feat(streaming): support external async token generators #1286
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
Add ability to pass custom async token generators to `stream_async`, enabling integration with external LLMs or custom streaming sources. Update docs and add tests for output rails interaction and edge cases with external generators.
Documentation preview |
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## develop #1286 +/- ##
===========================================
+ Coverage 69.78% 69.82% +0.03%
===========================================
Files 161 161
Lines 16057 16061 +4
===========================================
+ Hits 11206 11214 +8
+ Misses 4851 4847 -4
Flags with carried forward coverage won't be shown. Click here to find out more.
🚀 New features to boost your workflow:
|
trebedea
approved these changes
Jul 14, 2025
| - You want to use a different LLM provider that has its own streaming API | ||
| - You have pre-generated responses that you want to stream through guardrails | ||
| - You want to implement custom token generation logic | ||
| - You want to test your output rails or its config in streaming mode wihtout relying on an LLM which generates stochastic outputs. |
There was a problem hiding this comment.
Suggested change
| - You want to test your output rails or its config in streaming mode wihtout relying on an LLM which generates stochastic outputs. | |
| - You want to test your output rails or its config in streaming mode on predefined responses without actually relying on an actual LLM generation. |
There was a problem hiding this comment.
Corrected a mistype and also changed the wording. Is this the correct usage, on predefined / given assistant responses only with output rails?
| app = LLMRails(config) | ||
|
|
||
| async def my_token_generator() -> AsyncIterator[str]: | ||
| # this could be from OpenAI, Anthropic, or any other source |
There was a problem hiding this comment.
Suggested change
| # this could be from OpenAI, Anthropic, or any other source | |
| # This could be from OpenAI API, Anthropic API, or any other LLM API that already has a streaming token generator. Mocking the stream here, for a simple example. |
|
|
||
| # use the external generator with guardrails | ||
| async for chunk in app.stream_async( | ||
| messages=history, |
There was a problem hiding this comment.
missing history - can we put a simple example?
|
|
||
| When using an external generator: | ||
|
|
||
| - The internal LLM generation is completely bypassed |
There was a problem hiding this comment.
Suggested change
| - The internal LLM generation is completely bypassed | |
| - The internal LLM generation in the Guardrails runtime is completely bypassed, the LLM responses are given by the external generator |
| When using an external generator: | ||
|
|
||
| - The internal LLM generation is completely bypassed | ||
| - Output rails are still applied if configured |
There was a problem hiding this comment.
Suggested change
| - Output rails are still applied if configured | |
| - Output rails are still applied to the LLM responses returned by the external generator, if configured |
| ) | ||
| else: | ||
| return generator | ||
|
|
There was a problem hiding this comment.
Can this still be used with explain / generation options?
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.
Add ability to pass custom async token generators to
stream_async, enabling integration with external LLMs or custom streaming sources. Update docs and add tests for output rails interaction and edge cases with external generators.Note: