<
   ir_version: 10,
   opset_import: ["" : 23],
   producer_name: "pytorch"
>
main_graph (int32[batch,77] text) => (float[batch,1152] text_embedding) 
   <
      float[batch,77,1024] add_1011
      float[batch,77,1024] add_1084
      float[batch,77,1024] add_1113
      float[batch,77,1024] add_1186
      float[batch,77,1024] add_1215
      float[batch,77,1024] add_1288
      float[batch,77,1024] add_1317
      float[batch,77,1024] add_1390
      float[batch,77,1024] add_1419
      float[batch,77,1024] add_1492
      float[batch,77,1024] add_1521
      float[batch,77,1024] add_1594
      float[batch,77,1024] add_1623
      float[batch,77,1024] add_166
      float[batch,77,1024] add_1696
      float[batch,77,1024] add_1725
      float[batch,77,1024] add_1798
      float[batch,77,1024] add_1827
      float[batch,77,1024] add_1900
      float[batch,77,1024] add_1929
      float[batch,77,1024] add_195
      float[batch,77,1024] add_2002
      float[batch,77,1024] add_2031
      float[batch,77,1024] add_2104
      float[batch,77,1024] add_2133
      float[batch,77,1024] add_2206
      float[batch,77,1024] add_2235
      float[batch,77,1024] add_2308
      float[batch,77,1024] add_2337
      float[batch,77,1024] add_2410
      float[batch,77,1024] add_2439
      float[batch,1,1024] add_2439_pooled
      float[batch,1,1024] add_2512
      float[batch,1,1024] add_2541
      float[batch,77,1024] add_268
      float[batch,77,1024] add_297
      float[batch,77,1024] add_370
      float[batch,77,1024] add_399
      float[batch,77,1024] add_472
      float[batch,77,1024] add_501
      float[batch,77,1024] add_53
      float[batch,77,1024] add_574
      float[batch,77,1024] add_603
      float[batch,77,1024] add_676
      float[batch,77,1024] add_705
      float[batch,77,1024] add_778
      float[batch,77,1024] add_807
      float[batch,77,1024] add_880
      float[batch,77,1024] add_909
      float[batch,77,1024] add_982
      float[batch,1,1,77] bitwise_and_1_f
      float[batch,1] clamp_min
      float[batch,77,1024] layer_norm
      float[batch,77,1024] layer_norm_1
      float[batch,77,1024] layer_norm_10
      float[batch,77,1024] layer_norm_11
      float[batch,77,1024] layer_norm_12
      float[batch,77,1024] layer_norm_13
      float[batch,77,1024] layer_norm_14
      float[batch,77,1024] layer_norm_15
      float[batch,77,1024] layer_norm_16
      float[batch,77,1024] layer_norm_17
      float[batch,77,1024] layer_norm_18
      float[batch,77,1024] layer_norm_19
      float[batch,77,1024] layer_norm_2
      float[batch,77,1024] layer_norm_20
      float[batch,77,1024] layer_norm_21
      float[batch,77,1024] layer_norm_22
      float[batch,77,1024] layer_norm_23
      float[batch,77,1024] layer_norm_24
      float[batch,77,1024] layer_norm_25
      float[batch,77,1024] layer_norm_26
      float[batch,77,1024] layer_norm_27
      float[batch,77,1024] layer_norm_28
      float[batch,77,1024] layer_norm_29
      float[batch,77,1024] layer_norm_3
      float[batch,77,1024] layer_norm_30
      float[batch,77,1024] layer_norm_31
      float[batch,77,1024] layer_norm_32
      float[batch,77,1024] layer_norm_33
      float[batch,77,1024] layer_norm_34
      float[batch,77,1024] layer_norm_35
      float[batch,77,1024] layer_norm_36
      float[batch,77,1024] layer_norm_37
      float[batch,77,1024] layer_norm_38
      float[batch,77,1024] layer_norm_39
      float[batch,77,1024] layer_norm_4
      float[batch,77,1024] layer_norm_40
      float[batch,77,1024] layer_norm_41
      float[batch,77,1024] layer_norm_42
      float[batch,77,1024] layer_norm_43
      float[batch,77,1024] layer_norm_44
      float[batch,77,1024] layer_norm_45
      float[batch,77,1024] layer_norm_46
      float[batch,1,1024] layer_norm_47
      float[batch,77,1024] layer_norm_5
      float[batch,77,1024] layer_norm_6
      float[batch,77,1024] layer_norm_7
      float[batch,77,1024] layer_norm_8
      float[batch,77,1024] layer_norm_9
      float[batch,1] linalg_vector_norm
      float[batch,77,1024] linear
      float[batch,77,1024] linear_1
      float[batch,77,8192] linear_10
      float[batch,77,8192] linear_100
      float[batch,77,1024] linear_101
      float[batch,77,1024] linear_102
      float[batch,77,1024] linear_103
      float[batch,77,1024] linear_105
      float[batch,77,8192] linear_106
      float[batch,77,1024] linear_107
      float[batch,77,1024] linear_108
      float[batch,77,1024] linear_109
      float[batch,77,1024] linear_11
      float[batch,77,1024] linear_111
      float[batch,77,8192] linear_112
      float[batch,77,1024] linear_113
      float[batch,77,1024] linear_114
      float[batch,77,1024] linear_115
      float[batch,77,1024] linear_117
      float[batch,77,8192] linear_118
      float[batch,77,1024] linear_119
      float[batch,77,1024] linear_12
      float[batch,77,1024] linear_120
      float[batch,77,1024] linear_121
      float[batch,77,1024] linear_123
      float[batch,77,8192] linear_124
      float[batch,77,1024] linear_125
      float[batch,77,1024] linear_126
      float[batch,77,1024] linear_127
      float[batch,77,1024] linear_129
      float[batch,77,1024] linear_13
      float[batch,77,8192] linear_130
      float[batch,77,1024] linear_131
      float[batch,77,1024] linear_132
      float[batch,77,1024] linear_133
      float[batch,77,1024] linear_135
      float[batch,77,8192] linear_136
      float[batch,77,1024] linear_137
      float[batch,77,1024] linear_138
      float[batch,77,1024] linear_139
      float[batch,1,1024] linear_141
      float[batch,1,8192] linear_142
      float[batch,1,1024] linear_143
      float[batch,1152] linear_144
      float[batch,77,1024] linear_15
      float[batch,77,8192] linear_16
      float[batch,77,1024] linear_17
      float[batch,77,1024] linear_18
      float[batch,77,1024] linear_19
      float[batch,77,1024] linear_21
      float[batch,77,8192] linear_22
      float[batch,77,1024] linear_23
      float[batch,77,1024] linear_24
      float[batch,77,1024] linear_25
      float[batch,77,1024] linear_27
      float[batch,77,8192] linear_28
      float[batch,77,1024] linear_29
      float[batch,77,1024] linear_3
      float[batch,77,1024] linear_30
      float[batch,77,1024] linear_31
      float[batch,77,1024] linear_33
      float[batch,77,8192] linear_34
      float[batch,77,1024] linear_35
      float[batch,77,1024] linear_36
      float[batch,77,1024] linear_37
      float[batch,77,1024] linear_39
      float[batch,77,8192] linear_4
      float[batch,77,8192] linear_40
      float[batch,77,1024] linear_41
      float[batch,77,1024] linear_42
      float[batch,77,1024] linear_43
      float[batch,77,1024] linear_45
      float[batch,77,8192] linear_46
      float[batch,77,1024] linear_47
      float[batch,77,1024] linear_48
      float[batch,77,1024] linear_49
      float[batch,77,1024] linear_5
      float[batch,77,1024] linear_51
      float[batch,77,8192] linear_52
      float[batch,77,1024] linear_53
      float[batch,77,1024] linear_54
      float[batch,77,1024] linear_55
