API: Fix usage reporting for chat completions

kingbri1 · kingbri1 · commit 0e9385e02392 · 2024-05-17T00:03:15.000-04:00
Resolves #106 Signed-off-by: kingbri <bdashore3@proton.me>
diff --git a/endpoints/OAI/utils/chat_completion.py b/endpoints/OAI/utils/chat_completion.py
@@ -67,7 +67,7 @@ def _create_response(generation: dict, model_name: Optional[str]):
     )
 
     prompt_tokens = unwrap(generation.get("prompt_tokens"), 0)
-    completion_tokens = unwrap(generation.get("completion_tokens"), 0)
+    completion_tokens = unwrap(generation.get("generated_tokens"), 0)
 
     response = ChatCompletionResponse(
         choices=[choice],

Original file line number	Diff line number	Diff line change
`@@ -67,7 +67,7 @@ def _create_response(generation: dict, model_name: Optional[str]):`
`67`	`67`	`)`
`68`	`68`
`69`	`69`	`prompt_tokens = unwrap(generation.get("prompt_tokens"), 0)`
`70`		`- completion_tokens = unwrap(generation.get("completion_tokens"), 0)`
	`70`	`+ completion_tokens = unwrap(generation.get("generated_tokens"), 0)`
`71`	`71`
`72`	`72`	`response = ChatCompletionResponse(`
`73`	`73`	`choices=[choice],`