/packages/modules/NeuralNetworks/runtime/test/specs/V1_3/ |
D | bidirectional_sequence_lstm.mod.py | 26 n_output, argument 104 "{{{}, {}}}".format(n_cell, n_output)) 107 "{{{}, {}}}".format(n_cell, n_output)) 110 "{{{}, {}}}".format(n_cell, n_output)) 113 "{{{}, {}}}".format(n_cell, n_output)) 149 "{{{}, {}}}".format(n_cell, n_output)) 152 "{{{}, {}}}".format(n_cell, n_output)) 155 "{{{}, {}}}".format(n_cell, n_output)) 158 "{{{}, {}}}".format(n_cell, n_output)) 181 "{{{}, {}}}".format(n_batch, n_output)) [all …]
|
D | bidirectional_sequence_lstm_state_output.mod.py | 23 n_output = 4 variable 38 "fw_recurrent_to_input_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 40 "fw_recurrent_to_forget_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 42 "fw_recurrent_to_cell_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 44 "fw_recurrent_to_output_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 63 "fw_projection_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_output, n_cell)) 65 "fw_projection_bias", "TENSOR_FLOAT32", "{{{}}}".format(n_output)) 77 "bw_recurrent_to_input_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 79 "bw_recurrent_to_forget_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 81 "bw_recurrent_to_cell_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) [all …]
|
D | unidirectional_sequence_lstm_layer_norm_cifg_peephole_state_output.mod.py | 28 n_output = 3 variable 46 "{%d, %d}" % (n_cell, n_output)) 48 "{%d, %d}" % (n_cell, n_output)) 51 "{%d, %d}" % (n_cell, n_output)) 67 "{%d,%d}" % (n_output, n_cell)) 71 "{%d, %d}" % (n_batch, n_output)) 90 "{%d, %d, %d}" % (max_time, n_batch, n_output)) 92 "{%d, %d}" % (n_batch, n_output)) 181 golden_output[(max_time - 1) * (n_batch * n_output):], 189 input0[output_state_in] = [0 for _ in range(n_batch * n_output)]
|
/packages/modules/NeuralNetworks/runtime/test/specs/V1_2/ |
D | layer_norm_lstm.mod.py | 25 n_output = 3 variable 40 "{%d, %d}" % (n_cell, n_output)) 43 "{%d, %d}" % (n_cell, n_output)) 45 "{%d, %d}" % (n_cell, n_output)) 48 "{%d, %d}" % (n_cell, n_output)) 65 "{%d,%d}" % (n_output, n_cell)) 69 "{%d, %d}" % (n_batch, n_output)) 89 "{%d, %d}" % (n_batch, n_output)) 92 output = Output("output", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 197 n_output = 3 variable [all …]
|
D | quantized_lstm.mod.py | 23 n_output = n_cell variable 32 [n_output, n_input], weights_scale, weights_zero_point) 39 [n_output, n_output], weights_scale, weights_zero_point) 45 BiasType = ("TENSOR_INT32", [n_output], weights_scale / 128., 0) 52 OutputType = ("TENSOR_QUANT8_ASYMM", (n_batch, n_output), 1 / 128, 128) 110 n_output = n_cell variable 119 [n_output, n_input], weights_scale, weights_zero_point) 130 [n_output, n_output], weights_scale, weights_zero_point) 140 BiasType = ("TENSOR_INT32", [n_output], weights_scale / 128., 0) 151 OutputType = ("TENSOR_QUANT8_ASYMM", (n_batch, n_output), 1 / 128, 128)
|
D | bidirectional_sequence_lstm_float16_batch_major.mod.py | 23 n_output = 4 variable 38 "fw_recurrent_to_input_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 40 "fw_recurrent_to_forget_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 42 "fw_recurrent_to_cell_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 44 "fw_recurrent_to_output_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 63 "fw_projection_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_output, n_cell)) 65 "fw_projection_bias", "TENSOR_FLOAT16", "{{{}}}".format(n_output)) 77 "bw_recurrent_to_input_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 79 "bw_recurrent_to_forget_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 81 "bw_recurrent_to_cell_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) [all …]
|
D | bidirectional_sequence_lstm.mod.py | 23 n_output = 4 variable 38 "fw_recurrent_to_input_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 40 "fw_recurrent_to_forget_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 42 "fw_recurrent_to_cell_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 44 "fw_recurrent_to_output_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 63 "fw_projection_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_output, n_cell)) 65 "fw_projection_bias", "TENSOR_FLOAT32", "{{{}}}".format(n_output)) 77 "bw_recurrent_to_input_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 79 "bw_recurrent_to_forget_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 81 "bw_recurrent_to_cell_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) [all …]
|
