I’m a beginner, so please excuse me if there are some misunderstandings here. I’m trying to tune the YOLOv11n model through a modified version of the End-to-End Optimize Model tutorial. The tuning process seems to work fine, but there are some tasks that just get finished without running, so the tuning fails at the end.
2025-07-22 03:36:33 [INFO] [task_scheduler.cc:182] TaskScheduler picks Task #39: "fused_reshape6_split4"
2025-07-22 03:38:37 [INFO] [task_scheduler.cc:304] Task #39 has finished. Remaining task(s): 89
36 | fused_conv2d12_add11_tir_sigmoid9_multiply9 | 52838400 | 1 | 433.5437 | 121.8756 | 121.8756 | 5 |
37 | split | 1 | 1 | 0.0000 | 96.4153 | 96.4153 | 1 |
38 | fused_conv2d21_add18_tir_sigmoid14_multiply14 | 13209600 | 2 | 264.5756 | 49.9275 | 99.8550 | 5 |
39 | fused_reshape6_split4 | 1 | 1 | N/A | N/A | N/A | 0 | Y
40 | fused_conv2d38_add18_tir_sigmoid14_multiply14 | 118067200 | 1 | 324.6641 | 363.6595 | 363.6595 | 5 |
I think this has something to do with SampleInitPopulation() failing to generate valid initial samples, and nothing gets past the VerifyGPUCode() check.
2025-07-22 03:38:35 [INFO] [evolutionary_search.cc:533] Sample-Init-Population summary:
Postproc #0 [meta_schedule.DisallowDynamicLoop(0xaaaae58f5a08)]: 0 failure(s)
Postproc #1 [meta_schedule.RewriteCooperativeFetch(0xaaaae5dbfa88)]: 0 failure(s)
Postproc #2 [meta_schedule.RewriteUnboundBlock(0xaaaae61a4a58)]: 0 failure(s)
Postproc #3 [meta_schedule.RewriteParallelVectorizeUnroll(0xaaaae5920478)]: 0 failure(s)
Postproc #4 [meta_schedule.RewriteReductionBlock(0xaaaae61bbcb8)]: 0 failure(s)
Postproc #5 [meta_schedule.VerifyGPUCode(0xaaaae5dc0068)]: 25088 failure(s)
Postproc #6 [meta_schedule.RewriteTensorize(0xaaaae5dc02c8)]: 0 failure(s)
2025-07-22 03:38:37 [INFO] [evolutionary_search.cc:533] Sample-Init-Population summary:
Postproc #0 [meta_schedule.DisallowDynamicLoop(0xaaaae58f5a08)]: 0 failure(s)
Postproc #1 [meta_schedule.RewriteCooperativeFetch(0xaaaae5dbfa88)]: 0 failure(s)
Postproc #2 [meta_schedule.RewriteUnboundBlock(0xaaaae61a4a58)]: 0 failure(s)
Postproc #3 [meta_schedule.RewriteParallelVectorizeUnroll(0xaaaae5920478)]: 0 failure(s)
Postproc #4 [meta_schedule.RewriteReductionBlock(0xaaaae61bbcb8)]: 0 failure(s)
Postproc #5 [meta_schedule.VerifyGPUCode(0xaaaae5dc0068)]: 25600 failure(s)
Postproc #6 [meta_schedule.RewriteTensorize(0xaaaae5dc02c8)]: 0 failure(s)
2025-07-22 03:38:37 [WARNING] [evolutionary_search.cc:722] Cannot sample enough initial population, evolutionary search failed.
I’ve tried using other search strategies, modifying parameters in evolutionary_search.py and nothing seems to work so far. Is this a known issue or is there something I can do to get past this?