      float[batch,77,1024] linear_57
      float[batch,77,8192] linear_58
      float[batch,77,1024] linear_59
      float[batch,77,1024] linear_6
      float[batch,77,1024] linear_60
      float[batch,77,1024] linear_61
      float[batch,77,1024] linear_63
      float[batch,77,8192] linear_64
      float[batch,77,1024] linear_65
      float[batch,77,1024] linear_66
      float[batch,77,1024] linear_67
      float[batch,77,1024] linear_69
      float[batch,77,1024] linear_7
      float[batch,77,8192] linear_70
      float[batch,77,1024] linear_71
      float[batch,77,1024] linear_72
      float[batch,77,1024] linear_73
      float[batch,77,1024] linear_75
      float[batch,77,8192] linear_76
      float[batch,77,1024] linear_77
      float[batch,77,1024] linear_78
      float[batch,77,1024] linear_79
      float[batch,77,1024] linear_81
      float[batch,77,8192] linear_82
      float[batch,77,1024] linear_83
      float[batch,77,1024] linear_84
      float[batch,77,1024] linear_85
      float[batch,77,1024] linear_87
      float[batch,77,8192] linear_88
      float[batch,77,1024] linear_89
      float[batch,77,1024] linear_9
      float[batch,77,1024] linear_90
      float[batch,77,1024] linear_91
      float[batch,77,1024] linear_93
      float[batch,77,8192] linear_94
      float[batch,77,1024] linear_95
      float[batch,77,1024] linear_96
      float[batch,77,1024] linear_97
      float[batch,77,1024] linear_99
      float[batch,77,1024] mul_5
      float[batch,77,8192] relu
      float[batch,77,8192] relu_1
      float[batch,77,8192] relu_10
      float[batch,77,8192] relu_11
      float[batch,77,8192] relu_12
      float[batch,77,8192] relu_13
      float[batch,77,8192] relu_14
      float[batch,77,8192] relu_15
      float[batch,77,8192] relu_16
      float[batch,77,8192] relu_17
      float[batch,77,8192] relu_18
      float[batch,77,8192] relu_19
      float[batch,77,8192] relu_2
      float[batch,77,8192] relu_20
      float[batch,77,8192] relu_21
      float[batch,77,8192] relu_22
      float[batch,1,8192] relu_23
      float[batch,77,8192] relu_3
      float[batch,77,8192] relu_4
      float[batch,77,8192] relu_5
      float[batch,77,8192] relu_6
      float[batch,77,8192] relu_7
      float[batch,77,8192] relu_8
      float[batch,77,8192] relu_9
      float[batch,77,1024] scaled_dot_product_attention
      float[batch,77,1024] scaled_dot_product_attention_1
      float[batch,77,1024] scaled_dot_product_attention_10
      float[batch,77,1024] scaled_dot_product_attention_11
      float[batch,77,1024] scaled_dot_product_attention_12
      float[batch,77,1024] scaled_dot_product_attention_13
      float[batch,77,1024] scaled_dot_product_attention_14
      float[batch,77,1024] scaled_dot_product_attention_15
      float[batch,77,1024] scaled_dot_product_attention_16
      float[batch,77,1024] scaled_dot_product_attention_17
      float[batch,77,1024] scaled_dot_product_attention_18
      float[batch,77,1024] scaled_dot_product_attention_19
      float[batch,77,1024] scaled_dot_product_attention_2
      float[batch,77,1024] scaled_dot_product_attention_20
      float[batch,77,1024] scaled_dot_product_attention_21
      float[batch,77,1024] scaled_dot_product_attention_22
      float[batch,77,1024] scaled_dot_product_attention_23
      float[batch,1,1024] scaled_dot_product_attention_23_pooled
      float[batch,77,1024] scaled_dot_product_attention_3
      float[batch,77,1024] scaled_dot_product_attention_4
      float[batch,77,1024] scaled_dot_product_attention_5
      float[batch,77,1024] scaled_dot_product_attention_6
      float[batch,77,1024] scaled_dot_product_attention_7
      float[batch,77,1024] scaled_dot_product_attention_8
      float[batch,77,1024] scaled_dot_product_attention_9
      float[batch,1024] select
      float[batch,77] text_keep
      float[batch,77,1024] val_150
      float[batch,77,1024] val_151
      float[batch,77,1024] val_152
      float[batch,77,1024] val_153
      float[batch,77,8192] val_154
      float[batch,77,1024] val_155
      float[batch,77,1024] val_156
      float[batch,77,1024] val_157
      float[batch,77,1024] val_158
      float[batch,77,1024] val_159
      float[batch,77,8192] val_160
      float[batch,77,1024] val_161
      float[batch,77,1024] val_162
      float[batch,77,1024] val_163
      float[batch,77,1024] val_164
      float[batch,77,1024] val_165
      float[batch,77,8192] val_166
      float[batch,77,1024] val_167
      float[batch,77,1024] val_168
      float[batch,77,1024] val_169
      float[batch,77,1024] val_170
      float[batch,77,1024] val_171
      float[batch,77,8192] val_172
      float[batch,77,1024] val_173
      float[batch,77,1024] val_174
      float[batch,77,1024] val_175
      float[batch,77,1024] val_176
      float[batch,77,1024] val_177
      float[batch,77,8192] val_178
      float[batch,77,1024] val_179
      float[batch,77,1024] val_180
      float[batch,77,1024] val_181
      float[batch,77,1024] val_182
      float[batch,77,1024] val_183
      float[batch,77,8192] val_184
      float[batch,77,1024] val_185
      float[batch,77,1024] val_186
      float[batch,77,1024] val_187
      float[batch,77,1024] val_188
      float[batch,77,1024] val_189
      float[batch,77,8192] val_190
      float[batch,77,1024] val_191
      float[batch,77,1024] val_192
      float[batch,77,1024] val_193
      float[batch,77,1024] val_194
      float[batch,77,1024] val_195
      float[batch,77,8192] val_196
      float[batch,77,1024] val_197
      float[batch,77,1024] val_198
      float[batch,77,1024] val_199
      float[batch,77,1024] val_200
      float[batch,77,1024] val_201
      float[batch,77,8192] val_202
      float[batch,77,1024] val_203
      float[batch,77,1024] val_204
      float[batch,77,1024] val_205
      float[batch,77,1024] val_206
      float[batch,77,1024] val_207
      float[batch,77,8192] val_208
      float[batch,77,1024] val_209
      float[batch,77,1024] val_210
      float[batch,77,1024] val_211
      float[batch,77,1024] val_212
      float[batch,77,1024] val_213
      float[batch,77,8192] val_214
      float[batch,77,1024] val_215
      float[batch,77,1024] val_216
      float[batch,77,1024] val_217
      float[batch,77,1024] val_218
      float[batch,77,1024] val_219
      float[batch,77,8192] val_220
      float[batch,77,1024] val_221
      float[batch,77,1024] val_222
      float[batch,77,1024] val_223
      float[batch,77,1024] val_224
      float[batch,77,1024] val_225
      float[batch,77,8192] val_226
      float[batch,77,1024] val_227
      float[batch,77,1024] val_228
      float[batch,77,1024] val_229
      float[batch,77,1024] val_230
      float[batch,77,1024] val_231
      float[batch,77,8192] val_232
      float[batch,77,1024] val_233
      float[batch,77,1024] val_234
      float[batch,77,1024] val_235
      float[batch,77,1024] val_236
      float[batch,77,1024] val_237
      float[batch,77,8192] val_238
      float[batch,77,1024] val_239
      float[batch,77,1024] val_240
      float[batch,77,1024] val_241
