Yuanjing Shi

bert base on llvm 20k trials: ``` ID | Name | FLOP | Weight | Speed (GFLOPS) | Latency (us) | Weighted Latency (us) | Trials | Terminated --------------------------------------------------------------------------------------------------------------------------------------------------------------------------- 0...

@zxybazh The pandas warning should be suppressed now with the last commit.

I took a quick look and this PR should be in pretty good shape after CI is fixed. cc: @masahi Can you also take a look at this? Would really...

@areusch this should be closed by #12993

Hi @Sunny-Island, I wonder if you are still on this PR? There are quite some changes introduced to the [use of xgboost in meta schedule](https://github.com/apache/tvm/blob/main/python/tvm/meta_schedule/cost_model/xgb_model.py) after your PR. I wonder...

relay @shingjan torch @shingjan pytorch @shingjan tvmscript @shingjan tir @shingjan

Will vote for integrating synr into TVM as moving it into tlcpack would still leave the process of checking synr changes into TVM cumbersome.

@tkonolige Thanks for the clarification. My points: 1. It is a good thing that as a separate repo, synr could have its own CI. But almost every change we bring...

Thanks! @cchan It did for me as well. Just wonder if `zlib1g-dev` should be part of the triton's dependencies.

@Jokeren Thanks for the reply! Does that mean you guys tried different cuda codegen path for triton, like MLIR/CUTLASS/TensorRT and CUDA kernels generated by MLIR yield the best result? Would...