rl-rag/

personalfinance-gpt-oss-120b

Deep research agent evaluation on data/drtulu_personalfinance.jsonl (normal split).

	Results




	

Metric Value

pass@4 0.0%

avg@4 0.0%

Trajectory accuracy 0.0% (0/18772)

Questions 4693

Trajectories 18772 (4 per question)

Avg tool calls 15.7

Full conversations ✅

	Model & Setup




	




	

Model gpt-oss-120b

Judge gpt-4o

Max tool calls 50

Temperature 0.7

Blocked domains None