      float[batch,77,1024] val_242
      float[batch,77,1024] val_243
      float[batch,77,8192] val_244
      float[batch,77,1024] val_245
      float[batch,77,1024] val_246
      float[batch,77,1024] val_247
      float[batch,77,1024] val_248
      float[batch,77,1024] val_249
      float[batch,77,8192] val_250
      float[batch,77,1024] val_251
      float[batch,77,1024] val_252
      float[batch,77,1024] val_253
      float[batch,77,1024] val_254
      float[batch,77,1024] val_255
      float[batch,77,8192] val_256
      float[batch,77,1024] val_257
      float[batch,77,1024] val_258
      float[batch,77,1024] val_259
      float[batch,77,1024] val_260
      float[batch,77,1024] val_261
      float[batch,77,8192] val_262
      float[batch,77,1024] val_263
      float[batch,77,1024] val_264
      float[batch,77,1024] val_265
      float[batch,77,1024] val_266
      float[batch,77,1024] val_267
      float[batch,77,8192] val_268
      float[batch,77,1024] val_269
      float[batch,77,1024] val_270
      float[batch,77,1024] val_271
      float[batch,77,1024] val_272
      float[batch,77,1024] val_273
      float[batch,77,8192] val_274
      float[batch,77,1024] val_275
      float[batch,77,1024] val_276
      float[batch,77,1024] val_277
      float[batch,77,1024] val_278
      float[batch,77,1024] val_279
      float[batch,77,8192] val_280
      float[batch,77,1024] val_281
      float[batch,77,1024] val_282
      float[batch,77,1024] val_283
      float[batch,77,1024] val_284
      float[batch,77,1024] val_285
      float[batch,77,8192] val_286
      float[batch,77,1024] val_287
      float[batch,77,1024] val_288
      float[batch,77,1024] val_289
      float[batch,77,1024] val_290
      float[batch,1,1024] val_291
      float[batch,1,8192] val_292
      float[batch,1,1024] val_293
      float[batch,1024] val_294
      float[batch,1,1,77] val_53_f
      float[batch,1,1,77] val_53_f_bias
      float[batch,1,77,77] val_53_f_mask
      float[1,77,1024] view_1
   >
{
   val_149 = Gather <axis: int = 0> ("text.transformer.embed_tokens.weight_fp16", text)
   mul_5 = Cast <to: int = 1> (val_149)
   view_1 = Reshape <allowzero: int = 1> (index_select, view_1_target)
   add_53 = Add (mul_5, view_1)
   [node_layer_norm] layer_norm = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_53, "text.transformer.layers.0.self_attn_layer_norm.weight", "text.transformer.layers.0.self_attn_layer_norm.bias")
   val_150 = MatMul (layer_norm, val_5)
   [node_linear] linear = Add (val_150, "text.transformer.layers.0.self_attn.q_proj.bias")
   val_151 = MatMul (layer_norm, val_6)
   linear_1 = Add (val_151, "text.transformer.layers.0.self_attn.k_proj.bias")
   val_152 = MatMul (layer_norm, val_7)
   [pad_keep] text_keep = Gather <axis: int = 0> (text_pad_keep, text)
   [val_53_f] val_53_f = Unsqueeze (text_keep, text_row_axes)
   [bitwise_and_1_f] bitwise_and_1_f = Sub (val_53_f, text_one)
   val_53_f_bias = Mul (bitwise_and_1_f, text_scale)
   val_53_f_mask = Add (val_53_f_bias, text_q_axis)
   [node_scaled_dot_product_attention] scaled_dot_product_attention = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear, linear_1, val_152, val_53_f_mask)
   val_153 = MatMul (scaled_dot_product_attention, val_8)
   linear_3 = Add (val_153, "text.transformer.layers.0.self_attn.out_proj.bias")
   add_166 = Add (add_53, linear_3)
   layer_norm_1 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_166, "text.transformer.layers.0.final_layer_norm.weight", "text.transformer.layers.0.final_layer_norm.bias")
   val_154 = MatMul (layer_norm_1, val_9)
   linear_4 = Add (val_154, "text.transformer.layers.0.fc1.bias")
   [node_relu] relu = Relu (linear_4)
   val_155 = MatMul (relu, val_10)
   linear_5 = Add (val_155, "text.transformer.layers.0.fc2.bias")
   add_195 = Add (add_166, linear_5)
   layer_norm_2 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_195, "text.transformer.layers.1.self_attn_layer_norm.weight", "text.transformer.layers.1.self_attn_layer_norm.bias")
   val_156 = MatMul (layer_norm_2, val_11)
   linear_6 = Add (val_156, "text.transformer.layers.1.self_attn.q_proj.bias")
   val_157 = MatMul (layer_norm_2, val_12)
   linear_7 = Add (val_157, "text.transformer.layers.1.self_attn.k_proj.bias")
   val_158 = MatMul (layer_norm_2, val_13)
   scaled_dot_product_attention_1 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_6, linear_7, val_158, val_53_f_mask)
   val_159 = MatMul (scaled_dot_product_attention_1, val_14)
   linear_9 = Add (val_159, "text.transformer.layers.1.self_attn.out_proj.bias")
   add_268 = Add (add_195, linear_9)
   layer_norm_3 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_268, "text.transformer.layers.1.final_layer_norm.weight", "text.transformer.layers.1.final_layer_norm.bias")
   val_160 = MatMul (layer_norm_3, val_15)
   linear_10 = Add (val_160, "text.transformer.layers.1.fc1.bias")
   relu_1 = Relu (linear_10)
   val_161 = MatMul (relu_1, val_16)
   linear_11 = Add (val_161, "text.transformer.layers.1.fc2.bias")
   add_297 = Add (add_268, linear_11)
   layer_norm_4 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_297, "text.transformer.layers.2.self_attn_layer_norm.weight", "text.transformer.layers.2.self_attn_layer_norm.bias")
   val_162 = MatMul (layer_norm_4, val_17)
   linear_12 = Add (val_162, "text.transformer.layers.2.self_attn.q_proj.bias")
   val_163 = MatMul (layer_norm_4, val_18)
   linear_13 = Add (val_163, "text.transformer.layers.2.self_attn.k_proj.bias")
   val_164 = MatMul (layer_norm_4, val_19)
   scaled_dot_product_attention_2 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_12, linear_13, val_164, val_53_f_mask)
   val_165 = MatMul (scaled_dot_product_attention_2, val_20)
   linear_15 = Add (val_165, "text.transformer.layers.2.self_attn.out_proj.bias")
   add_370 = Add (add_297, linear_15)
   layer_norm_5 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_370, "text.transformer.layers.2.final_layer_norm.weight", "text.transformer.layers.2.final_layer_norm.bias")
   val_166 = MatMul (layer_norm_5, val_21)
   linear_16 = Add (val_166, "text.transformer.layers.2.fc1.bias")
   relu_2 = Relu (linear_16)
   val_167 = MatMul (relu_2, val_22)
   linear_17 = Add (val_167, "text.transformer.layers.2.fc2.bias")
   add_399 = Add (add_370, linear_17)
   layer_norm_6 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_399, "text.transformer.layers.3.self_attn_layer_norm.weight", "text.transformer.layers.3.self_attn_layer_norm.bias")
   val_168 = MatMul (layer_norm_6, val_23)
   linear_18 = Add (val_168, "text.transformer.layers.3.self_attn.q_proj.bias")
   val_169 = MatMul (layer_norm_6, val_24)
   linear_19 = Add (val_169, "text.transformer.layers.3.self_attn.k_proj.bias")
   val_170 = MatMul (layer_norm_6, val_25)
   scaled_dot_product_attention_3 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_18, linear_19, val_170, val_53_f_mask)
   val_171 = MatMul (scaled_dot_product_attention_3, val_26)
   linear_21 = Add (val_171, "text.transformer.layers.3.self_attn.out_proj.bias")
