['embedding']
         print("Embedding the text")
         embeddings.append(embedding)
     except OpenAIError as e:
         if e.status == 429: # Check if the error is a Rate Limit
             print("Rate limit error, waiting 20 seconds before trying again")
             time.sleep(20)
             # Retry the operation that caused the error
         else:
             # Handle other types of OpenAI errors here
             raise # Reraise the error so it can be handled at another level
     i+=1

[rattrayalex]

You're looking for openai.APIError, see https://github.com/openai/openai-python?tab=readme-ov-file#handling-errors

Note that the new library auto-retries automatically, so you may not need this code.

[Abs73]

After following the guide for adding to Windows. I get the same result. what could be the issue?

[Abs73]

[phongvants123]

i got the same error, how did you solve it ?

[Madtolla]

Hello
I tried the solution for linux on raspberry Pi4 but It didn't worked Traceback (most recent call last): File "/home/office/home/office/bin/openai", line 8, in <module> sys.exit(main()) ^^^^^^ File "/home/office/home/office/lib/python3.11/site-packages/openai/cli/_cli.py", line 129, in main _main() File "/home/office/home/office/lib/python3.11/site-packages/openai/cli/_cli.py", line 209, in _main parsed.func( File "/home/office/home/office/lib/python3.11/site-packages/openai/cli/_tools/migrate.py", line 56, in migrate subprocess.check_call([grit_path, "apply", "openai", *args.unknown_args]) File "/usr/lib/python3.11/subprocess.py", line 408, in check_call retcode = call(*popenargs, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/lib/python3.11/subprocess.py", line 389, in call with Popen(*popenargs, **kwargs) as p: ^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/lib/python3.11/subprocess.py", line 1024, in __init__ self._execute_child(args, executable, preexec_fn, close_fds, File "/usr/lib/python3.11/subprocess.py", line 1901, in _execute_child raise child_exception_type(errno_num, err_msg, err_filename) FileNotFoundError: [Errno 2] No such file or directory: PosixPath('/home/office/.cache/openai-python/.install/bin/marzano')
here is the error message
and what I did is just worte openai migrate
I need your help ASAP

[rattrayalex]

cc @morgante

[RomanG-PT]

how to enter in Ubuntu this 3 lines as it is accepts only one and than run it....
curl -fsSL https://docs.grit.io/install | bash
grit install
grit apply openai

[morgante]

Can you share what the output is?

[RomanG-PT]

here it is:

[morgante]

Try sourcing it, run the last line of the install output.

[RomanG-PT]

thank you for your help
I am so new to all this...
do you mean "run source /home/roman/.bashrc" ?

[morgante]

Yes, run this command:

source /home/roman/.bashrc

[RomanG-PT]

maybe I am doing something very stupid, but nothing happens:

[morgante]

It's expected for nothing you happen from that. You then need to actually run the migration.

If you join our Discord I'm happy to help you - https://discord.com/invite/ARExD4gvFB

[EdricSouza]

Can anyone help me with this code? I'm trying to run this function, but this error appears:
(At first the idea was to use openai.embeddings_utils, but with the openai update, I don't know what to do)

File "c:\Users", line 254, in distances_from_embeddings
distances = np.dot(embeddings, query_embedding) / (np.linalg.norm(embeddings) * np.linalg.norm(query_embedding))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: shapes (9,) and (1,1536) not aligned: 9 (dim 0) != 1 (dim 0)

def distances_from_embeddings(query_embedding, embeddings, distance_metric='cosine'):
    if distance_metric == 'cosine':
        query_embedding = np.expand_dims(query_embedding, 0)
        print("Forma original de embeddings:", embeddings.shape)
        print("Forma original de query_embedding:", query_embedding.shape)
        distances = np.dot(embeddings, query_embedding) / (np.linalg.norm(embeddings) * np.linalg.norm(query_embedding))
        distances = 1 - distances  
    else:
        raise ValueError("Métrica de distância não suportada. Utilize 'cosine'.")

    return distances

def create_context(question, df, max_len=1800, size="ada"):