D | bidirectional_sequence_lstm_cifg_peephole.mod.py | 23 n_output = 4 variable 38 "fw_recurrent_to_input_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 40 "fw_recurrent_to_forget_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 42 "fw_recurrent_to_cell_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 44 "fw_recurrent_to_output_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 63 "fw_projection_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_output, n_cell)) 65 "fw_projection_bias", "TENSOR_FLOAT32", "{{{}}}".format(n_output)) 77 "bw_recurrent_to_input_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 79 "bw_recurrent_to_forget_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 81 "bw_recurrent_to_cell_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) [all …]
|
D | unidirectional_sequence_lstm_batch_major_peephole_projection_bias.mod.py | 28 n_output = 16 variable 43 "{%d, %d}" % (n_cell, n_output)) 46 "{%d, %d}" % (n_cell, n_output)) 48 "{%d, %d}" % (n_cell, n_output)) 51 "{%d, %d}" % (n_cell, n_output)) 68 "{%d,%d}" % (n_output, n_cell)) 69 projection_bias = Input("projection_bias", "TENSOR_FLOAT32", "{%d}" % n_output) 72 "{%d, %d}" % (n_batch, n_output)) 86 output = Output("output", "TENSOR_FLOAT32", "{%d, %d, %d}" % (n_batch, max_time, n_output)) 702 input0[output_state_in] = [0 for _ in range(n_batch * n_output)]
|
D | lstm3_float16.mod.py | 25 n_output = 16 variable 34 …ut_weights = Input("recurrent_to_input_weights", "TENSOR_FLOAT16", "{%d, %d}" % (n_cell, n_output)) 35 …t_weights = Input("recurrent_to_forget_weights", "TENSOR_FLOAT16", "{%d, %d}" % (n_cell, n_output)) 36 …ell_weights = Input("recurrent_to_cell_weights", "TENSOR_FLOAT16", "{%d, %d}" % (n_cell, n_output)) 37 …t_weights = Input("recurrent_to_output_weights", "TENSOR_FLOAT16", "{%d, %d}" % (n_cell, n_output)) 48 projection_weights = Input("projection_weights", "TENSOR_FLOAT16", "{%d,%d}" % (n_output, n_cell)) 51 output_state_in = Input("output_state_in", "TENSOR_FLOAT16", "{%d, %d}" % (n_batch, n_output)) 59 output_state_out = Output("output_state_out", "TENSOR_FLOAT16", "{%d, %d}" % (n_batch, n_output)) 61 output = Output("output", "TENSOR_FLOAT16", "{%d, %d}" % (n_batch, n_output)) 621 input0[output_state_in] = [ 0 for _ in range(n_batch * n_output) ]
|
D | lstm3_state3_float16.mod.py | 25 n_output = 16 variable 34 …ut_weights = Input("recurrent_to_input_weights", "TENSOR_FLOAT16", "{%d, %d}" % (n_cell, n_output)) 35 …t_weights = Input("recurrent_to_forget_weights", "TENSOR_FLOAT16", "{%d, %d}" % (n_cell, n_output)) 36 …ell_weights = Input("recurrent_to_cell_weights", "TENSOR_FLOAT16", "{%d, %d}" % (n_cell, n_output)) 37 …t_weights = Input("recurrent_to_output_weights", "TENSOR_FLOAT16", "{%d, %d}" % (n_cell, n_output)) 48 projection_weights = Input("projection_weights", "TENSOR_FLOAT16", "{%d,%d}" % (n_output, n_cell)) 51 output_state_in = Input("output_state_in", "TENSOR_FLOAT16", "{%d, %d}" % (n_batch, n_output)) 59 …t_state_out = IgnoredOutput("output_state_out", "TENSOR_FLOAT16", "{%d, %d}" % (n_batch, n_output)) 61 output = Output("output", "TENSOR_FLOAT16", "{%d, %d}" % (n_batch, n_output)) 645 output_state_out: [ 0 for x in range(n_batch * n_output) ],
|
D | bidirectional_sequence_lstm_norm_fw_output.mod.py | 24 n_output = 3 variable 39 "fw_recurrent_to_input_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 41 "fw_recurrent_to_forget_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 43 "fw_recurrent_to_cell_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 45 "fw_recurrent_to_output_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 64 "fw_projection_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_output, n_cell)) 66 "fw_projection_bias", "TENSOR_FLOAT32", "{{{}}}".format(n_output)) 78 "bw_recurrent_to_input_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 80 "bw_recurrent_to_forget_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 82 "bw_recurrent_to_cell_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) [all …]
|
D | bidirectional_sequence_lstm_float16_batch_major_merge_outputs.mod.py | 24 n_output = 4 variable 39 "fw_recurrent_to_input_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 41 "fw_recurrent_to_forget_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 43 "fw_recurrent_to_cell_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 45 "fw_recurrent_to_output_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 64 "fw_projection_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_output, n_cell)) 66 "fw_projection_bias", "TENSOR_FLOAT16", "{{{}}}".format(n_output)) 78 "bw_recurrent_to_input_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 80 "bw_recurrent_to_forget_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 82 "bw_recurrent_to_cell_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) [all …]
|
