From 4512d2b76dd90c1387e4219eccee160ed47a1192 Mon Sep 17 00:00:00 2001
From: frob <rick+github@frob.com.au>
Date: Fri, 11 Sep 2026 00:46:58 +0200
Subject: [PATCH] llm: raise token repeat limit to 100 and return error instead
 of silently closing (#18374)

---
 llm/llama_server.go | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/llm/llama_server.go b/llm/llama_server.go
index fc3edfbbe..0f4019799 100644
--- a/llm/llama_server.go
+++ b/llm/llama_server.go
@@ -1700,9 +1700,9 @@ func (s *llamaServerRunner) Completion(ctx context.Context, req CompletionReques
 				lastToken = strings.TrimSpace(lsResp.Content)
 				tokenRepeat = 0
 			}
-			if tokenRepeat > 30 {
+			if tokenRepeat > 100 {
 				slog.Debug("prediction aborted, token repeat limit reached")
-				return ctx.Err()
+				return fmt.Errorf("prediction aborted, token repeat limit reached")
 			}
 
 			if lsResp.Content != "" && !lsResp.Stop {