   add_472 = Add (add_399, linear_21)
   layer_norm_7 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_472, "text.transformer.layers.3.final_layer_norm.weight", "text.transformer.layers.3.final_layer_norm.bias")
   val_172 = MatMul (layer_norm_7, val_27)
   linear_22 = Add (val_172, "text.transformer.layers.3.fc1.bias")
   relu_3 = Relu (linear_22)
   val_173 = MatMul (relu_3, val_28)
   linear_23 = Add (val_173, "text.transformer.layers.3.fc2.bias")
   add_501 = Add (add_472, linear_23)
   layer_norm_8 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_501, "text.transformer.layers.4.self_attn_layer_norm.weight", "text.transformer.layers.4.self_attn_layer_norm.bias")
   val_174 = MatMul (layer_norm_8, val_29)
   linear_24 = Add (val_174, "text.transformer.layers.4.self_attn.q_proj.bias")
   val_175 = MatMul (layer_norm_8, val_30)
   linear_25 = Add (val_175, "text.transformer.layers.4.self_attn.k_proj.bias")
   val_176 = MatMul (layer_norm_8, val_31)
   scaled_dot_product_attention_4 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_24, linear_25, val_176, val_53_f_mask)
   val_177 = MatMul (scaled_dot_product_attention_4, val_32)
   linear_27 = Add (val_177, "text.transformer.layers.4.self_attn.out_proj.bias")
   add_574 = Add (add_501, linear_27)
   layer_norm_9 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_574, "text.transformer.layers.4.final_layer_norm.weight", "text.transformer.layers.4.final_layer_norm.bias")
   val_178 = MatMul (layer_norm_9, val_33)
   linear_28 = Add (val_178, "text.transformer.layers.4.fc1.bias")
   relu_4 = Relu (linear_28)
   val_179 = MatMul (relu_4, val_34)
   linear_29 = Add (val_179, "text.transformer.layers.4.fc2.bias")
   add_603 = Add (add_574, linear_29)
   layer_norm_10 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_603, "text.transformer.layers.5.self_attn_layer_norm.weight", "text.transformer.layers.5.self_attn_layer_norm.bias")
   val_180 = MatMul (layer_norm_10, val_35)
   linear_30 = Add (val_180, "text.transformer.layers.5.self_attn.q_proj.bias")
   val_181 = MatMul (layer_norm_10, val_36)
   linear_31 = Add (val_181, "text.transformer.layers.5.self_attn.k_proj.bias")
   val_182 = MatMul (layer_norm_10, val_37)
   scaled_dot_product_attention_5 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_30, linear_31, val_182, val_53_f_mask)
   val_183 = MatMul (scaled_dot_product_attention_5, val_38)
   linear_33 = Add (val_183, "text.transformer.layers.5.self_attn.out_proj.bias")
   add_676 = Add (add_603, linear_33)
   layer_norm_11 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_676, "text.transformer.layers.5.final_layer_norm.weight", "text.transformer.layers.5.final_layer_norm.bias")
   val_184 = MatMul (layer_norm_11, val_39)
   linear_34 = Add (val_184, "text.transformer.layers.5.fc1.bias")
   relu_5 = Relu (linear_34)
   val_185 = MatMul (relu_5, val_40)
   linear_35 = Add (val_185, "text.transformer.layers.5.fc2.bias")
   add_705 = Add (add_676, linear_35)
   layer_norm_12 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_705, "text.transformer.layers.6.self_attn_layer_norm.weight", "text.transformer.layers.6.self_attn_layer_norm.bias")
   val_186 = MatMul (layer_norm_12, val_41)
   linear_36 = Add (val_186, "text.transformer.layers.6.self_attn.q_proj.bias")
   val_187 = MatMul (layer_norm_12, val_42)
   linear_37 = Add (val_187, "text.transformer.layers.6.self_attn.k_proj.bias")
   val_188 = MatMul (layer_norm_12, val_43)
   scaled_dot_product_attention_6 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_36, linear_37, val_188, val_53_f_mask)
   val_189 = MatMul (scaled_dot_product_attention_6, val_44)
   linear_39 = Add (val_189, "text.transformer.layers.6.self_attn.out_proj.bias")
   add_778 = Add (add_705, linear_39)
   layer_norm_13 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_778, "text.transformer.layers.6.final_layer_norm.weight", "text.transformer.layers.6.final_layer_norm.bias")
   val_190 = MatMul (layer_norm_13, val_45)
   linear_40 = Add (val_190, "text.transformer.layers.6.fc1.bias")
   relu_6 = Relu (linear_40)
   val_191 = MatMul (relu_6, val_46)
   linear_41 = Add (val_191, "text.transformer.layers.6.fc2.bias")
   add_807 = Add (add_778, linear_41)
   layer_norm_14 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_807, "text.transformer.layers.7.self_attn_layer_norm.weight", "text.transformer.layers.7.self_attn_layer_norm.bias")
   val_192 = MatMul (layer_norm_14, val_47)
   linear_42 = Add (val_192, "text.transformer.layers.7.self_attn.q_proj.bias")
   val_193 = MatMul (layer_norm_14, val_48)
   linear_43 = Add (val_193, "text.transformer.layers.7.self_attn.k_proj.bias")
   val_194 = MatMul (layer_norm_14, val_49)
   scaled_dot_product_attention_7 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_42, linear_43, val_194, val_53_f_mask)
   val_195 = MatMul (scaled_dot_product_attention_7, val_50)
   linear_45 = Add (val_195, "text.transformer.layers.7.self_attn.out_proj.bias")
   add_880 = Add (add_807, linear_45)
   layer_norm_15 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_880, "text.transformer.layers.7.final_layer_norm.weight", "text.transformer.layers.7.final_layer_norm.bias")
   val_196 = MatMul (layer_norm_15, val_51)
   linear_46 = Add (val_196, "text.transformer.layers.7.fc1.bias")
   relu_7 = Relu (linear_46)
   val_197 = MatMul (relu_7, val_52)
   linear_47 = Add (val_197, "text.transformer.layers.7.fc2.bias")
   add_909 = Add (add_880, linear_47)
   layer_norm_16 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_909, "text.transformer.layers.8.self_attn_layer_norm.weight", "text.transformer.layers.8.self_attn_layer_norm.bias")
   val_198 = MatMul (layer_norm_16, val_53)
   linear_48 = Add (val_198, "text.transformer.layers.8.self_attn.q_proj.bias")
   val_199 = MatMul (layer_norm_16, val_54)
   linear_49 = Add (val_199, "text.transformer.layers.8.self_attn.k_proj.bias")
   val_200 = MatMul (layer_norm_16, val_55)
   scaled_dot_product_attention_8 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_48, linear_49, val_200, val_53_f_mask)
   val_201 = MatMul (scaled_dot_product_attention_8, val_56)
   linear_51 = Add (val_201, "text.transformer.layers.8.self_attn.out_proj.bias")
   add_982 = Add (add_909, linear_51)
   layer_norm_17 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_982, "text.transformer.layers.8.final_layer_norm.weight", "text.transformer.layers.8.final_layer_norm.bias")
   val_202 = MatMul (layer_norm_17, val_57)
   linear_52 = Add (val_202, "text.transformer.layers.8.fc1.bias")
   relu_8 = Relu (linear_52)
   val_203 = MatMul (relu_8, val_58)
   linear_53 = Add (val_203, "text.transformer.layers.8.fc2.bias")
   add_1011 = Add (add_982, linear_53)
   layer_norm_18 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1011, "text.transformer.layers.9.self_attn_layer_norm.weight", "text.transformer.layers.9.self_attn_layer_norm.bias")
   val_204 = MatMul (layer_norm_18, val_59)
   linear_54 = Add (val_204, "text.transformer.layers.9.self_attn.q_proj.bias")
