Sign Language Translation on PHOENIX14T (test)
30.7BLEU-4ADAT
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ADATstrategy=HATL2026.02 | 30.7 | — | — | — | — | 68 | 48.4 | 35.6 | — | 52.8 | — | |
| ADATstrategy=HATL2026.02 | 29.4 | — | — | — | — | 59.5 | 48.2 | 35.4 | — | 54.2 | — | |
| CV-SLTSupervision level=Gloss-based2025.12 | 29.27 | — | — | — | — | 54.88 | 42.68 | 34.79 | — | 54.33 | — | |
| MSKASupervision level=Gloss-based2026.07 | 29.03 | — | 53.54 | — | — | 54.79 | 42.42 | 34.49 | — | — | — | |
| Two Stream-SLT2026.02 | 29 | — | — | — | — | 54.9 | 42.4 | 34.5 | — | 53.5 | — | |
| TS-SLTSupervision level=Gloss-based2026.01 | 28.95 | — | 53.48 | — | — | 54.9 | 42.43 | 34.46 | — | — | — | |
| TS-SLTSupervision level=Gloss-based2025.12 | 28.95 | — | — | — | — | 54.9 | 42.43 | 34.46 | — | 53.48 | — | |
| TS-SLTModality (Pose)=true, Modality (RGB)=true, Approach=Gloss-based2026.06 | 28.95 | — | 53.48 | — | — | 54.9 | — | — | — | — | — | |
| TS-SLTSupervision level=Gloss-based2026.07 | 28.95 | — | 53.48 | — | — | 54.9 | 42.43 | 34.46 | — | — | — | |
| SLTUNET2026.02 | 28.5 | — | — | — | — | 52.9 | 41.8 | 34 | — | 52.1 | — | |
| SLTUNETSupervision level=Gloss-based2026.01 | 28.47 | — | 52.11 | — | — | 52.92 | 41.76 | 33.99 | — | — | — | |
| SLTUNETModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-based2026.06 | 28.47 | — | 52.11 | — | — | 52.92 | — | — | — | — | — | |
| VL-Transfer2026.02 | 28.4 | — | — | — | — | 54 | 41.8 | 33.8 | — | 52.7 | — | |
| MMTLBSupervision level=Gloss-based2026.01 | 28.39 | — | 52.65 | — | — | 53.97 | 41.75 | 33.84 | — | — | — | |
| MMTLBModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-based2026.06 | 28.39 | — | 52.65 | — | — | 53.97 | — | — | — | — | — | |
| IP-SLTModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-based2026.06 | 27.97 | — | 53.72 | — | — | 54.25 | — | — | — | — | — | |
| SLT-GRPOSupervision level=Gloss-free2025.12 | 27.86 | — | — | — | — | 53.6 | 41.2 | 33.24 | — | 52.56 | — | |
| SIGNETModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-free2026.06 | 27.82 | — | 53.05 | — | — | 54.1 | — | — | — | — | — | |
| PGG-SLTModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-free2026.06 | 27.32 | — | 52.56 | — | — | 54.02 | — | — | — | — | — | |
| ViPo-MLLMSupervision level=Gloss-free2026.07 | 27.1 | — | 51.5 | — | — | 50.97 | 40 | 32.36 | — | — | — | |
| SLT-SFTSupervision level=Gloss-free2025.12 | 26.83 | — | — | — | — | 53.29 | 40.44 | 32.35 | — | 51.81 | — | |
| C2RLSupervision level=Gloss-free2025.12 | 26.75 | — | — | — | — | 52.81 | 40.2 | 32.2 | — | 50.96 | — | |
| Two Stream-SLT2026.02 | 26.7 | — | — | — | — | 52.1 | 39.8 | 32 | — | 51.6 | — | |
| VAPSupervision level=Weakly Supervised Gloss-free2026.07 | 26.62 | — | 51.47 | — | — | 52.78 | — | — | — | — | — | |
