List your company

Hallucination Mitigation on FinLLM-Eval

92.5Faithfulness ScoreBALTO
32.28447.91763.5579.183Jun 14, 2026

Evaluation results

MethodLinks
BALTOModel=Qwen3-8B2026.0692.557.953.6—
GRPODModel=Qwen3-8B2026.0691.751.146.937.2
GRPOBModel=Qwen3-8B2026.0690.239.135.344
BALTOModel=Qwen3-4B2026.0688.742.137.3—
GRPOBModel=Qwen3-4B2026.068223.319.138.8
GRPODModel=Qwen3-4B2026.0678.243.634.157.7
FSPOModel=Qwen3-8B2026.0659.48248.746.6
FSPOModel=Qwen3-4B2026.0659.48248.746.6
BaseModel=Qwen3-4B2026.0657.1—57.1—
SFTModel=Qwen3-8B2026.0648.170.734—
DPOModel=Qwen3-8B2026.0641.460.224.9—
BaseModel=Qwen3-8B2026.0634.6—34.6—