← Back to Quantization

mixed int7/int8

Quantization
Used in
6 PRs
Best BPB
1.0579
Avg BPB
1.0597

Hyperparameters Across PRs

pr_numberbitsscope
1925embeddings and row gate
2007embeddings and model weights
2060embeddings and blocks
2109embeddings and attention gate
21578embeddings and selected groups
21637embeddings and attention gates