| VAPSupervision level=Weakly supervised gloss-free2026.01 | 26.16 | — | 51.28 | — | — | 53.07 | — | — | — | — | — | |
| VAPSupervision level=Gloss-free2025.12 | 26.16 | — | — | — | — | 53.07 | — | — | — | 51.28 | — | |
| EASLTSupervision level=Gloss-free2026.01 | 26.15 | — | 48.68 | — | — | 52.42 | 39.63 | 31.59 | — | — | — | |
| SLTUNET2026.02 | 26 | — | — | — | — | 50.4 | 39.2 | 31.4 | — | 50 | — | |
| MMSLTSupervision level=Gloss-free2026.01 | 25.73 | — | 47.97 | — | — | 48.92 | 38.12 | 30.79 | — | — | — | |
| MMSLTModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-free2026.06 | 25.73 | — | 47.97 | — | — | 48.92 | — | — | — | — | — | |
| Transformerstrategy=HATL2026.02 | 25.7 | — | — | — | — | 67.7 | 48.2 | 35.5 | — | 50.5 | — | |
| MMSLTSupervision level=Gloss-free2026.07 | 25.7 | — | 48 | — | — | 48.9 | 38.1 | 30.8 | — | — | — | |
| BeyondGlossSupervision level=Gloss-free2026.01 | 25.49 | — | 52.89 | — | — | 52.38 | 38.57 | 30.74 | — | — | — | |
| BeyondGlossModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-free2026.06 | 25.49 | — | 52.89 | — | — | 52.38 | — | — | — | — | — | |
| STMC-Transformer2026.02 | 25.4 | — | — | — | — | 50.6 | 38.4 | 30.6 | — | 48.8 | — | |
| CMDA-SLTSupervision level=Weakly Supervised Gloss-free2026.07 | 25.36 | — | 49.87 | — | — | — | — | — | — | — | — | |
| VL-Transfer2026.02 | 24.6 | — | — | — | — | 49.9 | 37.3 | 29.7 | — | 49.6 | — | |
| BN-TIN-Transf+SignBTSupervision level=Gloss-based2026.01 | 24.32 | — | 49.54 | — | — | 50.8 | 37.75 | 29.72 | — | — | — | |
| SpaMo (Baseline)Supervision level=Gloss-free2026.01 | 24.32 | — | 46.57 | — | — | 49.8 | 37.32 | 29.5 | — | — | — | |
| SignBTSupervision level=Gloss-based2025.12 | 24.32 | — | — | — | — | 50.8 | 37.75 | 29.72 | — | 49.54 | — | |
| SpaMoSupervision level=Gloss-free2026.07 | 24.3 | — | 46.6 | — | — | 49.8 | 37.3 | 29.5 | — | — | — | |
| Ours w TxtCTCGloss-free=true, Type=one-to-one SLT2025.05 | 24.23 | — | 50.6 | — | — | — | — | — | — | — | — | |
| MLSLTSupervision level=Gloss-free2026.01 | 24.23 | — | 50.6 | — | — | — | — | — | — | — | — | |
| LLM-based SLT (Vid+PG)Input Modality=Video + Pseudo-glosses, Pseudo-gloss source=SlowFastSign [1]2025.01 | 23.8 | — | 46.11 | 227.05 | 38.49 | — | — | — | — | — | 52.8 | |
| STMC-Transformer2026.02 | 23.8 | — | — | — | — | 47.5 | 35.9 | 28.6 | — | 47.3 | — | |
| STM-Net2026.02 | 23.7 | — | — | — | — | 47 | 36.1 | 28.7 | — | 46.7 | — | |
| STN-SLTSupervision level=Gloss-based2026.07 | 23.65 | — | — | — | — | 48.61 | 35.97 | 28.37 | — | — | — | |
| STMC-TSupervision level=Gloss-based2026.07 | 23.65 | — | 46.65 | — | — | 46.98 | 36.09 | 28.7 | — | — | — | |
| LLaVA-SLTSupervision level=Gloss-free2025.12 | 23.42 | — | — | — | — | 51.2 | 37.51 | 29.39 | — | 50.44 | — | |
