Transfer Learning on SentEval Transfer tasks (test)
91.88MRSUBJPrompt-LLAMA2
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| SUBJPrompt-LLAMA2Params=7B, Training Mode=Without fine-tuning, Backbone=LLAMA22024.02 | 91.88 | 93.17 | 96.96 | 91.09 | — | — | 96 | 76.41 | 91.6 | 95.66 | |
| MRPrompt-LLAMA2Params=7B, Training Mode=Without fine-tuning, Backbone=LLAMA22024.02 | 91.82 | 92.88 | 97.07 | 91.6 | — | — | 95.8 | 74.61 | 91.47 | 96.54 | |
| SSTPrompt-LLAMA2Params=7B, Training Mode=Without fine-tuning, Backbone=LLAMA22024.02 | 91.82 | 92.88 | 97.07 | 91.6 | — | — | 95.8 | 74.61 | 91.47 | 96.54 | |
| CRPrompt-LLAMA2Params=7B, Training Mode=Without fine-tuning, Backbone=LLAMA22024.02 | 91.17 | 93.27 | 96.62 | 91.75 | — | — | 95.8 | 73.22 | 91.2 | 96.6 | |
| MPQAPrompt-LLAMA2Params=7B, Training Mode=Without fine-tuning, Backbone=LLAMA22024.02 | 91.1 | 93.04 | 96.3 | 91.82 | — | — | 96 | 75.42 | 91.34 | 95.72 | |
| MetaEOL-LLAMA2Params=7B, Training Mode=Without fine-tuning, Backbone=LLAMA22024.02 | 90.93 | 93.51 | 96.12 | 91.95 | — | — | 97.6 | 76.81 | 91.81 | 95.77 | |
| ST5-EncParams=4.8B, Training Mode=Fine-tuning on supervised datasets2024.02 | 90.83 | 94.44 | 96.33 | 91.68 | — | — | 95.4 | 77.91 | 91.63 | 94.84 | |
| PromptEOL-LLAMA2Params=7B, Training Mode=Without fine-tuning, Backbone=LLAMA22024.02 | 90.63 | 92.87 | 96.32 | 91.19 | — | — | 95.4 | 75.19 | 90.94 | 95 | |
| MRPCPrompt-LLAMA2Params=7B, Training Mode=Without fine-tuning, Backbone=LLAMA22024.02 | 90.33 | 93.32 | 96.36 | 91.45 | — | — | 96 | 75.13 | 91.04 | 94.67 | |
| TRECPrompt-LLAMA2Params=7B, Training Mode=Without fine-tuning, Backbone=LLAMA22024.02 | 88.97 | 92.19 | 96.23 | 91.45 | — | — | 96.8 | 74.72 | 90.65 | 94.18 | |
| DCLPre-train dataset=TweetEmoji-EN2022.03 | 86.83 | 90.36 | 92.61 | 87.51 | 91.43 | 50.77 | 75.8 | 71.54 | 80.86 | — | |
| SLPPre-train dataset=TweetEmoji-EN2022.03 | 86.79 | 89.69 | 92.24 | 87.93 | 92.53 | 52.31 | 78.6 | 74.61 | 81.84 | — | |
| InfoDCL-RoBERTaPre-train dataset=TweetEmoji-EN2022.03 | 86.66 | 89.75 | 93.71 | 87.12 | 92.59 | 53.08 | 83 | 73.22 | 82.39 | — | |
| InfoDCL-BERTweetPre-train dataset=TweetEmoji-EN2022.03 | 86.12 | 89.62 | 93.53 | 86.21 | 91.1 | 52.13 | 83.4 | 72 | 81.76 | — | |
| SimCSE-DistantPre-train dataset=TweetEmoji-EN2022.03 | 86.04 | 89.48 | 91.24 | 87.18 | 91.87 | 48.6 | 74.6 | 71.59 | 80.08 | — | |
| SimCSE-RoBERTaParams=123M, Training Mode=Fine-tuning on supervised datasets, Backbone=RoBERTa2024.02 | 84.92 | 92 | 94.11 | 89.82 | — | — | 88.8 | 75.65 | 88.08 | 91.27 | |
| E-MLMPre-train dataset=TweetEmoji-EN2022.03 | 80.62 | 84.79 | 91.01 | 78.54 | 88.14 | 46.65 | 72.2 | 74.09 | 77.01 | — | |
| BERTweetPre-train dataset=TweetEmoji-EN2022.03 | 79.58 | 80.82 | 93.03 | 71.78 | 86.66 | 43.71 | 80.8 | 72.35 | 76.09 | — | |
| WCLPre-train dataset=TweetEmoji-EN2022.03 | 78.96 | 83.74 | 92.91 | 85.3 | 85.28 | 43.48 | 84.2 | 71.88 | 78.22 | — | |
| MLMPre-train dataset=TweetEmoji-EN2022.03 | 76.85 | 77.35 | 90.63 | 80.4 | 85.5 | 41.81 | 73.2 | 73.04 | 74.85 | — | |
| Mir-BPre-train dataset=TweetEmoji-EN2022.03 | 76.72 | 81.48 | 91.57 | 85.39 | 81.05 | 44.48 | 87 | 74.67 | 77.8 | — | |
| RoBERTaPre-train dataset=TweetEmoji-EN2022.03 | 75.92 | 69.59 | 91.5 | 73.75 | 82.81 | 38.46 | 61.4 | 71.42 | 70.61 | — | |
| Sim-SPre-train dataset=TweetEmoji-EN2022.03 | 73.77 | 80.19 | 90.29 | 83.92 | 78.69 | 41.45 | 86 | 74.49 | 76.1 | — |