Financial Domain Reasoning on Finance Benchmarks ConvFinQA, OpenFinData
56.77ConvFinQA AccuracyFinance (WiroAI-Finance-Llama-8B)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Finance (WiroAI-Finance-Llama-8B)Model Family=Llama3.1-8B2026.01 | 56.77 | 42.08 | 38.4 | |
| ReasonAnyMerging=ReasonAny, Model Family=Llama3.1-8B2026.01 | 50.37 | 62.3 | 54.62 | |
| Task ArithmeticMerging=Task Arithmetic, Model Family=Llama3.1-8B2026.01 | 47.85 | 51.41 | 46.53 | |
| LEDMerging=LED, Model Family=Llama3.1-8B2026.01 | 47.17 | 52.23 | 46.2 | |
| LinearMerging=Linear, Model Family=Llama3.1-8B2026.01 | 44.75 | 61.88 | 49.08 | |
| FuseLLMMerging=FuseLLM, Model Family=Llama3.1-8B2026.01 | 42.33 | 46.92 | 38.06 | |
| DAREMerging=DARE, Model Family=Llama3.1-8B2026.01 | 39.97 | 63.33 | 45.91 | |
| TIESMerging=TIES, Model Family=Llama3.1-8B2026.01 | 36.14 | 58.3 | 47.21 | |
| Reasoning (DeepSeek-R1-Distill-Llama-8B)Model Family=Llama3.1-8B2026.01 | 28.13 | 60.72 | 57.62 |