Skip to content

docs(sagemaker): distinguish native sagemaker_chat streaming from legacy fake streaming #637

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

Open
devin-ai-integration wants to merge 1 commit into
base: main
Choose a base branch
from
Open

docs(sagemaker): distinguish native sagemaker_chat streaming from legacy fake streaming #637

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
4 changes: 3 additions & 1 deletion docs/providers/aws_sagemaker.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,7 +36,7 @@ response = completion(
```

### Usage - Streaming
Sagemaker currently does not support streaming - LiteLLM fakes streaming by returning chunks of the response string
The legacy `sagemaker/` completion route calls the `/invocations` endpoint, which does not stream; LiteLLM fakes streaming by chunking the full response string. If your endpoint speaks the HF Messages API, use the `sagemaker_chat/` route instead, which streams natively (see [Sagemaker Messages API](#sagemaker-messages-api) below)

```python
import os
Expand Down Expand Up @@ -423,6 +423,8 @@ Use route `sagemaker_chat/*` to route to Sagemaker Messages API
model: sagemaker_chat/<your-endpoint-name>
```

Unlike the legacy `sagemaker/` route, `sagemaker_chat/` streams natively; it calls the SageMaker `/invocations-response-stream` endpoint and forwards each decoded event as soon as it arrives, so client time-to-first-token and chunk cadence track the endpoint's own streaming. Pass `stream=True` (LiteLLM includes `"stream": true` in the signed request body) to enable it

<Tabs>
<TabItem value="sdk" label="SDK">

Expand Down