Natural Language Inference On Rte
Metrics
Accuracy
Results
Performance results of various models on this benchmark
Comparison Table
| Model Name | Accuracy |
|---|---|
| not-all-layers-are-equally-as-important-every | 53.7 |
| distilbert-a-distilled-version-of-bert | 62.9% |
| guess-the-instruction-making-language-models | 71.05 |
| ernie-enhanced-language-representation-with | 68.8% |
| smart-robust-and-efficient-fine-tuning-for | 92.5% |
| albert-a-lite-bert-for-self-supervised | 89.2% |
| bloomberggpt-a-large-language-model-for | 53.8% |
| palm-scaling-language-modeling-with-pathways-1 | 78.7% |
| knowledge-in-context-towards-knowledgeable | 74.00 |
| data2vec-a-general-framework-for-self-1 | 69.9% |
| lamini-lm-a-diverse-herd-of-distilled-models | 65% |
| palm-scaling-language-modeling-with-pathways-1 | 72.9% |
| q-bert-hessian-based-ultra-low-precision | 84.7 |
| palm-2-technical-report-1 | 79.3% |
| exploring-the-benefits-of-training-expert | 64.01 |
| smart-robust-and-efficient-fine-tuning-for | 71.2% |
| informer-transformer-likes-informed-attention | 73.7% |
| palm-2-technical-report-1 | 78.7% |
| xlnet-generalized-autoregressive-pretraining | 85.9% |
| big-bird-transformers-for-longer-sequences | 75.0% |
| hungry-hungry-hippos-towards-language | 53.1% |
| unifying-language-learning-paradigms | 60.7% |
| opt-iml-scaling-language-model-instruction | 60.3% |
| structbert-incorporating-language-structures | 88.7% |
| not-all-layers-are-equally-as-important-every | 55.4 |
| ask-me-anything-a-simple-strategy-for | 75.1% |
| lamini-lm-a-diverse-herd-of-distilled-models | 52.3% |
| bloomberggpt-a-large-language-model-for | 69.3% |
| q8bert-quantized-8bit-bert | 84.8 |
| entailment-as-few-shot-learner | 87.2% |
| hungry-hungry-hippos-towards-language | 58.1% |
| exploring-the-limits-of-transfer-learning | 87.2% |
| designing-effective-sparse-expert-models | 93.5% |
| smart-robust-and-efficient-fine-tuning-for | 92.0% |
| llm-int8-8-bit-matrix-multiplication-for | 85.4% |
| 190910351 | 62.9% |
| the-cot-collection-improving-zero-shot-and | 80.8% |
| bert-pre-training-of-deep-bidirectional | 70.1% |
| exploring-the-limits-of-transfer-learning | 80.1% |
| lamini-lm-a-diverse-herd-of-distilled-models | 57% |
| n-grammer-augmenting-transformers-with-latent-1 | 59.2% |
| ask-me-anything-a-simple-strategy-for | 58.8% |
| unifying-language-learning-paradigms | 92.1% |
| hungry-hungry-hippos-towards-language | 58.1% |
| alexatm-20b-few-shot-learning-using-a-large | 68.6% |
| opt-iml-scaling-language-model-instruction | 54.2% |
| palm-scaling-language-modeling-with-pathways-1 | 79.6% |
| lamini-lm-a-diverse-herd-of-distilled-models | 87.4% |
| lamini-lm-a-diverse-herd-of-distilled-models | 67.9% |
| palm-scaling-language-modeling-with-pathways-1 | 95.7% |
| palm-2-technical-report-1 | 81.9% |
| smart-robust-and-efficient-fine-tuning-for | 71.2% |
| designing-effective-sparse-expert-models | 92.1% |
| fnet-mixing-tokens-with-fourier-transforms | 69% |
| debertav3-improving-deberta-using-electra | 92.7% |
| opt-iml-scaling-language-model-instruction | 84.8% |
| Model 57 | 83.6% |
| entailment-as-few-shot-learner | 90.5% |
| squeezebert-what-can-computer-vision-teach | 73.2% |
| toward-efficient-language-model-pretraining | 96% |
| finetuned-language-models-are-zero-shot | 84.5% |
| toward-efficient-language-model-pretraining | 94.1% |
| hungry-hungry-hippos-towards-language | 52.3% |
| opt-iml-scaling-language-model-instruction | 66.8% |
| exploring-the-limits-of-transfer-learning | 91.1% |
| 190910351 | 66% |
| bloomberggpt-a-large-language-model-for | 54.9% |
| muppet-massive-multi-task-representations | 92.8% |
| ask-me-anything-a-simple-strategy-for | 61.7% |
| exploring-the-limits-of-transfer-learning | 92.5% |
| exploring-the-limits-of-transfer-learning | 69.9% |
| language-models-are-few-shot-learners | 69% |
| ernie-20-a-continual-pre-training-framework | 74.8% |
| sensebert-driving-some-sense-into-bert | 67.5% |
| finetuned-language-models-are-zero-shot | 84.1% |
| ernie-20-a-continual-pre-training-framework | 80.2% |
| roberta-a-robustly-optimized-bert-pretraining | 88.2% |
| not-all-layers-are-equally-as-important-every | 63 |
| opt-iml-scaling-language-model-instruction | 58.1% |
| finetuned-language-models-are-zero-shot | 91.7% |
| spanbert-improving-pre-training-by | 79.0% |
| deberta-decoding-enhanced-bert-with | 93.2% |
| hungry-hungry-hippos-towards-language | 59.2% |
| roberta-a-robustly-optimized-bert-pretraining | 88.2% |
| how-to-train-bert-with-an-academic-budget | 57.7% |
| not-all-layers-are-equally-as-important-every | 54.7 |
| clear-contrastive-learning-for-sentence | 79.8% |
| opt-iml-scaling-language-model-instruction | 83.8% |
| bloomberggpt-a-large-language-model-for | 57.4% |
| a-statistical-framework-for-low-bitwidth | 86.8 |