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

Increase timeout for reasoning models + make o1 available by default #3954

Merged
merged 1 commit into from
Feb 11, 2025
Merged

Increase timeout for reasoning models + make o1 available by default #3954

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
15 changes: 12 additions & 3 deletions backend/onyx/llm/chat_llm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,6 +27,7 @@

from onyx.configs.app_configs import LOG_DANSWER_MODEL_INTERACTIONS
from onyx.configs.app_configs import MOCK_LLM_RESPONSE
from onyx.configs.chat_configs import QA_TIMEOUT
from onyx.configs.model_configs import (
DISABLE_LITELLM_STREAMING,
)
Expand All @@ -35,6 +36,7 @@
from onyx.llm.interfaces import LLM
from onyx.llm.interfaces import LLMConfig
from onyx.llm.interfaces import ToolChoiceOptions
from onyx.llm.utils import model_is_reasoning_model
from onyx.server.utils import mask_string
from onyx.utils.logger import setup_logger
from onyx.utils.long_term_log import LongTermLogger
Expand Down Expand Up @@ -229,25 +231,32 @@ class DefaultMultiLLM(LLM):
def __init__(
self,
api_key: str | None,
timeout: int,
model_provider: str,
model_name: str,
timeout: int | None = None,
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

making these None by default and setting the default within the init to allow for more complicated logic here

api_base: str | None = None,
api_version: str | None = None,
deployment_name: str | None = None,
max_output_tokens: int | None = None,
custom_llm_provider: str | None = None,
temperature: float = GEN_AI_TEMPERATURE,
temperature: float | None = None,
custom_config: dict[str, str] | None = None,
extra_headers: dict[str, str] | None = None,
extra_body: dict | None = LITELLM_EXTRA_BODY,
model_kwargs: dict[str, Any] | None = None,
long_term_logger: LongTermLogger | None = None,
):
self._timeout = timeout
if timeout is None:
if model_is_reasoning_model(model_name):
self._timeout = QA_TIMEOUT * 10 # Reasoning models are slow
else:
self._timeout = QA_TIMEOUT

self._temperature = GEN_AI_TEMPERATURE if temperature is None else temperature

self._model_provider = model_provider
self._model_version = model_name
self._temperature = temperature
self._api_key = api_key
self._deployment_name = deployment_name
self._api_base = api_base
Expand Down
7 changes: 3 additions & 4 deletions backend/onyx/llm/factory.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,6 @@

from onyx.chat.models import PersonaOverrideConfig
from onyx.configs.app_configs import DISABLE_GENERATIVE_AI
from onyx.configs.chat_configs import QA_TIMEOUT
from onyx.configs.model_configs import GEN_AI_MODEL_FALLBACK_MAX_TOKENS
from onyx.configs.model_configs import GEN_AI_TEMPERATURE
from onyx.db.engine import get_session_context_manager
Expand Down Expand Up @@ -88,8 +87,8 @@ def _create_llm(model: str) -> LLM:


def get_default_llms(
timeout: int = QA_TIMEOUT,
temperature: float = GEN_AI_TEMPERATURE,
timeout: int | None = None,
temperature: float | None = None,
additional_headers: dict[str, str] | None = None,
long_term_logger: LongTermLogger | None = None,
) -> tuple[LLM, LLM]:
Expand Down Expand Up @@ -138,7 +137,7 @@ def get_llm(
api_version: str | None = None,
custom_config: dict[str, str] | None = None,
temperature: float | None = None,
timeout: int = QA_TIMEOUT,
timeout: int | None = None,
additional_headers: dict[str, str] | None = None,
long_term_logger: LongTermLogger | None = None,
) -> LLM:
Expand Down
4 changes: 2 additions & 2 deletions backend/onyx/llm/llm_provider_options.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,11 +29,11 @@ class WellKnownLLMProviderDescriptor(BaseModel):
OPEN_AI_MODEL_NAMES = [
"o3-mini",
"o1-mini",
"o1-preview",
"o1-2024-12-17",
"o1",
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

o1 should be higher up than preview

"gpt-4",
"gpt-4o",
"gpt-4o-mini",
"o1-preview",
"gpt-4-turbo",
"gpt-4-turbo-preview",
"gpt-4-1106-preview",
Expand Down
11 changes: 11 additions & 0 deletions backend/onyx/llm/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -543,3 +543,14 @@ def model_supports_image_input(model_name: str, model_provider: str) -> bool:
f"Failed to get model object for {model_provider}/{model_name}"
)
return False


def model_is_reasoning_model(model_name: str) -> bool:
_REASONING_MODEL_NAMES = [
"o1",
"o1-mini",
"o3-mini",
"deepseek-reasoner",
"deepseek-r1",
]
return model_name.lower() in _REASONING_MODEL_NAMES
19 changes: 6 additions & 13 deletions web/src/lib/hooks.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -647,11 +647,11 @@ export const useUserGroups = (): {

const MODEL_DISPLAY_NAMES: { [key: string]: string } = {
// OpenAI models
"o1-2025-12-17": "O1 (December 2025)",
"o3-mini": "O3 Mini",
"o1-mini": "O1 Mini",
"o1-preview": "O1 Preview",
o1: "O1",
"o1-2025-12-17": "o1 (December 2025)",
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

matching case of OpenAI's docs/chatgpt

"o3-mini": "o3 Mini",
"o1-mini": "o1 Mini",
"o1-preview": "o1 Preview",
o1: "o1",
"gpt-4": "GPT 4",
"gpt-4o": "GPT 4o",
"gpt-4o-2024-08-06": "GPT 4o (Structured Outputs)",
Expand Down Expand Up @@ -753,14 +753,7 @@ export function getDisplayNameForModel(modelName: string): string {
}

export const defaultModelsByProvider: { [name: string]: string[] } = {
openai: [
"gpt-4",
"gpt-4o",
"gpt-4o-mini",
"o3-mini",
"o1-mini",
"o1-preview",
],
openai: ["gpt-4", "gpt-4o", "gpt-4o-mini", "o3-mini", "o1-mini", "o1"],
bedrock: [
"meta.llama3-1-70b-instruct-v1:0",
"meta.llama3-1-8b-instruct-v1:0",
Expand Down
Loading