[Apache TVM Discuss] [Questions] Repeatable build/compile

[cui via Apache TVM Discuss <no...@discuss.tvm.ai>]

as far as I know, traditional compiler would not use some ML tech to opt their opt. but tvm can use autoTVM/auto-scheduler some ML tech, but some ML tech will use probability. Whether if authTVM/auto-scheduler will generate different assembly code when input is same model file in difference build?





---
[Visit Topic](https://discuss.tvm.apache.org/t/repeatable-build-compile/11516/1) to respond.

You are receiving this because you enabled mailing list mode.

To unsubscribe from these emails, [click here](https://discuss.tvm.apache.org/email/unsubscribe/3ff33ef2a7ab96ca65aa5d4a3680a101cd2b18b8d70a1a803b3ac2c98c6ae0cd).

[Apache TVM Discuss] [Questions] Repeatable build/compile

[cronos via Apache TVM Discuss <no...@discuss.tvm.ai>]

Yes, different autoTVM/autoscheduler runs on the same network can yield different implementations.

If you know your model doesnt change, you would autoschedule once and save the log files of the optimized implementation. If for some reason you need to recompile your model you would tell TVM to just look for this logfile with the implementations so you get the same as before





---
[Visit Topic](https://discuss.tvm.apache.org/t/repeatable-build-compile/11516/2) to respond.

You are receiving this because you enabled mailing list mode.

To unsubscribe from these emails, [click here](https://discuss.tvm.apache.org/email/unsubscribe/20d8a339bb1f7afd4e3c102b1e257483d0f6dffb84058bea9b8041d10d869af9).