Skip to content

Fix responses structured outputs + cache retrieval error #9123

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

Closed
Olocool17 wants to merge 2 commits into from
+9 −3
Closed

Fix responses structured outputs + cache retrieval error #9123

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
ed237fe
fix: structured outputs when using responses model
Olocool17 Dec 11, 2025
5349ce2
fix: cache retrieval error when using responses model
Olocool17 Dec 11, 2025
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
1 change: 0 additions & 1 deletion dspy/clients/cache.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -122,7 +122,6 @@ def get(self, request: dict[str, Any], ignored_args_for_cache_key: list[str] | N
if hasattr(response, "usage"):
# Clear the usage data when cache is hit, because no LM call is made
response.usage = {}
response.cache_hit = True
return response

def put(
Expand Down
11 changes: 9 additions & 2 deletions dspy/clients/lm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,6 +6,7 @@
from typing import Any, Literal, cast

import litellm
import pydantic
from anyio.streams.memory import MemoryObjectSendStream
from asyncer import syncify

Expand Down Expand Up @@ -163,7 +164,7 @@ def forward(

self._check_truncation(results)

if not getattr(results, "cache_hit", False) and dspy.settings.usage_tracker and hasattr(results, "usage"):
if dspy.settings.usage_tracker and hasattr(results, "usage"):
settings.usage_tracker.add_usage(self.model, dict(results.usage))
return results

Expand Down Expand Up @@ -201,7 +202,7 @@ async def aforward(

self._check_truncation(results)

if not getattr(results, "cache_hit", False) and dspy.settings.usage_tracker and hasattr(results, "usage"):
if dspy.settings.usage_tracker and hasattr(results, "usage"):
settings.usage_tracker.add_usage(self.model, dict(results.usage))
return results

Expand Down Expand Up @@ -501,6 +502,12 @@ def _convert_chat_request_to_responses_request(request: dict[str, Any]):
# Convert `response_format` to `text.format` for Responses API
if "response_format" in request:
response_format = request.pop("response_format")
if isinstance(response_format, type) and issubclass(response_format, pydantic.BaseModel):
Copy link
Collaborator

@chenmoneygithub chenmoneygithub Dec 11, 2025

Choose a reason for hiding this comment

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

Instead of one more serialization, can we use the text_format arg?

response_format = request.pop("response_format")
request["text_format"] = response_format

Copy link
Contributor Author

@Olocool17 Olocool17 Dec 15, 2025

Choose a reason for hiding this comment

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

Unfortunately not, I tried just the text_format arg at first, but I think litellm refused to play nice with it.
Serializing it ourselves might a bit more verbose but being explicit is warranted here, since we modify the Pydantic model's model_json_schema in JSONAdapter manually anyways.

response_format = {
"name": response_format.__name__,
"type": "json_schema",
"schema": response_format.model_json_schema(),
}
text = request.pop("text", {})
request["text"] = {**text, "format": response_format}

Expand Down