   val_205 = MatMul (layer_norm_18, val_60)
   linear_55 = Add (val_205, "text.transformer.layers.9.self_attn.k_proj.bias")
   val_206 = MatMul (layer_norm_18, val_61)
   scaled_dot_product_attention_9 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_54, linear_55, val_206, val_53_f_mask)
   val_207 = MatMul (scaled_dot_product_attention_9, val_62)
   linear_57 = Add (val_207, "text.transformer.layers.9.self_attn.out_proj.bias")
   add_1084 = Add (add_1011, linear_57)
   layer_norm_19 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1084, "text.transformer.layers.9.final_layer_norm.weight", "text.transformer.layers.9.final_layer_norm.bias")
   val_208 = MatMul (layer_norm_19, val_63)
   linear_58 = Add (val_208, "text.transformer.layers.9.fc1.bias")
   relu_9 = Relu (linear_58)
   val_209 = MatMul (relu_9, val_64)
   linear_59 = Add (val_209, "text.transformer.layers.9.fc2.bias")
   add_1113 = Add (add_1084, linear_59)
   layer_norm_20 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1113, "text.transformer.layers.10.self_attn_layer_norm.weight", "text.transformer.layers.10.self_attn_layer_norm.bias")
   val_210 = MatMul (layer_norm_20, val_65)
   linear_60 = Add (val_210, "text.transformer.layers.10.self_attn.q_proj.bias")
   val_211 = MatMul (layer_norm_20, val_66)
   linear_61 = Add (val_211, "text.transformer.layers.10.self_attn.k_proj.bias")
   val_212 = MatMul (layer_norm_20, val_67)
   scaled_dot_product_attention_10 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_60, linear_61, val_212, val_53_f_mask)
   val_213 = MatMul (scaled_dot_product_attention_10, val_68)
   linear_63 = Add (val_213, "text.transformer.layers.10.self_attn.out_proj.bias")
   add_1186 = Add (add_1113, linear_63)
   layer_norm_21 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1186, "text.transformer.layers.10.final_layer_norm.weight", "text.transformer.layers.10.final_layer_norm.bias")
   val_214 = MatMul (layer_norm_21, val_69)
   linear_64 = Add (val_214, "text.transformer.layers.10.fc1.bias")
   relu_10 = Relu (linear_64)
   val_215 = MatMul (relu_10, val_70)
   linear_65 = Add (val_215, "text.transformer.layers.10.fc2.bias")
   add_1215 = Add (add_1186, linear_65)
   layer_norm_22 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1215, "text.transformer.layers.11.self_attn_layer_norm.weight", "text.transformer.layers.11.self_attn_layer_norm.bias")
   val_216 = MatMul (layer_norm_22, val_71)
   linear_66 = Add (val_216, "text.transformer.layers.11.self_attn.q_proj.bias")
   val_217 = MatMul (layer_norm_22, val_72)
   linear_67 = Add (val_217, "text.transformer.layers.11.self_attn.k_proj.bias")
   val_218 = MatMul (layer_norm_22, val_73)
   scaled_dot_product_attention_11 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_66, linear_67, val_218, val_53_f_mask)
   val_219 = MatMul (scaled_dot_product_attention_11, val_74)
   linear_69 = Add (val_219, "text.transformer.layers.11.self_attn.out_proj.bias")
   add_1288 = Add (add_1215, linear_69)
   layer_norm_23 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1288, "text.transformer.layers.11.final_layer_norm.weight", "text.transformer.layers.11.final_layer_norm.bias")
   val_220 = MatMul (layer_norm_23, val_75)
   linear_70 = Add (val_220, "text.transformer.layers.11.fc1.bias")
   relu_11 = Relu (linear_70)
   val_221 = MatMul (relu_11, val_76)
   linear_71 = Add (val_221, "text.transformer.layers.11.fc2.bias")
   add_1317 = Add (add_1288, linear_71)
   layer_norm_24 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1317, "text.transformer.layers.12.self_attn_layer_norm.weight", "text.transformer.layers.12.self_attn_layer_norm.bias")
   val_222 = MatMul (layer_norm_24, val_77)
   linear_72 = Add (val_222, "text.transformer.layers.12.self_attn.q_proj.bias")
   val_223 = MatMul (layer_norm_24, val_78)
   linear_73 = Add (val_223, "text.transformer.layers.12.self_attn.k_proj.bias")
   val_224 = MatMul (layer_norm_24, val_79)
   scaled_dot_product_attention_12 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_72, linear_73, val_224, val_53_f_mask)
   val_225 = MatMul (scaled_dot_product_attention_12, val_80)
   linear_75 = Add (val_225, "text.transformer.layers.12.self_attn.out_proj.bias")
   add_1390 = Add (add_1317, linear_75)
   layer_norm_25 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1390, "text.transformer.layers.12.final_layer_norm.weight", "text.transformer.layers.12.final_layer_norm.bias")
   val_226 = MatMul (layer_norm_25, val_81)
   linear_76 = Add (val_226, "text.transformer.layers.12.fc1.bias")
   relu_12 = Relu (linear_76)
   val_227 = MatMul (relu_12, val_82)
   linear_77 = Add (val_227, "text.transformer.layers.12.fc2.bias")
   add_1419 = Add (add_1390, linear_77)
   layer_norm_26 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1419, "text.transformer.layers.13.self_attn_layer_norm.weight", "text.transformer.layers.13.self_attn_layer_norm.bias")
   val_228 = MatMul (layer_norm_26, val_83)
   linear_78 = Add (val_228, "text.transformer.layers.13.self_attn.q_proj.bias")
   val_229 = MatMul (layer_norm_26, val_84)
   linear_79 = Add (val_229, "text.transformer.layers.13.self_attn.k_proj.bias")
   val_230 = MatMul (layer_norm_26, val_85)
   scaled_dot_product_attention_13 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_78, linear_79, val_230, val_53_f_mask)
   val_231 = MatMul (scaled_dot_product_attention_13, val_86)
   linear_81 = Add (val_231, "text.transformer.layers.13.self_attn.out_proj.bias")
   add_1492 = Add (add_1419, linear_81)
   layer_norm_27 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1492, "text.transformer.layers.13.final_layer_norm.weight", "text.transformer.layers.13.final_layer_norm.bias")
   val_232 = MatMul (layer_norm_27, val_87)
   linear_82 = Add (val_232, "text.transformer.layers.13.fc1.bias")
   relu_13 = Relu (linear_82)
   val_233 = MatMul (relu_13, val_88)
   linear_83 = Add (val_233, "text.transformer.layers.13.fc2.bias")
   add_1521 = Add (add_1492, linear_83)
   layer_norm_28 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1521, "text.transformer.layers.14.self_attn_layer_norm.weight", "text.transformer.layers.14.self_attn_layer_norm.bias")
   val_234 = MatMul (layer_norm_28, val_89)
   linear_84 = Add (val_234, "text.transformer.layers.14.self_attn.q_proj.bias")
   val_235 = MatMul (layer_norm_28, val_90)
   linear_85 = Add (val_235, "text.transformer.layers.14.self_attn.k_proj.bias")
   val_236 = MatMul (layer_norm_28, val_91)
   scaled_dot_product_attention_14 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_84, linear_85, val_236, val_53_f_mask)
   val_237 = MatMul (scaled_dot_product_attention_14, val_92)
   linear_87 = Add (val_237, "text.transformer.layers.14.self_attn.out_proj.bias")
   add_1594 = Add (add_1521, linear_87)
   layer_norm_29 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1594, "text.transformer.layers.14.final_layer_norm.weight", "text.transformer.layers.14.final_layer_norm.bias")
   val_238 = MatMul (layer_norm_29, val_93)
