You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Describe the bug
According to this documentation, logprobs is supported. However the response doesn't contain them. In order to ensure that I was sending a correct request, I simply swapped out OpenAI's endpoint and everything worked as expected. All logprobs were returned.
To Reproduce
async def get_openrouter_response(
messages: list[Message],
models: list[Model],
temperature: Temperature | None = None,
response_format: ResponseFormat | None = None,
*,
seed: int | None = None,
logprobs: bool = False,
top_logprobs: int | None = None,
) -> OpenRouterResponse:
"""
Get a response from the OpenRouter API.
"""
temperature = Temperature(value=0.7) if temperature is None else temperature
async with httpx.AsyncClient(timeout=5.0) as client:
response = await client.post(
url="https://openrouter.ai/api/v1/chat/completions",
# url="https://api.openai.com/v1/chat/completions",
headers={
"Authorization": f"Bearer {SETTINGS.openrouter_api_key}",
# "Authorization": f"Bearer {SETTINGS.openai_api_key}",
"HTTP-Referer": SITE_URL, # Optional, for including your app on openrouter.ai rankings.
"X-Title": APP_NAME, # Optional. Shows in rankings on openrouter.ai.
},
json={
# "models": models,
# "model": "gpt-4o",
"model": "openai/gpt-4o-2024-05-13",
"messages": [message.model_dump() for message in messages],
"temperature": temperature.value,
"seed": seed,
"response_format": (
None if response_format is None else {"type": response_format}
),
"logprobs": True,
"top_logprobs": top_logprobs,
},
)
Expected behavior
Return logprobs from supported providers and models.
The text was updated successfully, but these errors were encountered:
Describe the bug
According to this documentation, logprobs is supported. However the response doesn't contain them. In order to ensure that I was sending a correct request, I simply swapped out OpenAI's endpoint and everything worked as expected. All logprobs were returned.
To Reproduce
Expected behavior
Return logprobs from supported providers and models.
The text was updated successfully, but these errors were encountered: