%0 = "tf.MatMul"(%a, %b) {transpose_a = true, transpose_b = true} : (tensor<7x5xf32>, tensor<11x7xf32>) -> tensor<5x11xf32>

  func.return %0 : tensor<5x11xf32>
}

// Verify that MatMul with ranked inputs are lowered to HLO.
// CHECK-LABEL: matmul_ranked
func.func @matmul_ranked(%a: tensor<?x7xf32>, %b: tensor<7x?xf32>) -> tensor<?x?xf32> {
  // CHECK: "mhlo.dot"

}

// CHECK-LABEL: @FuseReshapeAroundBMMRHS
func.func @FuseReshapeAroundBMMRHS(%arg0: tensor<1x3x6x5x1024xf32>) -> tensor<1x3x6x5x8192xf32> attributes {tf.entry_function = {control_outputs = "", inputs = "inputs", outputs = "Identity_1"}} {
  %cst = arith.constant dense_resource<__elided__> : tensor<1x1024x8192xf32>
  %cst_0 = arith.constant dense_resource<__elided__> : tensor<5xi32>

plugin");return t}};fc.exports=Pt;Pt.default=Pt;ov.registerProcessor(Pt);av.registerProcessor(Pt)});var dc=v((bT,pc)=>{l();"use strict";var lv=gr(),uv=Ps(),fv=yr(),cv=$i(),pv=Vi(),dv=Et(),hv=ji();function Cr(i,e){if(Array.isArray(i))return i.map(n=>Cr(n));let{inputs:t,...r}=i;if(t){e=[];for(let n of t){let a={...n,__proto__:pv.prototype};a.map&&(a.map={...a.map,__proto__:uv.prototype}),e.push(a)}}if(r.nodes&&(r.nodes=i.nodes.map(n=>Cr(n,e))),r.source){let{inputId:n,...a}=r.source;r.source=a,n!=null&&(r.source...