   linear_88 = Add (val_238, "text.transformer.layers.14.fc1.bias")
   relu_14 = Relu (linear_88)
   val_239 = MatMul (relu_14, val_94)
   linear_89 = Add (val_239, "text.transformer.layers.14.fc2.bias")
   add_1623 = Add (add_1594, linear_89)
   layer_norm_30 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1623, "text.transformer.layers.15.self_attn_layer_norm.weight", "text.transformer.layers.15.self_attn_layer_norm.bias")
   val_240 = MatMul (layer_norm_30, val_95)
   linear_90 = Add (val_240, "text.transformer.layers.15.self_attn.q_proj.bias")
   val_241 = MatMul (layer_norm_30, val_96)
   linear_91 = Add (val_241, "text.transformer.layers.15.self_attn.k_proj.bias")
   val_242 = MatMul (layer_norm_30, val_97)
   scaled_dot_product_attention_15 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_90, linear_91, val_242, val_53_f_mask)
   val_243 = MatMul (scaled_dot_product_attention_15, val_98)
   linear_93 = Add (val_243, "text.transformer.layers.15.self_attn.out_proj.bias")
   add_1696 = Add (add_1623, linear_93)
   layer_norm_31 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1696, "text.transformer.layers.15.final_layer_norm.weight", "text.transformer.layers.15.final_layer_norm.bias")
   val_244 = MatMul (layer_norm_31, val_99)
   linear_94 = Add (val_244, "text.transformer.layers.15.fc1.bias")
   relu_15 = Relu (linear_94)
   val_245 = MatMul (relu_15, val_100)
   linear_95 = Add (val_245, "text.transformer.layers.15.fc2.bias")
   add_1725 = Add (add_1696, linear_95)
   layer_norm_32 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1725, "text.transformer.layers.16.self_attn_layer_norm.weight", "text.transformer.layers.16.self_attn_layer_norm.bias")
   val_246 = MatMul (layer_norm_32, val_101)
   linear_96 = Add (val_246, "text.transformer.layers.16.self_attn.q_proj.bias")
   val_247 = MatMul (layer_norm_32, val_102)
   linear_97 = Add (val_247, "text.transformer.layers.16.self_attn.k_proj.bias")
   val_248 = MatMul (layer_norm_32, val_103)
   scaled_dot_product_attention_16 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_96, linear_97, val_248, val_53_f_mask)
   val_249 = MatMul (scaled_dot_product_attention_16, val_104)
   linear_99 = Add (val_249, "text.transformer.layers.16.self_attn.out_proj.bias")
   add_1798 = Add (add_1725, linear_99)
   layer_norm_33 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1798, "text.transformer.layers.16.final_layer_norm.weight", "text.transformer.layers.16.final_layer_norm.bias")
   val_250 = MatMul (layer_norm_33, val_105)
   linear_100 = Add (val_250, "text.transformer.layers.16.fc1.bias")
   relu_16 = Relu (linear_100)
   val_251 = MatMul (relu_16, val_106)
   linear_101 = Add (val_251, "text.transformer.layers.16.fc2.bias")
   add_1827 = Add (add_1798, linear_101)
   layer_norm_34 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1827, "text.transformer.layers.17.self_attn_layer_norm.weight", "text.transformer.layers.17.self_attn_layer_norm.bias")
   val_252 = MatMul (layer_norm_34, val_107)
   linear_102 = Add (val_252, "text.transformer.layers.17.self_attn.q_proj.bias")
   val_253 = MatMul (layer_norm_34, val_108)
   linear_103 = Add (val_253, "text.transformer.layers.17.self_attn.k_proj.bias")
   val_254 = MatMul (layer_norm_34, val_109)
   scaled_dot_product_attention_17 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_102, linear_103, val_254, val_53_f_mask)
   val_255 = MatMul (scaled_dot_product_attention_17, val_110)
   linear_105 = Add (val_255, "text.transformer.layers.17.self_attn.out_proj.bias")
   add_1900 = Add (add_1827, linear_105)
   layer_norm_35 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1900, "text.transformer.layers.17.final_layer_norm.weight", "text.transformer.layers.17.final_layer_norm.bias")
   val_256 = MatMul (layer_norm_35, val_111)
   linear_106 = Add (val_256, "text.transformer.layers.17.fc1.bias")
   relu_17 = Relu (linear_106)
   val_257 = MatMul (relu_17, val_112)
   linear_107 = Add (val_257, "text.transformer.layers.17.fc2.bias")
   add_1929 = Add (add_1900, linear_107)
   layer_norm_36 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_1929, "text.transformer.layers.18.self_attn_layer_norm.weight", "text.transformer.layers.18.self_attn_layer_norm.bias")
   val_258 = MatMul (layer_norm_36, val_113)
   linear_108 = Add (val_258, "text.transformer.layers.18.self_attn.q_proj.bias")
   val_259 = MatMul (layer_norm_36, val_114)
   linear_109 = Add (val_259, "text.transformer.layers.18.self_attn.k_proj.bias")
   val_260 = MatMul (layer_norm_36, val_115)
   scaled_dot_product_attention_18 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_108, linear_109, val_260, val_53_f_mask)
   val_261 = MatMul (scaled_dot_product_attention_18, val_116)
   linear_111 = Add (val_261, "text.transformer.layers.18.self_attn.out_proj.bias")
   add_2002 = Add (add_1929, linear_111)
   layer_norm_37 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_2002, "text.transformer.layers.18.final_layer_norm.weight", "text.transformer.layers.18.final_layer_norm.bias")
   val_262 = MatMul (layer_norm_37, val_117)
   linear_112 = Add (val_262, "text.transformer.layers.18.fc1.bias")
   relu_18 = Relu (linear_112)
   val_263 = MatMul (relu_18, val_118)
   linear_113 = Add (val_263, "text.transformer.layers.18.fc2.bias")
   add_2031 = Add (add_2002, linear_113)
   layer_norm_38 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_2031, "text.transformer.layers.19.self_attn_layer_norm.weight", "text.transformer.layers.19.self_attn_layer_norm.bias")
   val_264 = MatMul (layer_norm_38, val_119)
   linear_114 = Add (val_264, "text.transformer.layers.19.self_attn.q_proj.bias")
   val_265 = MatMul (layer_norm_38, val_120)
   linear_115 = Add (val_265, "text.transformer.layers.19.self_attn.k_proj.bias")
   val_266 = MatMul (layer_norm_38, val_121)
   scaled_dot_product_attention_19 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_114, linear_115, val_266, val_53_f_mask)
   val_267 = MatMul (scaled_dot_product_attention_19, val_122)
   linear_117 = Add (val_267, "text.transformer.layers.19.self_attn.out_proj.bias")
   add_2104 = Add (add_2031, linear_117)
   layer_norm_39 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_2104, "text.transformer.layers.19.final_layer_norm.weight", "text.transformer.layers.19.final_layer_norm.bias")
   val_268 = MatMul (layer_norm_39, val_123)
   linear_118 = Add (val_268, "text.transformer.layers.19.fc1.bias")
   relu_19 = Relu (linear_118)
   val_269 = MatMul (relu_19, val_124)
   linear_119 = Add (val_269, "text.transformer.layers.19.fc2.bias")
   add_2133 = Add (add_2104, linear_119)
   layer_norm_40 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_2133, "text.transformer.layers.20.self_attn_layer_norm.weight", "text.transformer.layers.20.self_attn_layer_norm.bias")
   val_270 = MatMul (layer_norm_40, val_125)
   linear_120 = Add (val_270, "text.transformer.layers.20.self_attn.q_proj.bias")
   val_271 = MatMul (layer_norm_40, val_126)
   linear_121 = Add (val_271, "text.transformer.layers.20.self_attn.k_proj.bias")
