From cc1fab44a357781f6f065c260ed3d8858fd2443d Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Fri, 16 Jan 2026 22:16:39 +0000
Subject: [PATCH] fix: show error details in verbose benchmark output
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

When a sample errors during benchmark execution, the verbose output
now shows the actual error message, making debugging easier:

  [✗] Sample niah-0000: {'correct': 0.0, 'f1': 0.0}
      Error: <actual error message here>
---
 benchmarks/runner.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/benchmarks/runner.py b/benchmarks/runner.py
index a7b9700..e03a258 100644
--- a/benchmarks/runner.py
+++ b/benchmarks/runner.py
@@ -292,6 +292,8 @@ def _update_progress(
         if self.config.verbose:
             status = "✓" if sample_result.is_correct else "✗"
             print(f"  [{status}] Sample {sample_result.sample_id}: {sample_result.metrics}")
+            if sample_result.error:
+                print(f"      Error: {sample_result.error}")
 
     def _run_sequential(
         self,