| Benchmark | Model | Rank | Results |
|---|---|---|---|
| language-modelling-on-lambada | Llama-3-70B with SQ-format (W4A(SQ6), static, bb=16) | – | Perplexity: 2.64 |
| question-answering-on-piqa | Llama-3-70B with SQ-format (W4A(SQ6)) | – | Accuracy: 84.06 |
| sentence-completion-on-hellaswag | Llama-3-70B with SQ-format (W4A(SQ6)) | – | Accuracy: 84.44 |