   val_272 = MatMul (layer_norm_40, val_127)
   scaled_dot_product_attention_20 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_120, linear_121, val_272, val_53_f_mask)
   val_273 = MatMul (scaled_dot_product_attention_20, val_128)
   linear_123 = Add (val_273, "text.transformer.layers.20.self_attn.out_proj.bias")
   add_2206 = Add (add_2133, linear_123)
   layer_norm_41 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_2206, "text.transformer.layers.20.final_layer_norm.weight", "text.transformer.layers.20.final_layer_norm.bias")
   val_274 = MatMul (layer_norm_41, val_129)
   linear_124 = Add (val_274, "text.transformer.layers.20.fc1.bias")
   relu_20 = Relu (linear_124)
   val_275 = MatMul (relu_20, val_130)
   linear_125 = Add (val_275, "text.transformer.layers.20.fc2.bias")
   add_2235 = Add (add_2206, linear_125)
   layer_norm_42 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_2235, "text.transformer.layers.21.self_attn_layer_norm.weight", "text.transformer.layers.21.self_attn_layer_norm.bias")
   val_276 = MatMul (layer_norm_42, val_131)
   linear_126 = Add (val_276, "text.transformer.layers.21.self_attn.q_proj.bias")
   val_277 = MatMul (layer_norm_42, val_132)
   linear_127 = Add (val_277, "text.transformer.layers.21.self_attn.k_proj.bias")
   val_278 = MatMul (layer_norm_42, val_133)
   scaled_dot_product_attention_21 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_126, linear_127, val_278, val_53_f_mask)
   val_279 = MatMul (scaled_dot_product_attention_21, val_134)
   linear_129 = Add (val_279, "text.transformer.layers.21.self_attn.out_proj.bias")
   add_2308 = Add (add_2235, linear_129)
   layer_norm_43 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_2308, "text.transformer.layers.21.final_layer_norm.weight", "text.transformer.layers.21.final_layer_norm.bias")
   val_280 = MatMul (layer_norm_43, val_135)
   linear_130 = Add (val_280, "text.transformer.layers.21.fc1.bias")
   relu_21 = Relu (linear_130)
   val_281 = MatMul (relu_21, val_136)
   linear_131 = Add (val_281, "text.transformer.layers.21.fc2.bias")
   add_2337 = Add (add_2308, linear_131)
   layer_norm_44 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_2337, "text.transformer.layers.22.self_attn_layer_norm.weight", "text.transformer.layers.22.self_attn_layer_norm.bias")
   val_282 = MatMul (layer_norm_44, val_137)
   linear_132 = Add (val_282, "text.transformer.layers.22.self_attn.q_proj.bias")
   val_283 = MatMul (layer_norm_44, val_138)
   linear_133 = Add (val_283, "text.transformer.layers.22.self_attn.k_proj.bias")
   val_284 = MatMul (layer_norm_44, val_139)
   scaled_dot_product_attention_22 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_132, linear_133, val_284, val_53_f_mask)
   val_285 = MatMul (scaled_dot_product_attention_22, val_140)
   linear_135 = Add (val_285, "text.transformer.layers.22.self_attn.out_proj.bias")
   add_2410 = Add (add_2337, linear_135)
   layer_norm_45 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_2410, "text.transformer.layers.22.final_layer_norm.weight", "text.transformer.layers.22.final_layer_norm.bias")
   val_286 = MatMul (layer_norm_45, val_141)
   linear_136 = Add (val_286, "text.transformer.layers.22.fc1.bias")
   relu_22 = Relu (linear_136)
   val_287 = MatMul (relu_22, val_142)
   linear_137 = Add (val_287, "text.transformer.layers.22.fc2.bias")
   add_2439 = Add (add_2410, linear_137)
   layer_norm_46 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_2439, "text.transformer.layers.23.self_attn_layer_norm.weight", "text.transformer.layers.23.self_attn_layer_norm.bias")
   val_288 = MatMul (layer_norm_46, val_143)
   linear_138 = Add (val_288, "text.transformer.layers.23.self_attn.q_proj.bias")
   val_289 = MatMul (layer_norm_46, val_144)
   linear_139 = Add (val_289, "text.transformer.layers.23.self_attn.k_proj.bias")
   val_290 = MatMul (layer_norm_46, val_145)
   scaled_dot_product_attention_23 = Attention <is_causal: int = 0, kv_num_heads: int = 16, q_num_heads: int = 16, qk_matmul_output_mode: int = 0, scale: float = 0.125, softcap: float = 0> (linear_138, linear_139, val_290, val_53_f_mask)
   [pool_hoist_scaled_dot_product_attention_23] scaled_dot_product_attention_23_pooled = Slice (scaled_dot_product_attention_23, val_4, val_3, val_3)
   val_291 = MatMul (scaled_dot_product_attention_23_pooled, val_146)
   linear_141 = Add (val_291, "text.transformer.layers.23.self_attn.out_proj.bias")
   [pool_hoist_add_2439] add_2439_pooled = Slice (add_2439, val_4, val_3, val_3)
   add_2512 = Add (add_2439_pooled, linear_141)
   layer_norm_47 = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (add_2512, "text.transformer.layers.23.final_layer_norm.weight", "text.transformer.layers.23.final_layer_norm.bias")
   val_292 = MatMul (layer_norm_47, val_147)
   linear_142 = Add (val_292, "text.transformer.layers.23.fc1.bias")
   relu_23 = Relu (linear_142)
   val_293 = MatMul (relu_23, val_148)
   linear_143 = Add (val_293, "text.transformer.layers.23.fc2.bias")
   add_2541 = Add (add_2512, linear_143)
   val_294 = Squeeze (add_2541, val_3)
   select = LayerNormalization <axis: int = -1, epsilon: float = 1e-05, stash_type: int = 1> (val_294, "text.transformer.layer_norm.weight", "text.transformer.layer_norm.bias")
   linear_144 = Gemm <alpha: float = 1, beta: float = 1, transA: int = 0, transB: int = 1> (select, "text.proj.weight")
   [node_linalg_vector_norm] linalg_vector_norm = ReduceL2 <keepdims: int = 1, noop_with_empty_axes: int = 0> (linear_144, val_0)
   [node_clamp_min] clamp_min = Clip (linalg_vector_norm, val_1)
   [node_div] text_embedding = Div (linear_144, clamp_min)
}

weights:
index_select FLOAT[77,1024] 5557f9a2e312
text.proj.weight FLOAT[1152,1024] 744c7b230868
text.transformer.embed_tokens.weight_fp16 FLOAT16[256206,1024] d6117c49aed7
text.transformer.layer_norm.bias FLOAT[1024] 48281ba98023
text.transformer.layer_norm.weight FLOAT[1024] 9324b64e988e
text.transformer.layers.0.fc1.bias FLOAT[8192] 8d953ce823cd
text.transformer.layers.0.fc2.bias FLOAT[1024] 7969d5ddd275
text.transformer.layers.0.final_layer_norm.bias FLOAT[1024] 5e1739ccb2b9
text.transformer.layers.0.final_layer_norm.weight FLOAT[1024] 095af9e11072
text.transformer.layers.0.self_attn.k_proj.bias FLOAT[1024] 65731d08d1c0
text.transformer.layers.0.self_attn.out_proj.bias FLOAT[1024] 6e4c1071de2c
text.transformer.layers.0.self_attn.q_proj.bias FLOAT[1024] 35645368d0b9
text.transformer.layers.0.self_attn_layer_norm.bias FLOAT[1024] 75ac4d7ceafe
text.transformer.layers.0.self_attn_layer_norm.weight FLOAT[1024] 8dba51ae5413
text.transformer.layers.1.fc1.bias FLOAT[8192] 634d0ae4e955
text.transformer.layers.1.fc2.bias FLOAT[1024] a0e20c8df601
text.transformer.layers.1.final_layer_norm.bias FLOAT[1024] cca8ef8c97c2
text.transformer.layers.1.final_layer_norm.weight FLOAT[1024] afb2e24055a0
text.transformer.layers.1.self_attn.k_proj.bias FLOAT[1024] 940ebffc283f
text.transformer.layers.1.self_attn.out_proj.bias FLOAT[1024] 3718ff2a88f0
