File size: 6,979 Bytes
b5d19c4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
{
  "op": "com.microsoft.GatedAdd",
  "cases": [
    {
      "name": "rank3_rows_f32",
      "inputs": {
        "X": {
          "dtype": "float32",
          "shape": [2, 3, 8],
          "data": { "kind": "fillFloat32", "sinStep": 0.13, "cosStep": 0.29 }
        },
        "Y": {
          "dtype": "float32",
          "shape": [2, 3, 8],
          "data": { "kind": "fillFloat32", "sinStep": 0.31, "cosStep": 0.07 }
        },
        "gate": {
          "dtype": "float32",
          "shape": [2, 3, 1],
          "data": { "kind": "fillFloat32", "sinStep": 0.41, "cosStep": 0.17, "offset": 0.75 }
        }
      },
      "outputs": {
        "output": { "dtype": "float32", "shape": [2, 3, 8], "tolerance": 0.000001, "relTolerance": 0.000001 }
      }
    },
    {
      "name": "rank2_channels_odd_f32",
      "inputs": {
        "X": {
          "dtype": "float32",
          "shape": [5, 7],
          "data": { "kind": "fillFloat32", "sinStep": 0.19, "cosStep": 0.23 }
        },
        "Y": {
          "dtype": "float32",
          "shape": [5, 7],
          "data": { "kind": "fillFloat32", "sinStep": 0.37, "cosStep": 0.11 }
        },
        "gate": {
          "dtype": "float32",
          "shape": [5, 1],
          "data": { "kind": "fillFloat32", "sinStep": 0.53, "cosStep": 0.29, "offset": -1.25 }
        }
      },
      "outputs": { "output": { "dtype": "float32", "shape": [5, 7], "tolerance": 0.000001, "relTolerance": 0.000001 } }
    },
    {
      "name": "rank2_channels_not_vec4_aligned_f32",
      "inputs": {
        "X": {
          "dtype": "float32",
          "shape": [4, 6],
          "data": { "kind": "fillFloat32", "sinStep": 0.17, "cosStep": 0.43 }
        },
        "Y": {
          "dtype": "float32",
          "shape": [4, 6],
          "data": { "kind": "fillFloat32", "sinStep": 0.61, "cosStep": 0.13 }
        },
        "gate": { "dtype": "float32", "shape": [4, 1], "data": { "kind": "values", "values": [2.0, -3.0, 0.5, 7.0] } }
      },
      "outputs": { "output": { "dtype": "float32", "shape": [4, 6], "tolerance": 0.000001, "relTolerance": 0.000001 } }
    },
    {
      "name": "rank1_single_row_f32",
      "inputs": {
        "X": { "dtype": "float32", "shape": [16], "data": { "kind": "fillFloat32", "sinStep": 0.27, "cosStep": 0.09 } },
        "Y": { "dtype": "float32", "shape": [16], "data": { "kind": "fillFloat32", "sinStep": 0.47, "cosStep": 0.21 } },
        "gate": { "dtype": "float32", "shape": [1], "data": { "kind": "values", "values": [-1.75] } }
      },
      "outputs": { "output": { "dtype": "float32", "shape": [16], "tolerance": 0.000001, "relTolerance": 0.000001 } }
    },
    {
      "name": "rank4_rows_f32",
      "inputs": {
        "X": {
          "dtype": "float32",
          "shape": [2, 2, 3, 4],
          "data": { "kind": "fillFloat32", "sinStep": 0.11, "cosStep": 0.37 }
        },
        "Y": {
          "dtype": "float32",
          "shape": [2, 2, 3, 4],
          "data": { "kind": "fillFloat32", "sinStep": 0.23, "cosStep": 0.59 }
        },
        "gate": {
          "dtype": "float32",
          "shape": [2, 2, 3, 1],
          "data": { "kind": "fillFloat32", "sinStep": 0.71, "cosStep": 0.19, "offset": 1.5 }
        }
      },
      "outputs": {
        "output": { "dtype": "float32", "shape": [2, 2, 3, 4], "tolerance": 0.000001, "relTolerance": 0.000001 }
      }
    },
    {
      "name": "gate_broadcast_rows_pinned",
      "provenance": {
        "notes": "Hand-computed from the schema formula output = X + round_to_T(Y * gate); every value is exact in float32, so the expectation is independent of the reference."