    # Obter a embeddings para a pergunta que foi feita
    q_embeddings = client.embeddings.create(input=question,model='text-embedding-ada-002').data[0].embedding

    # Obter as distâncias a partir dos embeddings
    df['distances'] = distances_from_embeddings(q_embeddings, df['embeddings'].values, distance_metric='cosine')


    returns = []
    cur_len = 0

    # Classifique por distância e adicione o texto ao contexto
    for i, row in df.sort_values('distances', ascending=True).iterrows():
        
        # Adicionar o comprimento do texto ao comprimento atual
        cur_len += row['n_tokens'] + 4
        
        # Se o contexto for muito longo, quebre
        if cur_len > max_len:
            break
        
        # Caso contrário, adicione-o ao texto que está sendo retornado
        returns.append(row["text"])

    # Retornar o contexto
    return "\n\n###\n\n".join(returns)

[rattrayalex]

For help developing with the OpenAI API from fellow developers, I recommend the OpenAI Discord server!

[Stratosss]

Hello, I am a little bit confused with the steps:
"In the WSL terminal, cd into the appropriate directory (e.g., cd /mnt/c/Users/Myself/my/code/) and then run the following commands:
curl -fsSL https://docs.grit.io/install | bash
grit install
grit apply openai"

1. What do you mean by "appopriate directory"? Is it a random directory i have to cd in?
2. I selected a random directory, run the first command which was executed fine and returned: "grit was installed successfully to ~/.grit/bin/grit"
3. When I try to perform "grit install" no matter the directory I'm in it says:

Command 'grit' not found, did you mean:
  command 'grip' from deb grip (4.2.0-3)
  command 'grig' from deb grig (0.8.1-6)
  command 'grim' from deb grim (1.4.0+ds-1)
  command 'crit' from deb criu (3.16.1-2)
  command 'git' from deb git (1:2.34.1-1ubuntu1.10)

I tried the above while being in the randomly chosen directory and in ~/.grit/bin/grit. Nothing works. Please help!
Running it on Windows 11 Version 10.0.22621 Build 22621
and have WSL already installed

thanks!

[morgante]

You need to CD into the directory where your OpenAI code is.

After installing Grit, you need to source it - or you can just reference it directly by path. Try this, from your OpenAI dir:

~/.grit/bin/grit apply openai

If you require further assistance, please join our Discord: https://docs.grit.io/discord

[MeranTumeh]

Hi @rattrayalex,
I'm facing an issue with (openai.Deployment) since it was removed
I'm using Azure OpenAI, what is the alternative for it?

[rattrayalex]

I believe that using Azure's SDKs is best for managing Azure resources like cognitive services deployments!

cc @kristapratico - is that right?

[kristapratico]

That's correct. Deployments APIs were deprecated in the v0.X version of the openai library and so they were not added for v1.X. For all code managing deployments, please use the azure-mgmt-cognitiveservices client library. This library has additional options for working with deployments, like setting a deployment name or specific model version, and is the place where new features for deployments will be added.

Here are some samples for interacting with deployments:

Create a deployment
Retrieve a deployment
List deployments
Delete a deployment

[rattrayalex]

Thanks Krista! I've updated the migration guide above to specify this.

[PierrunoYT]

The Docs on the Open AI Website is outdated

[rattrayalex]

Can you provide a link to the part(s) of the docs which need updating?

[PierrunoYT]

This is the only doc which works https://platform.openai.com/docs/api-reference/images and Chat GPT is messing up things with it

[PierrunoYT]

This does not work https://platform.openai.com/docs/guides/images/usage?context=node

[aiakubovich]

terrible update. mess up all scripts. APIs are not consistent e.g. base_url for OpenAI() and azure_endpoint for AzureOpenAI()

[PierrunoYT]

https://platform.openai.com/docs/api-reference/chat

[411A]

Thanks to migration from aiohttp to httpx and not providing any methods or proper documentations, your update broke my aiohttp optimization functions (openai.aiosession.set(ClientSession()) that all my openai functions depend on. Will stay on previous version without TTS. ZERO backward compatibility.

[PierrunoYT]

https://platform.openai.com/docs/api-reference/chat