← Back to Quantization
mixed int5/int6 QAT
QuantizationUsed in
5 PRs
Best BPB
1.1466
Avg BPB
1.1779
Submissions
Hyperparameters Across PRs
| pr_number | bits | scope |
|---|---|---|
| 351 | — | MLP weights and attention weights |
| 352 | — | MLP weights int5, attention weights int6, embeddings fp16 |
| 421 | — | MLP int5, attention int6, embeddings int8 |
| 694 | 5 | MLP and attention |
| 822 | — | MLP in int5, attention in int6 |