D | bidirectional_sequence_lstm_merge_outputs.mod.py | 24 n_output = 4 variable 39 "fw_recurrent_to_input_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 41 "fw_recurrent_to_forget_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 43 "fw_recurrent_to_cell_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 45 "fw_recurrent_to_output_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 64 "fw_projection_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_output, n_cell)) 66 "fw_projection_bias", "TENSOR_FLOAT32", "{{{}}}".format(n_output)) 78 "bw_recurrent_to_input_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 80 "bw_recurrent_to_forget_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 82 "bw_recurrent_to_cell_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) [all …]
|
D | bidirectional_sequence_lstm_aux_input.mod.py | 25 n_output = 4 variable 40 "fw_recurrent_to_input_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 42 "fw_recurrent_to_forget_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 44 "fw_recurrent_to_cell_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 46 "fw_recurrent_to_output_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 65 "fw_projection_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_output, n_cell)) 67 "fw_projection_bias", "TENSOR_FLOAT32", "{{{}}}".format(n_output)) 79 "bw_recurrent_to_input_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 81 "bw_recurrent_to_forget_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) 83 "bw_recurrent_to_cell_weights", "TENSOR_FLOAT32", "{{{}, {}}}".format(n_cell, n_output)) [all …]
|
D | bidirectional_sequence_lstm_float16_batch_major_aux_input.mod.py | 26 n_output = 4 variable 41 "fw_recurrent_to_input_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 43 "fw_recurrent_to_forget_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 45 "fw_recurrent_to_cell_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 47 "fw_recurrent_to_output_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 66 "fw_projection_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_output, n_cell)) 68 "fw_projection_bias", "TENSOR_FLOAT16", "{{{}}}".format(n_output)) 80 "bw_recurrent_to_input_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 82 "bw_recurrent_to_forget_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) 84 "bw_recurrent_to_cell_weights", "TENSOR_FLOAT16", "{{{}, {}}}".format(n_cell, n_output)) [all …]
|
D | lstm3_state2_float16.mod.py | 25 n_output = 16 variable 34 …ut_weights = Input("recurrent_to_input_weights", "TENSOR_FLOAT16", "{%d, %d}" % (n_cell, n_output)) 35 …t_weights = Input("recurrent_to_forget_weights", "TENSOR_FLOAT16", "{%d, %d}" % (n_cell, n_output)) 36 …ell_weights = Input("recurrent_to_cell_weights", "TENSOR_FLOAT16", "{%d, %d}" % (n_cell, n_output)) 37 …t_weights = Input("recurrent_to_output_weights", "TENSOR_FLOAT16", "{%d, %d}" % (n_cell, n_output)) 48 projection_weights = Input("projection_weights", "TENSOR_FLOAT16", "{%d,%d}" % (n_output, n_cell)) 51 output_state_in = Input("output_state_in", "TENSOR_FLOAT16", "{%d, %d}" % (n_batch, n_output)) 59 output_state_out = Output("output_state_out", "TENSOR_FLOAT16", "{%d, %d}" % (n_batch, n_output)) 61 output = Output("output", "TENSOR_FLOAT16", "{%d, %d}" % (n_batch, n_output))
|
/packages/modules/NeuralNetworks/runtime/test/specs/V1_0/ |
D | lstm3.mod.py | 25 n_output = 16 variable 34 …ut_weights = Input("recurrent_to_input_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 35 …t_weights = Input("recurrent_to_forget_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 36 …ell_weights = Input("recurrent_to_cell_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 37 …t_weights = Input("recurrent_to_output_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 48 projection_weights = Input("projection_weights", "TENSOR_FLOAT32", "{%d,%d}" % (n_output, n_cell)) 51 output_state_in = Input("output_state_in", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 59 output_state_out = Output("output_state_out", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 61 output = Output("output", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 621 input0[output_state_in] = [ 0 for _ in range(n_batch * n_output) ]
|
D | lstm3_state3.mod.py | 25 n_output = 16 variable 34 …ut_weights = Input("recurrent_to_input_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 35 …t_weights = Input("recurrent_to_forget_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 36 …ell_weights = Input("recurrent_to_cell_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 37 …t_weights = Input("recurrent_to_output_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 48 projection_weights = Input("projection_weights", "TENSOR_FLOAT32", "{%d,%d}" % (n_output, n_cell)) 51 output_state_in = Input("output_state_in", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 59 …t_state_out = IgnoredOutput("output_state_out", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 61 output = Output("output", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 645 output_state_out: [ 0 for x in range(n_batch * n_output) ],
|
