[torchlib] Fix and improve quantization support #7421
| Job | Run time |
|---|---|
| 5s | |
| 2m 51s | |
| 4m 22s | |
| 2m 15s | |
| 1m 20s | |
| 2m 19s | |
| 2m 13s | |
| 3m 3s | |
| 1m 20s | |
| 2m 4s | |
| 2m 12s | |
| 3m 3s | |
| 1m 20s | |
| 3m 35s | |
| 2m 28s | |
| 3m 45s | |
| 4m 11s | |
| 3m 7s | |
| 3m 10s | |
| 3m 4s | |
| 3m 8s | |
| 3m 25s | |
| 2m 15s | |
| 3m 17s | |
| 3m 5s | |
| 1m 56s | |
| 1m 46s | |
| 1m 45s | |
| 1m 43s | |
| 1m 38s | |
| 1m 40s | |
| 1m 45s | |
| 1m 58s | |
| 17s | |
| 1h 21m 25s |