Panic with gorgonia, but why ?

panic_test.go

package main

import (
        "log"
        "testing"

        G "gorgonia.org/gorgonia"
        "gorgonia.org/tensor"
)

func TestMul(t *testing.T) {
        g := G.NewGraph()

        vecT := tensor.New(tensor.WithBacking([]float32{1, 1}), tensor.WithShape(2))
        vec := G.NewVector(g, tensor.Float32, G.WithName("x"), G.WithShape(2), G.WithValue(vecT))
        matT := tensor.New(tensor.WithBacking([]float32{0.95, 0.8}), tensor.WithShape(1, 2))
        mat := G.NewMatrix(g, tensor.Float32, G.WithName("W"), G.WithShape(1, 2), G.WithValue(matT))

        z, err := G.Mul(mat, vec)
        if err != nil {
                t.Fatal(err)
        }

        // create a VM to run the program on
        machine := G.NewLispMachine(g, G.ExecuteFwdOnly())

        if err = machine.RunAll(); err != nil {
                t.Fatal(err)
        }

        if z.Value().Data() != 0.95+0.8 {
                t.Fatal("Expectecd %v, got %v", 0.95+0.8, z.Value().Data())
        }

}

stack.txt

panic: blas: index of a out of range

goroutine 18 [running]:
gonum.org/v1/gonum/blas/gonum.Implementation.Sgemv(0x6f, 0x1, 0x2, 0x3f800000, 0xc42015e130, 0x2, 0x2, 0x1, 0xc42015e0a0, 0x2, ...)
        /Users/olivierwulveryck/GOPROJECTS/src/gonum.org/v1/gonum/blas/gonum/level2single.go:33 +0x96f
gonum.org/v1/gonum/blas/gonum.(*Implementation).Sgemv(0x18c2b20, 0x6f, 0x1, 0x2, 0x3f800000, 0xc42015e130, 0x2, 0x2, 0x1, 0xc42015e0a0, ...)
        <autogenerated>:1 +0x135
gorgonia.org/tensor.StdEng.MatVecMul(0x18777e0, 0xc42017a0b0, 0x18777e0, 0xc42017a000, 0x18777e0, 0xc42017a2c0, 0xc42017a201, 0x1a733f8)
        /Users/olivierwulveryck/GOPROJECTS/src/gorgonia.org/tensor/defaultengine_linalg.go:422 +0x615
gorgonia.org/tensor.(*StdEng).MatVecMul(0x18c2b20, 0x18777e0, 0xc42017a0b0, 0x18777e0, 0xc42017a000, 0x18777e0, 0xc42017a2c0, 0x0, 0xc42017a2c0)
        <autogenerated>:1 +0x7b
gorgonia.org/tensor.(*Dense).MatVecMul(0xc42017a0b0, 0x18777e0, 0xc42017a000, 0x0, 0x0, 0x0, 0xc42017a2c0, 0x0, 0x0)
        /Users/olivierwulveryck/GOPROJECTS/src/gorgonia.org/tensor/dense_linalg.go:97 +0x463
gorgonia.org/tensor.MatVecMul(0x18777e0, 0xc42017a0b0, 0x18777e0, 0xc42017a000, 0x0, 0x0, 0x0, 0xc42017a0b8, 0x2, 0xc420156f00, ...)
        /Users/olivierwulveryck/GOPROJECTS/src/gorgonia.org/tensor/api_arith.go:454 +0x2a0
gorgonia.org/gorgonia.linAlgBinOp.do(0x1000001, 0xc420156fe0, 0x2, 0x2, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, ...)
        /Users/olivierwulveryck/GOPROJECTS/src/gorgonia.org/gorgonia/op_math.go:790 +0x2f3
gorgonia.org/gorgonia.linAlgBinOp.Do(0x1000001, 0xc420156fe0, 0x2, 0x2, 0x100, 0x80, 0x7f, 0xc4200429c8)
        /Users/olivierwulveryck/GOPROJECTS/src/gorgonia.org/gorgonia/op_math.go:657 +0x7c
gorgonia.org/gorgonia.(*linAlgBinOp).Do(0xc42015e1d8, 0xc420156fe0, 0x2, 0x2, 0x199a868, 0x1537b40, 0xc4200429f0, 0x100e74e)
        <autogenerated>:1 +0x71
gorgonia.org/gorgonia.(*ExternalOp).Do(0xc42015a690, 0xc420156fe0, 0x2, 0x2, 0x2, 0x2, 0xc420042b70, 0xc420156fc0)
        /Users/olivierwulveryck/GOPROJECTS/src/gorgonia.org/gorgonia/execution.go:109 +0xa1
gorgonia.org/gorgonia.dvBindVar(0x18746e0, 0xc42015a690, 0xc4201543b0, 0x2, 0x2, 0x0, 0x0, 0x0)
        /Users/olivierwulveryck/GOPROJECTS/src/gorgonia.org/gorgonia/dual.go:284 +0x75
gorgonia.org/gorgonia.(*lispMachine).forward(0xc4201761e0, 0x0, 0x0)
        /Users/olivierwulveryck/GOPROJECTS/src/gorgonia.org/gorgonia/vm_genera.go:381 +0x1f8b
gorgonia.org/gorgonia.(*lispMachine).runall(0xc4201761e0, 0xc4201741e0, 0xc420174240)
        /Users/olivierwulveryck/GOPROJECTS/src/gorgonia.org/gorgonia/vm_genera.go:255 +0x44
created by gorgonia.org/gorgonia.(*lispMachine).RunAll
        /Users/olivierwulveryck/GOPROJECTS/src/gorgonia.org/gorgonia/vm_genera.go:125 +0x1bb
exit status 2
FAIL    tmp/strides     0.015s

When I change the LispMachine for a TapeMachine, it gives me:

=== RUN   TestMul
--- FAIL: TestMul (0.00s)
        stride_test.go:28: PC: 3: PC 3. Failed to execute instruction A × b     [CPU1 CPU0]     CPU2    true    false   true: Happened while attempting to execute A × b        [CPU1 CPU0]       CPU2    true    false   true. Node is 2. Register was: 2 : Expected Tensor as preallocated value. Got 0 of *gorgonia.F32 instead
FAIL
exit status 1
FAIL    tmp/strides     0.010s