| SignLLMGloss-free=true, Type=one-to-one SLT2025.05 | 23.4 | — | 44.49 | — | — | — | — | — | — | — | — | |
| SignLLMSupervision level=Gloss-free2026.01 | 23.4 | — | 44.49 | — | — | 45.21 | 34.78 | 28.05 | — | — | — | |
| Sign-llmSupervision level=Gloss-free2025.12 | 23.4 | — | — | — | — | 45.21 | 34.78 | 28.05 | — | 44.49 | — | |
| SignLLMModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-free2026.06 | 23.4 | — | 44.49 | — | — | 45.21 | — | — | — | — | — | |
| SignLLMSupervision level=Gloss-free2026.07 | 23.4 | — | 44.49 | — | — | 45.21 | 34.78 | 28.05 | — | — | — | |
| SimulSLT2026.02 | 23.1 | — | — | — | — | 48.2 | 35.6 | 28 | — | 49.2 | — | |
| Fla-LLMGloss-free=true, Type=one-to-one SLT2025.05 | 23.09 | — | 45.27 | — | — | — | — | — | — | — | — | |
| FLa-LLMSupervision level=Gloss-free2026.01 | 23.09 | — | 45.27 | — | — | 46.29 | 35.33 | 28.03 | — | — | — | |
| FLa-LLMSupervision level=Gloss-free2025.12 | 23.09 | — | — | — | — | 46.29 | 35.33 | 28.03 | — | 45.27 | — | |
| FLa-LLMModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-free2026.06 | 23.09 | — | 45.27 | — | — | 46.29 | — | — | — | — | — | |
| FLa-LLMSupervision level=Gloss-free2026.07 | 23.09 | — | 45.27 | — | — | 46.29 | 35.33 | 28.03 | — | — | — | |
| Sign2GPTGloss-free=true, Type=one-to-one SLT2025.05 | 22.52 | — | 48.9 | — | — | — | — | — | — | — | — | |
| BaselineGloss-free=true, Type=one-to-one SLT2025.05 | 22.52 | — | 49.85 | — | — | — | — | — | — | — | — | |
| Sign2GPTSupervision level=Gloss-free2026.01 | 22.52 | — | 48.9 | — | — | 49.54 | 35.96 | 28.83 | — | — | — | |
| Sign2GPTSupervision level=Gloss-free2025.12 | 22.52 | — | — | — | — | 49.54 | 35.96 | 28.83 | — | 48.9 | — | |
| Sign2GPT (w/PGP)Pseudo-gloss pre-training (PGP)=true2025.01 | 22.52 | — | 48.9 | — | — | — | — | — | — | — | — | |
| Sign2GPTModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-free2026.06 | 22.52 | — | 48.9 | — | — | 49.54 | — | — | — | — | — | |
| Sign2GPTSupervision level=Gloss-free2026.07 | 22.52 | — | 48.9 | — | — | 49.54 | 35.96 | 28.83 | — | — | — | |
| HST-GNN2026.02 | 22.3 | — | — | — | — | 45.2 | 34.7 | 27.1 | — | — | — | |
| ConSLTSupervision level=Weakly supervised gloss-free2026.01 | 21.59 | — | 47.69 | — | — | — | — | — | — | — | — | |
| GFSLT-VLPGloss-free=true, Type=one-to-one SLT2025.05 | 21.44 | — | 42.49 | — | — | — | — | — | — | — | — | |
| GFSLT-VLPSupervision level=Gloss-free2026.01 | 21.44 | — | 42.49 | — | — | 43.71 | 33.18 | 26.11 | — | — | — | |
| GFSLT-VLPSupervision level=Gloss-free2025.12 | 21.44 | — | — | — | — | 43.71 | 33.18 | 26.11 | — | 42.49 | — | |
| GFSLT2025.01 | 21.44 | — | 42.49 | — | — | — | — | — | — | — | — | |
