| language-modelling-on-c4 | Zeropoint LLM.int8 13B (vector-wise + decomp) | #2 | Perplexity: 12.45 |
| language-modelling-on-c4 | LLM.float32 6.7B | #5 | Perplexity: 13.3 |
| language-modelling-on-c4 | LLM.float32 2.7B | #6 | Perplexity: 14.43 |
| language-modelling-on-c4 | LLM.float32 1.3B | #9 | Perplexity: 15.91 |
| linguistic-acceptability-on-cola | RoBERTa-large 355M (MLP quantized vector-wise, fine-tuned) | #12 | Accuracy: 68.6% |
| natural-language-inference-on-multinli | RoBERTa-large 355M (MLP quantized vector-wise, fine-tuned) | #8 | Matched: 90.2 |
| natural-language-inference-on-qnli | RoBERTa-large 355M (MLP quantized vector-wise, fine-tuned) | #11 | Accuracy: 94.7% |
| natural-language-inference-on-rte | RoBERTa-large 355M (MLP quantized vector-wise, fine-tuned) | #22 | Accuracy: 85.4% |
| semantic-textual-similarity-on-mrpc | RoBERTa-large 355M (MLP quantized vector-wise, fine-tuned) | #6 | Accuracy: 91.0% |
| semantic-textual-similarity-on-sts-benchmark | RoBERTa-large 355M (MLP quantized vector-wise, fine-tuned) | #7 | Pearson Correlation: 0.919 |
| sentiment-analysis-on-sst-2-binary | RoBERTa-large 355M (MLP quantized vector-wise, fine-tuned) | #15 | Accuracy: 96.4 |