| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| Syntactic Control (Q ∝ pq) (test) | GPT2-large + SIS | Log Probability Q(y)0.0001 | 12 | 4d ago | |
| Syntactic Control (Q = p) (test) | Llama3-8B (5-shot) + SIS | Log Probability p(y)-22.71 | 12 | 4d ago | |
| RealToxicityPrompts 10K nontoxic prompts | DEXPERTS | Avg Max Toxicity30.2 | 9 | 4d ago | |
| Base Language Model Efficiency Comparison | PPLM | Speed Ratio270.11 | 8 | 4d ago | |
| SST-5 No-Pos | GENhance | Positiveness Score70 | 8 | 4d ago | |
| SST-5 200-Pos | GENhance | Positiveness Score91 | 8 | 4d ago | |
| Yelp Formality (test) | LATENTOPS | Accuracy97 | 4 | 4d ago | |
| Amazon Tense (test) | Accuracy97 | 4 | 4d ago | ||
| Single-Attribute Control prompts PPLM (test) | PriorControl | Average Score4.13 | 3 | 4d ago | |
| Single-Attribute Control | Sentiment Avg99.9 | 3 | 4d ago |
