Our new X account is live! Follow @wizwand_team for updates

LLM-as-a-judge evaluation on Vicuna Bench

0.605Pearson Correlation (r)

TRACT

Updated 4d ago

Evaluation Results

Method	Links
TRACT 2025.03		0.605	0.65
TRACT 2025.03		0.593	0.552
Mistral-7B-Instruct (RAFT on GPT-4) 2025.03		0.567	0.519
TRACT (Ablation: Objective CE) 2025.03		0.562	0.526
TRACT (Ablation: Data GPT-4) 2025.03		0.528	0.513
Baseline B.4 (RAFT) 2025.03		0.509	0.541
TRACT (Ablation: Stage 1 Init) 2025.03		0.505	0.477
Prometheus-2-7B 2025.03		0.488	0.48
Baseline B.3 (RAIL) 2025.03		0.485	0.487
Baseline B.2 (CE) 2025.03		0.467	0.483
Mistral-7B-Instruct (CE on GPT-4 CoT) 2025.03		0.463	0.456
Mistral-7B-Instruct (CE on GPT-4 scores) 2025.03		0.429	0.414
Mistral-7B-Instruct (CE on Self-gen CoT) 2025.03		0.418	0.404
Baseline B.1 (CE) 2025.03		0.4	0.423
Mistral-7B-Instruct (Zero-shot RAIL) 2025.03		0.281	0.165
CLoud (Reward model) 2025.03		0.229	0.311