Repository navigation
Commit b253af8
authored
Honor configured A16 weight ranges (pytorch#22769) (pytorch#22769)
Summary:
Forward the existing weight_qmin and weight_qmax arguments into the Arm
symmetric A16W8 quantization configuration in both the fbcode and xplat
mirrors. Callers that omit the arguments retain the existing [-127, 127]
weight range; restricted-range recipes receive the requested qspec.
Reviewed By: WongJohnson
Differential Revision: D1197792051 parent 3235f58 commit b253af8
1 file changed
Lines changed: 2 additions & 0 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
484 | 484 | | |
485 | 485 | | |
486 | 486 | | |
| 487 | + | |
| 488 | + | |
487 | 489 | | |
488 | 490 | | |
489 | 491 | | |
| |||
0 commit comments