| GFSLT-VLPModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-free2026.06 | 21.44 | — | 42.49 | — | — | 43.71 | — | — | — | — | — | |
| GFSLT-VLPSupervision level=Gloss-free2026.07 | 21.44 | — | 42.49 | — | — | 43.71 | 33.18 | 26.11 | — | — | — | |
| SLRTSupervision level=Gloss-based2026.01 | 21.32 | — | — | — | — | 46.61 | 33.73 | 26.19 | — | — | — | |
| SLRTSupervision level=Gloss-based2025.12 | 21.32 | — | — | — | — | 46.61 | 33.73 | 26.19 | — | — | — | |
| SLRTSupervision level=Gloss-based2026.07 | 21.32 | — | — | — | — | 46.61 | 33.73 | 26.19 | — | — | — | |
| Joint-SLRT2026.02 | 21.3 | — | — | — | — | 46.6 | 33.7 | 26.2 | — | — | — | |
| ConSLTSupervision level=Weakly Supervised Gloss-free2026.07 | 21.11 | — | 47.74 | — | — | — | — | — | — | — | — | |
| LLM-based SLT (Vid)Input Modality=Video only2025.01 | 20.58 | — | 41.2 | 190.4 | 34.05 | — | — | — | — | — | 52.25 | |
| ADATstrategy=Full Fine-tuning2026.02 | 20.2 | — | — | — | — | 41 | 30.6 | 23.6 | — | 43.5 | — | |
| Joint-SLRT2026.02 | 20.2 | — | — | — | — | 45.3 | 32.3 | 24.8 | — | — | — | |
| Multi-View SLT2026.02 | 20.1 | — | — | — | — | 46.1 | 32.8 | 25 | — | — | — | |
| Transformerstrategy=Full Fine-tuning2026.02 | 19.7 | — | — | — | — | 42 | 31 | 23.8 | — | 43.2 | — | |
| Sign2GPT2025.01 | 19.42 | — | 45.23 | — | — | — | — | — | — | — | — | |
| Transformerstrategy=Classical Fine-tuning2026.02 | 18.8 | — | — | — | — | 41.3 | 30.3 | 23.3 | — | 41.7 | — | |
| Transformerstrategy=HATL2026.02 | 18.6 | — | — | — | — | 45.1 | 31.5 | 22.5 | — | 42.6 | — | |
| ADATstrategy=Full Fine-tuning2026.02 | 18.6 | — | — | — | — | 37 | 28 | 22.4 | — | 42.8 | — | |
| ADATstrategy=Classical Fine-tuning2026.02 | 18.5 | — | — | — | — | 39.7 | 29.3 | 22.8 | — | 40.6 | — | |
| NSLT2026.02 | 18.1 | — | — | — | — | 43.3 | 30.4 | 22.8 | — | 43.8 | — | |
| Transformerstrategy=Full Fine-tuning2026.02 | 17.5 | — | — | — | — | 41.2 | 31.2 | 23.4 | — | 37.1 | — | |
| MC-SLT2026.02 | 17 | — | — | — | — | 43.7 | — | — | — | 43.4 | — | |
| ADATstrategy=Classical Fine-tuning2026.02 | 17 | — | — | — | — | 36 | 27.6 | 21.8 | — | 41.1 | — | |
| Transformerstrategy=Classical Fine-tuning2026.02 | 16.4 | — | — | — | — | 41 | 29.9 | 23.1 | — | 36.8 | — | |
| GASLTSupervision level=Weakly supervised gloss-free2026.01 | 15.74 | — | 39.86 | — | — | 39.07 | 26.74 | 21.86 | — | — | — | |
| GASLTSupervision level=Gloss-free2025.12 | 15.74 | — | — | — | — | 39.07 | 26.74 | 21.86 | — | 39.86 | — | |
| CSGCRGloss-free=true, Type=one-to-one SLT2025.05 | 15.18 | — | 38.85 | — | — | — | — | — | — | — | — | |
| CSGCRSupervision level=Gloss-free2025.12 | 15.18 | — | — | — | — | 36.71 | 25.4 | 18.86 | — | 38.85 | — | |
| CSGCRSupervision level=Gloss-free2026.07 | 15.18 | — | 38.85 | — | — | 36.71 | 25.4 | 18.86 | — | — | — |