      },
      "inputs": {
        "X": {
          "dtype": "float32",
          "shape": [2, 3],
          "data": { "kind": "values", "values": [1.0, 2.0, 3.0, 4.0, 5.0, 6.0] }
        },
        "Y": {
          "dtype": "float32",
          "shape": [2, 3],
          "data": { "kind": "values", "values": [0.5, -1.0, 2.0, 10.0, -0.25, 0.125] }
        },
        "gate": { "dtype": "float32", "shape": [2, 1], "data": { "kind": "values", "values": [2.0, -0.5] } }
      },
      "outputs": {
        "output": {
          "dtype": "float32",
          "shape": [2, 3],
          "data": { "kind": "values", "values": [2.0, 0.0, 7.0, -1.0, 5.125, 5.9375] }
        }
      }
    },
    {
      "name": "f16_product_rounds_to_type_pinned",
      "provenance": {
        "notes": "Pins the round_to_T rule that separates this op from a wider-precision fused multiply-add. Row 0 uses gate = 1 + 2^-10, so Y = 1025 gives a real product of 1026.0009765625 that float16 rounds to 1026.0; X = -1026 then cancels it exactly to 0. A kernel that let the product stay unrounded -- by contracting the multiply into the add -- would return 2^-10 there instead. Every other value is exact in float16, so the whole expectation is hand-computable."
      },
      "inputs": {
        "X": {
          "dtype": "float16",
          "shape": [2, 4],
          "data": { "kind": "values", "values": [-1026.0, -1000.0, 0.5, -8.0, 0.5, -1.0, 0.25, 3.0] }
        },
        "Y": {
          "dtype": "float16",
          "shape": [2, 4],
          "data": { "kind": "values", "values": [1025.0, 1024.0, 512.0, 8.0, 1.0, 2.0, 3.0, 4.0] }
        },
        "gate": { "dtype": "float16", "shape": [2, 1], "data": { "kind": "values", "values": [1.0009765625, 2.0] } }
      },
      "outputs": {
        "output": {
          "dtype": "float16",
          "shape": [2, 4],
          "data": { "kind": "values", "values": [0.0, 25.0, 513.0, 0.0078125, 2.5, 3.0, 6.25, 11.0] },
          "tolerance": 0.0005
        }
      }
    },
    {
      "name": "f16_rows",
      "inputs": {
        "X": {
          "dtype": "float16",
          "shape": [3, 16],
          "data": { "kind": "fillFloat32", "sinStep": 0.13, "cosStep": 0.29, "offset": 1.25 }
        },
        "Y": {
          "dtype": "float16",
          "shape": [3, 16],
          "data": { "kind": "fillFloat32", "sinStep": 0.31, "cosStep": 0.07, "offset": -0.75 }
        },
        "gate": { "dtype": "float16", "shape": [3, 1], "data": { "kind": "values", "values": [1.5, -0.5, 2.25] } }
      },
      "outputs": { "output": { "dtype": "float16", "shape": [3, 16], "tolerance": 0.0005, "relTolerance": 0.002 } }
    },
    {
      "name": "f16_channels_odd",
      "inputs": {
        "X": {
          "dtype": "float16",
          "shape": [4, 5],
          "data": { "kind": "fillFloat32", "sinStep": 0.43, "cosStep": 0.17, "offset": -1.5 }
        },
        "Y": {
          "dtype": "float16",
          "shape": [4, 5],
          "data": { "kind": "fillFloat32", "sinStep": 0.19, "cosStep": 0.53, "offset": 0.875 }
        },
        "gate": { "dtype": "float16", "shape": [4, 1], "data": { "kind": "values", "values": [0.75, -1.25, 3.0, 0.5] } }
      },
      "outputs": { "output": { "dtype": "float16", "shape": [4, 5], "tolerance": 0.0005, "relTolerance": 0.002 } }
    }
  ]
}