flsol demo: fix top_p/claude/gemini, force coordinates, add reflection comments, screenshot mode

2026-03-22 13:57:33 +08:00
parent 093b779045
commit f32e5f9e64
8 changed files with 881 additions and 50 deletions
--- a/evaluation_examples/test_flsol.json
+++ b/evaluation_examples/test_flsol.json
@@ -1,5 +1,5 @@
 {
  "flsol": [
-    "flsol_task4_measure"
+    "flsol_taskE_auto_optimize_scan"
  ]
 }