fix: Remove vec wrapping from OpenAI-compatible response (#273)

predibase · Feb 23, 2024 · 48c29e1 · 48c29e1
1 parent eb69759
commit 48c29e1
Show file tree

Hide file tree

Showing 3 changed files with 5 additions and 8 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/docs/reference/openai_api.md b/docs/reference/openai_api.md
@@ -26,7 +26,7 @@ resp = client.chat.completions.create(
     ],
     max_tokens=100,
 )
-print("Response:", resp[0].choices[0].message.content)
+print("Response:", resp.choices[0].message.content)
 ```
 
 ### Streaming
@@ -111,7 +111,7 @@ completion = client.completions.create(
     model=adapter_id,
     prompt=prompt,
 )
-print("Completion result:", completion[0].choices[0].text)
+print("Completion result:", completion.choices[0].text)
 
 # streaming completions
 completion_stream = client.completions.create(

diff --git a/router/src/server.rs b/router/src/server.rs
@@ -141,7 +141,7 @@ async fn completions_v1(
     } else {
         let (headers, generation) = generate(infer, Json(gen_req.into())).await?;
         // wrap generation inside a Vec to match api-inference
-        Ok((headers, Json(vec![CompletionResponse::from(generation.0)])).into_response())
+        Ok((headers, Json(CompletionResponse::from(generation.0))).into_response())
     }
 }
 
@@ -201,11 +201,7 @@ async fn chat_completions_v1(
     } else {
         let (headers, generation) = generate(infer, Json(gen_req.into())).await?;
         // wrap generation inside a Vec to match api-inference
-        Ok((
-            headers,
-            Json(vec![ChatCompletionResponse::from(generation.0)]),
-        )
-            .into_response())
+        Ok((headers, Json(ChatCompletionResponse::from(generation.0))).into_response())
     }
 }