Narrow finetuning on Finance
76.9EM-F1TReFT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TReFTModel=Llama-3.1-8B2026.06 | 76.9 | 0.3 | |
| Data interleavingModel=Llama-3.1-8B2026.06 | 71.9 | 0.1 | |
| TReFTModel=Qwen-2.5-7B2026.06 | 68.8 | 1.6 | |
| TReFTModel=Qwen-2.5-32B2026.06 | 68.1 | -0.4 | |
| TReFTModel=GPT-OSS-20B2026.06 | 62.6 | -2.3 | |
| SFTModel=Qwen-2.5-32B2026.06 | 55.6 | -2.4 | |
| SFTModel=Llama-3.1-8B2026.06 | 53.4 | -0.3 | |
| SFTModel=Qwen-2.5-7B2026.06 | 50.1 | -0.5 | |
| Data interleavingModel=GPT-OSS-20B2026.06 | 46.2 | -2.9 | |
| SFTModel=GPT-OSS-20B2026.06 | 45.4 | -4.5 | |
| Data interleavingModel=Qwen-2.5-7B2026.06 | 42.5 | 1.1 | |
| Data interleavingModel=Qwen-2.5-32B2026.06 | 28 | -0.8 |