Text Summarization on CNNDM (ROUGE-2, Rej., Speed)
15.7ROUGE-2 ScoreFaster Cascades
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Faster CascadesModel=T5 S→L, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 15.7 | 33 | 2.12 | |
| VIA-SDModel=T5 S→XL, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 13.05 | 24 | 2.4 | |
| VIA-SDModel=T5 S→XL, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 12.99 | 27 | 2.75 | |
| Faster CascadesModel=T5 S→XL, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 12.95 | 32 | 2.1 | |
| Faster CascadesModel=T5 S→XL, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 12.95 | 34 | 1.95 | |
| Speculative DecodingModel=T5 S→XL, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 12.9 | 38 | 1.7 | |
| BiLDModel=T5 S→XL, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 12.9 | 36 | 1.82 | |
| Cascade SDModel=T5 S→XL, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 12.9 | 35 | 1.85 | |
| SWiFTModel=T5 S→XL, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 12.9 | 34 | 1.95 | |
| CLaSPModel=T5 S→XL, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 12.9 | 33 | 2 | |
| VIA-SDModel=T5 S→L, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 12.81 | 26 | 2.1 | |
| Speculative DecodingModel=T5 S→XL, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 12.8 | 35 | 1.95 | |
| BiLDModel=T5 S→XL, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 12.8 | 34 | 2 | |
| Cascade SDModel=T5 S→XL, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 12.8 | 33 | 2.05 | |
| SWiFTModel=T5 S→XL, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 12.8 | 32 | 2.12 | |
| CLaSPModel=T5 S→XL, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 12.8 | 31 | 2.15 | |
| VIA-SDModel=T5 S→L, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 12.7 | 28 | 2.35 | |
| Faster CascadesModel=T5 S→L, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 12.63 | 34 | 1.88 | |
| Speculative DecodingModel=T5 S→L, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 11.2 | 36 | 1.79 | |
| BiLDModel=T5 S→L, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 11.2 | 34 | 1.85 | |
| Cascade SDModel=T5 S→L, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 11.2 | 33 | 1.9 | |
| SWiFTModel=T5 S→L, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 11.2 | 32 | 1.92 | |
| CLaSPModel=T5 S→L, Decoding Setting=Non-Greedy Sampling, Temperature=12026.06 | 11.2 | 31 | 1.95 | |
| Speculative DecodingModel=T5 S→L, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 11 | 34 | 2.05 | |
| BiLDModel=T5 S→L, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 11 | 33 | 2.1 | |
| Cascade SDModel=T5 S→L, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 11 | 32 | 2.15 | |
| SWiFTModel=T5 S→L, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 11 | 31 | 2.18 | |
| CLaSPModel=T5 S→L, Decoding Setting=Greedy Decoding, Temperature=02026.06 | 11 | 30 | 2.2 |