FlowerTune LLM — Finance flowertune-finance Leaderboard
FlowerTune Finance track — federated fine-tuning of LLMs on financial instruction data. Avg Score aggregates accuracy over FPB, FIQA-SA, and TFNS financial sentiment/QA benchmarks. Tests federated adaptation where client data (financial text) is privacy-sensitive. · Metric: Avg Score (higher is better)
| # | Model | Avg Score | Paper |
|---|---|---|---|
| 1 | ZeroOne.AI — Qwen2.5-7B | 84.58 | — |
| 2 | FL-finetune-JB-DC — Qwen2.5-7B | 75.51 | — |
| 3 | Massimo R. Scamarcia — SmolLM2-1.7B-Instruct | 67.90 | — |
| 4 | Nut Chukamphaeng — Qwen2.5-7B | 65.52 | — |
| 5 | Md Mahinur Alam — Qwen2.5-7B | 61.53 | — |
| 6 | ZJUDAI — Mistral-7B-Instruct-v0.3 | 61.45 | — |
| 7 | Baseline — Llama-3.2-3B | 54.78 | — |
| 8 | T-IoI@UNR — SmolLM2-360M-Instruct | 52.22 | — |
| 9 | Massimo R. Scamarcia — SmolLM2-360M-Instruct | 45.36 | — |