Financial Reasoning on FinQA

77.1Accuracy

QianfanHuijin-70B

Updated 4d ago

Evaluation Results

Method
QianfanHuijin-70B 2025.12	77.1	86.2	-	-
DeepSeek-V3.2 2025.12	72.2	83.82	-	-
QianfanHuijin-8B 2025.12	68.3	81.47	-	-
GenICL 2026.02	67.3	-	-	-
DICL 2026.02	67.1	-	-	-
theory-guided context selection strategy 2026.02	66.7	-	-	-
Qwen3-8B 2025.12	65.6	75.88	-	-
DeepSeek-R1 2025.12	65.5	82.56	-	-
BM25 2026.02	64.2	-	-	-
TopicK 2026.02	64.2	-	-	-
Qwen3-235B 2025.12	63.5	82.61	-	-
Zero 2026.02	60.1	-	-	-
DianJin-R1-7B 2025.12	56.7	71.85	-	-
GenICL 2026.02	51.6	-	-	-
DICL 2026.02	51.3	-	-	-
theory-guided context selection strategy 2026.02	51.1	-	-	-
TopicK 2026.02	49.8	-	-	-
BM25 2026.02	49.2	-	-	-
Zero 2026.02	48.6	-	-	-
BERT Base Internal Retriever + RoBERTa Large Encoder Generator (Baseline) 2025.12	-	-	57.87	60.02
BERT Base Internal Retriever + DPR-FAISS External Retriever + RoBERTa Large Encoder Generator 2025.12	-	-	58.49	60.96
SecBERT Internal Retriever + RoBERTa Large Encoder Generator 2025.12	-	-	59.7	62.11
SecBERT Internal Retriever + DPR-FAISS External Retriever + RoBERTa Large Encoder Generator 2025.12	-	-	60.54	63.48
SecBERT Internal Retriever + RoBERTa Base Encoder Generator 2025.12	-	-	56.49	58.33
SecBERT Internal Retriever + DPR-FAISS External Retriever + RoBERTa Base Encoder Generator 2025.12	-	-	56.75	58.81
SecBERT Internal Retriever + BERT Large Encoder Generator 2025.12	-	-	53.95	55.76
SecBERT Internal Retriever + DPR-FAISS External Retriever + BERT Large Encoder Generator 2025.12	-	-	54.64	56.97
SecBERT Internal Retriever + FinBERT Encoder Generator 2025.12	-	-	49.53	52.08
SecBERT Internal Retriever + DPR-FAISS External Retriever + FinBERT Encoder Generator 2025.12	-	-	48.77	50.9