huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-04-19 22:02:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fixing CI

Browse Source
This commit is contained in:
Nicolas Patry 2025-04-18 12:48:07 +02:00
parent b400c275e4
commit f17367e883
No known key found for this signature in database
GPG Key ID: 87B37D879D09DEB4
1 changed files with 13 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
router/src/sagemaker.rs
View File

@ -67,16 +67,26 @@ pub(crate) async fn sagemaker_compatibility(
default_return_full_text: Extension<bool>,
infer: Extension<Infer>,
compute_type: Extension<ComputeType>,
context: Extension<Option<opentelemetry::Context>>,
info: Extension<Info>,
Json(req): Json<SagemakerRequest>,
) -> Result<Response, (StatusCode, Json<ErrorResponse>)> {
match req {
SagemakerRequest::Generate(req) => {
compat_generate(default_return_full_text, infer, compute_type, Json(req)).await
compat_generate(
default_return_full_text,
infer,
compute_type,
context,
Json(req),
)
.await
}
SagemakerRequest::Chat(req) => {
chat_completions(infer, compute_type, info, context, Json(req)).await
}
SagemakerRequest::Chat(req) => chat_completions(infer, compute_type, info, Json(req)).await,
SagemakerRequest::Completion(req) => {
completions(infer, compute_type, info, Json(req)).await
completions(infer, compute_type, info, context, Json(req)).await
}
}
}