List your company

Code Generation on BigCodeBench (Pass rate)

54Pass RateTarun (Llama-3.3-70B password-locked organism)
42.5645.5348.551.47Jun 28, 2026

Evaluation results

MethodLinks
Tarun (Llama-3.3-70B password-locked organism)Condition=Password (unlocked)2026.0654
Causal Perturbative Elicitation (CPE)Condition=Locked + CPE, Base Model=Tarun (Llama-3.3-70B password-locked organism)2026.0652
Tarun (Llama-3.3-70B password-locked organism)Condition=Locked (no password)2026.0643