From 367cc7200ee6aa60cea36134b627ca526b40cc1c Mon Sep 17 00:00:00 2001 From: zhanghao Date: Thu, 8 Oct 2026 16:59:35 +0800 Subject: [PATCH] docs(profile): list continuation_gateway under scaling and health It is a released project in the organisation but was missing from the front page. It sits beside hang-watcher: both keep a model serving when an engine stops making progress, hang-watcher by restarting the engine, continuation_gateway by resuming the stream the client was receiving. --- profile/README.md | 1 + 1 file changed, 1 insertion(+) diff --git a/profile/README.md b/profile/README.md index 4490260..cd016f8 100644 --- a/profile/README.md +++ b/profile/README.md @@ -20,6 +20,7 @@ the deployment repository, with the install guide and the architecture. | | [slo-scaler-decision-gen](https://github.com/modelsphere/slo-scaler-decision-gen) | Turns SLO targets and live signals into replica decisions | | | [slo-api](https://github.com/modelsphere/slo-api) | HTTP API to read and change a service's SLO | | | [hang-watcher](https://github.com/modelsphere/hang-watcher) | Restarts an engine that stopped making progress | +| | [continuation_gateway](https://github.com/modelsphere/continuation_gateway) | Resumes a streamed completion that stalls or drops mid-generation, so the client gets one complete stream | | Operate | [swiss](https://github.com/modelsphere/swiss) | Deploy control plane for the engine charts | | | [console](https://github.com/modelsphere/console) | Web console: users and roles, model deployment, playground | | Tune and measure | [llm-autotune](https://github.com/modelsphere/llm-autotune) | Searches serving configurations on idle GPUs |