Style Alignment on Charles Dickens (test)
0.807Mean Style ScoreFT-Model
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| FT-ModelModel=GRPO fine-tuned 8B story generator2025.12 | 0.807 | 0.148 | |
| FT-AgenticEvaluation Protocol=Few-shot calibrated, Number of Parameters=8B2025.12 | 0.798 | — | |
| GPT-oss-120BEvaluation Protocol=Few-shot calibrated, Number of Parameters=120B2025.12 | 0.413 | — | |
| GPT-oss-120BModel=Reference model, Evaluation protocol=Zero-shot2025.12 | 0.37 | 0.126 | |
| Gemma3-27BModel=Open-weight baseline, Evaluation protocol=Zero-shot2025.12 | 0.298 | 0.099 | |
| Llama3.3-70BEvaluation Protocol=Few-shot calibrated, Number of Parameters=70B2025.12 | 0.297 | — | |
| Qwen2.5-32BModel=Open-weight baseline, Evaluation protocol=Zero-shot2025.12 | 0.278 | 0.092 | |
| Llama3-70BModel=Open-weight baseline, Evaluation protocol=Zero-shot2025.12 | 0.271 | 0.088 | |
| Qwen2.5-32BEvaluation Protocol=Few-shot calibrated, Number of Parameters=32B2025.12 | 0.262 | — | |
| Gemma3-27BEvaluation Protocol=Few-shot calibrated, Number of Parameters=27B2025.12 | 0.244 | — |