InstrumentedModel and FallbackModel fixes #1121
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Docs Preview
|
alexmojaki
changed the title
alexmojaki
changed the title
alexmojaki
commented
Mar 14, 2025
dmontagu
reviewed
Mar 14, 2025
| """The system / model provider, n/a for fallback models.""" | ||
| return None | ||
| def system(self) -> str: | ||
| return 'fallback' |
There was a problem hiding this comment.
Should this include the systems of models that are fallen back to? E.g. 'fallback:openai:anthropic:function' or similar?
Related, if this never ends up in the otel system (e.g. because we use the value from the actual model that made the request), I think we don't necessarily need to include the fallback systems here but I would appreciate a comment explaining that.
(If it does end up in the otel, then it seems problematic. I'm assuming it doesn't though.)
dmontagu
reviewed
Mar 14, 2025
dmontagu
reviewed
Mar 14, 2025
dmontagu
reviewed
Mar 14, 2025
Comment on lines
69
to
73
| span = get_current_span() | ||
| if span.is_recording(): | ||
| attributes = getattr(span, 'attributes', {}) | ||
| if attributes.get('gen_ai.request.model') == self.model_name: | ||
| span.set_attributes(InstrumentedModel.model_attributes(model)) |
There was a problem hiding this comment.
I feel like this should be outside this try-except. Like, errors with otel shouldn't trigger fallback, it means something more fundamental is wrong
dmontagu
approved these changes
Mar 14, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
TypeError: cannot pickle '_thread.RLock' objecterror reported in https://pydanticlogfire.slack.com/archives/C083V7PMHHA/p1741909209116159 by no longer settingmodel_usedonModelResponse.Model.systemnon-optional, becausegen_ai.systemis an important attribute and OTel doesn't support null attribute values.Model.systemdirectly returns value used forgen_ai.systemto help with Fix instrumentation of FallbackModel #1076 (comment). @Kludex we still need to do something about bedrock, azure, and maybe others, but I've stuffed too much in here already.