D | lstm2_state2.mod.py | 25 n_output = 4 variable 34 …ut_weights = Input("recurrent_to_input_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 35 …t_weights = Input("recurrent_to_forget_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 36 …ell_weights = Input("recurrent_to_cell_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 37 …t_weights = Input("recurrent_to_output_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 51 output_state_in = Input("output_state_in", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 59 …t_state_out = IgnoredOutput("output_state_out", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 61 output = Output("output", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 134 output_state_out: [ 0 for x in range(n_batch * n_output) ],
|
D | lstm3_state2.mod.py | 25 n_output = 16 variable 34 …ut_weights = Input("recurrent_to_input_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 35 …t_weights = Input("recurrent_to_forget_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 36 …ell_weights = Input("recurrent_to_cell_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 37 …t_weights = Input("recurrent_to_output_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 48 projection_weights = Input("projection_weights", "TENSOR_FLOAT32", "{%d,%d}" % (n_output, n_cell)) 51 output_state_in = Input("output_state_in", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 59 output_state_out = Output("output_state_out", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 61 output = Output("output", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output))
|
D | lstm3_state.mod.py | 25 n_output = 16 variable 34 …ut_weights = Input("recurrent_to_input_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 35 …t_weights = Input("recurrent_to_forget_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 36 …ell_weights = Input("recurrent_to_cell_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 37 …t_weights = Input("recurrent_to_output_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 48 projection_weights = Input("projection_weights", "TENSOR_FLOAT32", "{%d,%d}" % (n_output, n_cell)) 51 output_state_in = Input("output_state_in", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 59 output_state_out = Output("output_state_out", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 61 output = Output("output", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output))
|
/packages/modules/NeuralNetworks/common/cpu_operations/ |
D | LSTMTest.cpp | 78 LSTMOpModel(uint32_t n_batch, uint32_t n_input, uint32_t n_cell, uint32_t n_output, in LSTMOpModel() argument 83 n_output_(n_output), in LSTMOpModel() 94 input_shapes.push_back({n_batch, n_output}); in LSTMOpModel() 118 {n_batch, n_output}, in LSTMOpModel() 120 {n_batch, n_output}, in LSTMOpModel() 138 OutputStateIn_.insert(OutputStateIn_.end(), n_batch * n_output, 0.f); in LSTMOpModel() 277 const int n_output = 4; in TEST() local 279 LSTMOpModel lstm(n_batch, n_input, n_cell, n_output, in TEST() 292 {n_cell, n_output}, // recurrent_to_input_weight tensor in TEST() 293 {n_cell, n_output}, // recurrent_to_forget_weight tensor in TEST() [all …]
|
/packages/modules/NeuralNetworks/runtime/test/specs/V1_1/ |
D | lstm3_relaxed.mod.py | 25 n_output = 16 variable 34 …ut_weights = Input("recurrent_to_input_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 35 …t_weights = Input("recurrent_to_forget_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 36 …ell_weights = Input("recurrent_to_cell_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 37 …t_weights = Input("recurrent_to_output_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 48 projection_weights = Input("projection_weights", "TENSOR_FLOAT32", "{%d,%d}" % (n_output, n_cell)) 51 output_state_in = Input("output_state_in", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 59 output_state_out = Output("output_state_out", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 61 output = Output("output", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 622 input0[output_state_in] = [ 0 for _ in range(n_batch * n_output) ]
|
D | lstm3_state3_relaxed.mod.py | 25 n_output = 16 variable 34 …ut_weights = Input("recurrent_to_input_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 35 …t_weights = Input("recurrent_to_forget_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 36 …ell_weights = Input("recurrent_to_cell_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 37 …t_weights = Input("recurrent_to_output_weights", "TENSOR_FLOAT32", "{%d, %d}" % (n_cell, n_output)) 48 projection_weights = Input("projection_weights", "TENSOR_FLOAT32", "{%d,%d}" % (n_output, n_cell)) 51 output_state_in = Input("output_state_in", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 59 …t_state_out = IgnoredOutput("output_state_out", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 61 output = Output("output", "TENSOR_FLOAT32", "{%d, %d}" % (n_batch, n_output)) 646 output_state_out: [ 0 for x in range(n_batch * n_output) ],
|