Support attention_bias on LLaMA architecture #6658
| Job | Run time |
|---|---|
| 3m 48s | |
| 1m 34s | |
| 1m 27s | |
| 1m 23s | |
| 1m 17s | |
| 1m 32s | |
| 2m 18s | |
| 1m 36s | |
| 3m 47s | |
| 1m 31s | |
| 3m 23s | |
| 3m 12s | |
| 8m 1s | |
| 4m 56s | |
| 2m 36s | |
| 4m 41s | |
| 20m 34s | |
| 2m 43s | |
| 3m 16s | |
| 2m 11s | |
| 17m 13s | |
| 3m 22s | |
| 2m 7s | |
| 5m 12s | |
| 1m 39s | |
| 3m 5s | |
| 0s | |
| 1h 48m 24s |