StyleRec - 1-Shot - Token F1: leaderboard
StyleRec reconstructs a hidden style-rewriting prompt from original and rewritten English text, using 1 direct in-context examples and no fine-tuning. Published Token F1 score (source scale x100) for Mistral 7B Instruct v0.3 and Llama 3 8B Instruct on the 1,019-item test split of a 33-style dataset. Similarity and template overlap can reward an incorrect style label; this is not exact prompt recovery. Protocols and metrics remain separate.
Metric: Token F1 (points; source scale x100, higher is better). Source: arxiv.org. Saturation forecast: Estimated already saturated. 2 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3 8B Instruct | 79.64 | #1123 |
| 2 | Mistral 7B Instruct (v0.3) | 31.88 | #1277 |
No result here: #5 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=stylerec-1-shot-token-f1 · How It Works · Data refreshed daily, snapshot 2026-10-25.