Uses torch._inductor.compile instead of torch.compile in benchmark script to avoid segmentation by Dynamo #5231
Job | Run time |
---|---|
3s | |
12s | |
35s | |
23s | |
2m 39s | |
2m 51s | |
1m 7s | |
49s | |
3m 23s | |
3m 44s | |
1s | |
15m 47s |
Job | Run time |
---|---|
3s | |
12s | |
35s | |
23s | |
2m 39s | |
2m 51s | |
1m 7s | |
49s | |
3m 23s | |
3m 44s | |
1s | |
15m 47s |