phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-27 03:34:10 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

(feat) Support audio, modalities params (#6304)

Browse source 
* add audio, modalities param

* add test for gpt audio models

* add get_supported_openai_params for GPT audio models

* add supported params for audio

* test_audio_output_from_model

* bump openai to openai==1.52.0

* bump openai on pyproject

* fix audio test

* fix test mock_chat_response

* handle audio for Message

* fix handling audio for OAI compatible API endpoints

* fix linting

* fix mock dbrx test
This commit is contained in:
Ishaan Jaff 2024-10-18 19:14:25 +05:30 • committed by GitHub
parent e35fc3203e
commit 13e0b3f626
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
15 changed files with 290 additions and 23 deletions
Download patch file Download diff file Expand all files Collapse all files

18
litellm/main.py
View file

@ -147,6 +147,8 @@ from .llms.vertex_ai_and_google_ai_studio.vertex_embeddings.embedding_handler im
from .llms.watsonx import IBMWatsonXAI
from .types.llms.openai import (
ChatCompletionAssistantMessage,
ChatCompletionAudioParam,
ChatCompletionModality,
ChatCompletionUserMessage,
HttpxBinaryResponseContent,
)
@ -287,6 +289,8 @@ async def acompletion(
stop=None,
max_tokens: Optional[int] = None,
max_completion_tokens: Optional[int] = None,
modalities: Optional[List[ChatCompletionModality]] = None,
audio: Optional[ChatCompletionAudioParam] = None,
presence_penalty: Optional[float] = None,
frequency_penalty: Optional[float] = None,
logit_bias: Optional[dict] = None,
@ -327,6 +331,8 @@ async def acompletion(
stop(string/list, optional): - Up to 4 sequences where the LLM API will stop generating further tokens.
max_tokens (integer, optional): The maximum number of tokens in the generated completion (default is infinity).
max_completion_tokens (integer, optional): An upper bound for the number of tokens that can be generated for a completion, including visible output tokens and reasoning tokens.
modalities (List[ChatCompletionModality], optional): Output types that you would like the model to generate for this request. You can use `["text", "audio"]`
audio (ChatCompletionAudioParam, optional): Parameters for audio output. Required when audio output is requested with modalities: ["audio"]
presence_penalty (float, optional): It is used to penalize new tokens based on their existence in the text so far.
frequency_penalty: It is used to penalize new tokens based on their frequency in the text so far.
logit_bias (dict, optional): Used to modify the probability of specific tokens appearing in the completion.
@ -366,6 +372,8 @@ async def acompletion(
"stop": stop,
"max_tokens": max_tokens,
"max_completion_tokens": max_completion_tokens,
"modalities": modalities,
"audio": audio,
"presence_penalty": presence_penalty,
"frequency_penalty": frequency_penalty,
"logit_bias": logit_bias,
@ -670,6 +678,8 @@ def completion( # type: ignore # noqa: PLR0915
stop=None,
max_completion_tokens: Optional[int] = None,
max_tokens: Optional[int] = None,
modalities: Optional[List[ChatCompletionModality]] = None,
audio: Optional[ChatCompletionAudioParam] = None,
presence_penalty: Optional[float] = None,
frequency_penalty: Optional[float] = None,
logit_bias: Optional[dict] = None,
@ -712,6 +722,8 @@ def completion( # type: ignore # noqa: PLR0915
stop(string/list, optional): - Up to 4 sequences where the LLM API will stop generating further tokens.
max_tokens (integer, optional): The maximum number of tokens in the generated completion (default is infinity).
max_completion_tokens (integer, optional): An upper bound for the number of tokens that can be generated for a completion, including visible output tokens and reasoning tokens.
modalities (List[ChatCompletionModality], optional): Output types that you would like the model to generate for this request.. You can use `["text", "audio"]`
audio (ChatCompletionAudioParam, optional): Parameters for audio output. Required when audio output is requested with modalities: ["audio"]
presence_penalty (float, optional): It is used to penalize new tokens based on their existence in the text so far.
frequency_penalty: It is used to penalize new tokens based on their frequency in the text so far.
logit_bias (dict, optional): Used to modify the probability of specific tokens appearing in the completion.
@ -816,6 +828,8 @@ def completion( # type: ignore # noqa: PLR0915
"stream_options",
"stop",
"max_completion_tokens",
"modalities",
"audio",
"max_tokens",
"presence_penalty",
"frequency_penalty",
@ -975,6 +989,8 @@ def completion( # type: ignore # noqa: PLR0915
stop=stop,
max_tokens=max_tokens,
max_completion_tokens=max_completion_tokens,
modalities=modalities,
audio=audio,
presence_penalty=presence_penalty,
frequency_penalty=frequency_penalty,
logit_bias=logit_bias,
@ -1515,7 +1531,7 @@ def completion( # type: ignore # noqa: PLR0915
## COMPLETION CALL
try:
if litellm.OpenAIO1Config().is_model_o1_reasoning_model(model=model):
if litellm.openAIO1Config.is_model_o1_reasoning_model(model=model):
response = openai_o1_chat_completions.completion(
model=model,
messages=messages,