Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

response_format support #12

Open
wants to merge 16 commits into
base: main
Choose a base branch
from
Open

response_format support #12

Changes from all commits
Commits
Show all changes
16 commits
Select commit Hold shift + click to select a range
a5744d0
response_format support
thiwa8 Jun 17, 2024
b0d4816
"documentation to response_format"
thiwa8 Jun 17, 2024
f3bbedf
"fomat:"
thiwa8 Jun 17, 2024
6bd33e6
"bugfix"
thiwa8 Jun 17, 2024
58065f9
} bug
thiwa8 Jun 17, 2024
079a02b
"response format"
thiwa8 Jun 17, 2024
9b2536e
"
thiwa8 Jun 17, 2024
4a943d8
s
thiwa8 Jun 17, 2024
475d056
t
thiwa8 Jun 17, 2024
a31153f
serialize json
thiwa8 Jun 17, 2024
226e665
revert to repsonse_format=json object
thiwa8 Jun 17, 2024
d23a15e
fix
thiwa8 Jun 17, 2024
e3c67a0
s
thiwa8 Jun 17, 2024
208b426
sss
thiwa8 Jun 17, 2024
25c4ca7
update anthropic
thiwa8 Jun 17, 2024
be7a2b8
response_format instad of format param
thiwa8 Jun 17, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
27 changes: 21 additions & 6 deletions unify/clients.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,4 @@
from typing import AsyncGenerator, Dict, Generator, List, Optional, Union

import openai
import requests
from unify.exceptions import BadRequestError, UnifyError, status_error_map
Expand Down Expand Up @@ -130,6 +129,7 @@ def generate( # noqa: WPS234, WPS211
temperature: Optional[float] = 1.0,
stop: Optional[List[str]] = None,
stream: bool = False,
response_format: Optional[dict] = {"type": "text"}
) -> Union[Generator[str, None, None], str]: # noqa: DAR101, DAR201, DAR401
"""Generate content using the Unify API.

Expand Down Expand Up @@ -157,6 +157,9 @@ def generate( # noqa: WPS234, WPS211
If False, generates content as a single response.
Defaults to False.

response_format (Optional[str]): The format of the response that will be used for openai models
Will be sent as a parameter inside response_format= {type : format} in the request body
format can be either "json_object" or "text". User has to also explicitly ask for JSON in the prompt!
Returns:
Union[Generator[str, None, None], str]: If stream is True,
returns a generator yielding chunks of content.
Expand All @@ -179,11 +182,11 @@ def generate( # noqa: WPS234, WPS211
return self._generate_stream(contents, self._endpoint,
max_tokens=max_tokens,
temperature=temperature,
stop=stop)
stop=stop, response_format=response_format)
return self._generate_non_stream(contents, self._endpoint,
max_tokens=max_tokens,
temperature=temperature,
stop=stop)
stop=stop, response_format=response_format)

def get_credit_balance(self) -> float:
# noqa: DAR201, DAR401
Expand Down Expand Up @@ -218,6 +221,7 @@ def _generate_stream(
max_tokens: Optional[int] = 1024,
temperature: Optional[float] = 1.0,
stop: Optional[List[str]] = None,
response_format: Optional[dict] = {"type": "text"},
) -> Generator[str, None, None]:
try:
chat_completion = self.client.chat.completions.create(
Expand All @@ -227,6 +231,7 @@ def _generate_stream(
temperature=temperature,
stop=stop,
stream=True,
response_format=response_format,
)
for chunk in chat_completion:
content = chunk.choices[0].delta.content # type: ignore[union-attr]
Expand All @@ -243,6 +248,7 @@ def _generate_non_stream(
max_tokens: Optional[int] = 1024,
temperature: Optional[float] = 1.0,
stop: Optional[List[str]] = None,
response_format: Optional[dict] = {"type": "text"},
) -> str:
try:
chat_completion = self.client.chat.completions.create(
Expand All @@ -252,7 +258,7 @@ def _generate_non_stream(
temperature=temperature,
stop=stop,
stream=False,
)
response_format= response_format)
self.set_provider(
chat_completion.model.split( # type: ignore[union-attr]
"@",
Expand Down Expand Up @@ -413,6 +419,7 @@ async def generate( # noqa: WPS234, WPS211
temperature: Optional[float] = 1.0,
stop: Optional[List[str]] = None,
stream: bool = False,
response_format: Optional[dict] = {"type": "text"},
) -> Union[AsyncGenerator[str, None], str]: # noqa: DAR101, DAR201, DAR401
"""Generate content asynchronously using the Unify API.

Expand Down Expand Up @@ -440,6 +447,10 @@ async def generate( # noqa: WPS234, WPS211
If False, generates content as a single response.
Defaults to False.

format (Optional[str]): The format of the response that will be used for openai models
Will be sent as a parameter inside response_format= {type : format} in the request body
format can be either "json_object" or "text". User has to also explicitly ask for JSON in the prompt!

Returns:
Union[AsyncGenerator[str, None], List[str]]: If stream is True,
returns an asynchronous generator yielding chunks of content.
Expand All @@ -460,8 +471,8 @@ async def generate( # noqa: WPS234, WPS211
raise UnifyError("You must provide either the user_prompt or messages!")

if stream:
return self._generate_stream(contents, self._endpoint, max_tokens=max_tokens, stop=stop, temperature=temperature)
return await self._generate_non_stream(contents, self._endpoint, max_tokens=max_tokens, stop=stop, temperature=temperature)
return self._generate_stream(contents, self._endpoint, max_tokens=max_tokens, stop=stop, temperature=temperature,response_format=response_format)
return await self._generate_non_stream(contents, self._endpoint, max_tokens=max_tokens, stop=stop, temperature=temperature,response_format=response_format)

async def _generate_stream(
self,
Expand All @@ -470,6 +481,7 @@ async def _generate_stream(
max_tokens: Optional[int] = None,
temperature: Optional[float] = 1.0,
stop: Optional[List[str]] = None,
response_format: Optional[dict] = {"type": "text"},
) -> AsyncGenerator[str, None]:
try:
async_stream = await self.client.chat.completions.create(
Expand All @@ -479,6 +491,7 @@ async def _generate_stream(
temperature=temperature,
stop=stop,
stream=True,
response_format= response_format,
)
async for chunk in async_stream: # type: ignore[union-attr]
self.set_provider(chunk.model.split("@")[-1])
Expand All @@ -493,6 +506,7 @@ async def _generate_non_stream(
max_tokens: Optional[int] = None,
temperature: Optional[float] = 1.0,
stop: Optional[List[str]] = None,
response_format: Optional[dict] = {"type": "text"}
) -> str:
try:
async_response = await self.client.chat.completions.create(
Expand All @@ -502,6 +516,7 @@ async def _generate_non_stream(
temperature=temperature,
stop=stop,
stream=False,
response_format = response_format,
)
self.set_provider(async_response.model.split("@")[-1]) # type: ignore
return async_response.choices[0].message.content.strip(" ") # type: ignore # noqa: E501, WPS219
Expand Down