Large Language Model Inference on Llama 3.2 1B
1.94TPOTHBaseline
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| BaselineBackbone=Llama-3.2-1B2026.03 | 1.94 | 7.69 | 3.6 | |
| Vocab. TrimmingBackbone=Llama-3.2-1B2026.03 | 1.07 | 6.82 | 2.73 | |
| SVDSoftmaxBackbone=Llama-3.2-1B2026.03 | 0.61 | 6.36 | 2.27 | |
| FlashHeadBackbone=Llama-3.2-1B2026.03 | 0.4 | 6.15 | 2.06 |