Matmul op for relay

yongwww · February 1, 2019, 12:41am

Feel free to close the topic. Just wanna know the reason why we don’t implement matmul op for relay, I know we can use dense instead, is there any other reason?

junrushao · February 2, 2019, 11:28pm

What’s the diff between the two? bias?

yongwww · February 4, 2019, 8:30am

yup, bias. Y = XW^T + b

junrushao · February 4, 2019, 8:44am

I guess that you don’t have to provide the bias term, so that dense degenerates to matmul.

github.com

dmlc/tvm/blob/3a75b13d663a3bb14e11f5b7fd91980163f9dd75/topi/include/topi/nn/dense.h#L47


auto in_dim = data->shape[1];
auto out_dim = weight->shape[0];


auto k = tvm::reduce_axis(Range(0, in_dim), "k");
auto matmul = tvm::compute(
  { batch, out_dim },
  [&](Var i, Var j) {
    return tvm::sum(data(i, k) * weight(j, k), { k });
  }, "tensor", "dense");


if (bias.defined()) {
  matmul = tvm::compute(
    { batch, out_dim },
    [&](Var i, Var j) {
      return matmul(i, j) + bias(j);
    }, "tensor", kBroadcast);
}


return matmul;
}

yongwww · February 4, 2019, 6:12pm

yeah, dense will do the same thing as matmul without specifying bias. I like the name tvm.relay.matmul more than tvm.relay.nn.dense for matrix mulitiplication. Anyway, dense is good.

junrushao · February 4, 2019, 6:24pm

Sure. That sounds reasonable.

junrushao · February 4, 2019, 6:34pm

I guess I misread relay for topi…anyway…

jroesch · February 4, 2019, 9:10pm

We could expose matmul as a synonym. I think this might actually be a good idea, the overloaded use of dense has confused many users (including me when I started working on TVM).

yongwww · February 4, 2019, 9:26pm

thanks, I agree. Confused me, too

yongwww · February 4, 2019, 10:04pm

I will send out a pr to expose matmul then