erhan/abap-llm
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Trajectory runs: tool-call budget 100 for tasks with a CDS contract object (eval keeps 60)

Browse Source 
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
This commit is contained in:
Kral
2026-10-05 14:08:33 +02:00
parent 13360292dd
commit c062df2ec5
4 changed files with 12 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
harness/record.py
View File

@@ -56,7 +56,7 @@ def write_record(run_dir, task, agent, rep, pool):
"cost_ledger_usd": cost, "cost_usage_usd": round(cost / LEDGER_TO_USAGE, 5),
"end_reason": rep.get("end_reason"), "tool_calls": rep.get("tool_calls"),
"activations": rep.get("activations"), "activation_failures": rep.get("activation_failures"),
"max_fail_streak": rep.get("max_fail_streak"), "syntax_hints": rep.get("syntax_hints"),
"max_fail_streak": rep.get("max_fail_streak"), "tool_budget": rep.get("tool_budget"), "syntax_hints": rep.get("syntax_hints"),
"adt_fallbacks": bool(rep.get("adt_fallbacks")), "agent_seconds": rep.get("agent_seconds"),
"turn_usage": getattr(agent, "turn_usage", []),
"setup_failed": rep.get("setup_failed", False),