text.transformer.layers.1.self_attn.q_proj.bias FLOAT[1024] ff9c2e0ac2ad
text.transformer.layers.1.self_attn_layer_norm.bias FLOAT[1024] a62e28a0a25b
text.transformer.layers.1.self_attn_layer_norm.weight FLOAT[1024] 7e85d036ddf3
text.transformer.layers.10.fc1.bias FLOAT[8192] ac6c449cb1c6
text.transformer.layers.10.fc2.bias FLOAT[1024] ac4ff2db740e
text.transformer.layers.10.final_layer_norm.bias FLOAT[1024] 93839f314968
text.transformer.layers.10.final_layer_norm.weight FLOAT[1024] 018fc3162bd4
text.transformer.layers.10.self_attn.k_proj.bias FLOAT[1024] 6ab70d08c65b
text.transformer.layers.10.self_attn.out_proj.bias FLOAT[1024] 6288c57684bd
text.transformer.layers.10.self_attn.q_proj.bias FLOAT[1024] 039d1b068926
text.transformer.layers.10.self_attn_layer_norm.bias FLOAT[1024] 94c2f475accf
text.transformer.layers.10.self_attn_layer_norm.weight FLOAT[1024] 1b6aca7dd565
text.transformer.layers.11.fc1.bias FLOAT[8192] e73774db48b2
text.transformer.layers.11.fc2.bias FLOAT[1024] b4745d651a3f
text.transformer.layers.11.final_layer_norm.bias FLOAT[1024] e4778248a316
text.transformer.layers.11.final_layer_norm.weight FLOAT[1024] 31cbeb370d06
text.transformer.layers.11.self_attn.k_proj.bias FLOAT[1024] ce1bd3ab4e16
text.transformer.layers.11.self_attn.out_proj.bias FLOAT[1024] fc8b431b0b03
text.transformer.layers.11.self_attn.q_proj.bias FLOAT[1024] a7db4616a905
text.transformer.layers.11.self_attn_layer_norm.bias FLOAT[1024] cab7e0a9eb6d
text.transformer.layers.11.self_attn_layer_norm.weight FLOAT[1024] b74270a1b523
text.transformer.layers.12.fc1.bias FLOAT[8192] 7b12de7e90b8
text.transformer.layers.12.fc2.bias FLOAT[1024] 90a0efbdb8b9
text.transformer.layers.12.final_layer_norm.bias FLOAT[1024] fd06f55d6346
text.transformer.layers.12.final_layer_norm.weight FLOAT[1024] 22fbc9d632f0
text.transformer.layers.12.self_attn.k_proj.bias FLOAT[1024] 9bdad57d1ccc
text.transformer.layers.12.self_attn.out_proj.bias FLOAT[1024] e0f32332f599
text.transformer.layers.12.self_attn.q_proj.bias FLOAT[1024] 9af8f83c9dcb
text.transformer.layers.12.self_attn_layer_norm.bias FLOAT[1024] e7cce26bdfae
text.transformer.layers.12.self_attn_layer_norm.weight FLOAT[1024] 55e28eaff68f
text.transformer.layers.13.fc1.bias FLOAT[8192] da1960b1944f
text.transformer.layers.13.fc2.bias FLOAT[1024] 722716093406
text.transformer.layers.13.final_layer_norm.bias FLOAT[1024] 3e55c2933f8e
text.transformer.layers.13.final_layer_norm.weight FLOAT[1024] 3fe9c4b996b3
text.transformer.layers.13.self_attn.k_proj.bias FLOAT[1024] a542f4179f42
text.transformer.layers.13.self_attn.out_proj.bias FLOAT[1024] 312e19abb3c9
text.transformer.layers.13.self_attn.q_proj.bias FLOAT[1024] 7029b537b44b
text.transformer.layers.13.self_attn_layer_norm.bias FLOAT[1024] f642860f6c5e
text.transformer.layers.13.self_attn_layer_norm.weight FLOAT[1024] f9a7a209416f
text.transformer.layers.14.fc1.bias FLOAT[8192] 7bd5972dad7e
text.transformer.layers.14.fc2.bias FLOAT[1024] cf23ffdbe69a
text.transformer.layers.14.final_layer_norm.bias FLOAT[1024] 265db779952c
text.transformer.layers.14.final_layer_norm.weight FLOAT[1024] 999130cb456d
text.transformer.layers.14.self_attn.k_proj.bias FLOAT[1024] d2a7db997583
text.transformer.layers.14.self_attn.out_proj.bias FLOAT[1024] e4affcf7f5b2
text.transformer.layers.14.self_attn.q_proj.bias FLOAT[1024] 6d767355ab1a
text.transformer.layers.14.self_attn_layer_norm.bias FLOAT[1024] 748f4b7fdaa7
text.transformer.layers.14.self_attn_layer_norm.weight FLOAT[1024] 3729de94b711
text.transformer.layers.15.fc1.bias FLOAT[8192] 83df89b40dc4
text.transformer.layers.15.fc2.bias FLOAT[1024] e525ded24e16
text.transformer.layers.15.final_layer_norm.bias FLOAT[1024] 3a2f16421d34
text.transformer.layers.15.final_layer_norm.weight FLOAT[1024] def1ad839a27
text.transformer.layers.15.self_attn.k_proj.bias FLOAT[1024] 6541c732ecd5
text.transformer.layers.15.self_attn.out_proj.bias FLOAT[1024] 043fffee2684
text.transformer.layers.15.self_attn.q_proj.bias FLOAT[1024] f2c4aee51a85
text.transformer.layers.15.self_attn_layer_norm.bias FLOAT[1024] b9ade2c3977d
text.transformer.layers.15.self_attn_layer_norm.weight FLOAT[1024] eac823b10a8e
text.transformer.layers.16.fc1.bias FLOAT[8192] 58c984a2e7d7
text.transformer.layers.16.fc2.bias FLOAT[1024] c3e5902e3e92
text.transformer.layers.16.final_layer_norm.bias FLOAT[1024] dd27746a7ae1
text.transformer.layers.16.final_layer_norm.weight FLOAT[1024] 0324686a38eb
text.transformer.layers.16.self_attn.k_proj.bias FLOAT[1024] aa2ee20a93c5
text.transformer.layers.16.self_attn.out_proj.bias FLOAT[1024] d58089b2c6a1
text.transformer.layers.16.self_attn.q_proj.bias FLOAT[1024] c53af9b40081
text.transformer.layers.16.self_attn_layer_norm.bias FLOAT[1024] 8742307348d3
text.transformer.layers.16.self_attn_layer_norm.weight FLOAT[1024] 9b35f53fc54a
text.transformer.layers.17.fc1.bias FLOAT[8192] 55d28fa32c91
text.transformer.layers.17.fc2.bias FLOAT[1024] 17accc9ae93a
text.transformer.layers.17.final_layer_norm.bias FLOAT[1024] e6052fb3181c
text.transformer.layers.17.final_layer_norm.weight FLOAT[1024] 90bf9241b92d
text.transformer.layers.17.self_attn.k_proj.bias FLOAT[1024] 153137491039
text.transformer.layers.17.self_attn.out_proj.bias FLOAT[1024] c12e10a19e9a
text.transformer.layers.17.self_attn.q_proj.bias FLOAT[1024] ef29cc9ec00a
text.transformer.layers.17.self_attn_layer_norm.bias FLOAT[1024] 1e02d63fde79
text.transformer.layers.17.self_attn_layer_norm.weight FLOAT[1024] 54e9c4e994ef
text.transformer.layers.18.fc1.bias FLOAT[8192] 794de8f87e3b
text.transformer.layers.18.fc2.bias FLOAT[1024] 576c66ddeedf
text.transformer.layers.18.final_layer_norm.bias FLOAT[1024] 5df0a386e456
text.transformer.layers.18.final_layer_norm.weight FLOAT[1024] 3d310d35ca70
text.transformer.layers.18.self_attn.k_proj.bias FLOAT[1024] 2be7fdfe99d6
text.transformer.layers.18.self_attn.out_proj.bias FLOAT[1024] a42fd6071a2e
text.transformer.layers.18.self_attn.q_proj.bias FLOAT[1024] 78ef0e235767
text.transformer.layers.18.self_attn_layer_norm.bias FLOAT[1024] d1611a33a659
text.transformer.layers.18.self_attn_layer_norm.weight FLOAT[1024] 330a42d60dac
text.transformer.layers.19.fc1.bias FLOAT[8192] b4752ec3c94f
text.transformer.layers.19.fc2.bias FLOAT[1024] 9b30fbb1ced7
text.transformer.layers.19.final_layer_norm.bias FLOAT[1024] 45f056424d4b
text.transformer.layers.19.final_layer_norm.weight FLOAT[1024] 42a4a537ba67
text.transformer.layers.19.self_attn.k_proj.bias FLOAT[1024] 034bf64d41d3
text.transformer.layers.19.self_attn.out_proj.bias FLOAT[1024] 785e50678dc5
text.transformer.layers.19.self_attn.q_proj.bias FLOAT[1024] d7f02bc21e1d
text.transformer.layers.19.self_attn_layer_norm.bias FLOAT[1024] 03efce85b098
text.transformer.layers.19.self_attn_layer_norm.weight FLOAT[1024] 37a1d251d66c
text.transformer.layers.2.fc1.bias FLOAT[8192] 0fcb7da13f3b
text.transformer.layers.2.fc2.bias FLOAT[1024] 146632c9a394
text.transformer.layers.2.final_layer_norm.bias FLOAT[1024] e0ebc95314e0
text.transformer.layers.2.final_layer_norm.weight FLOAT[1024] 3940a51915a7
text.transformer.layers.2.self_attn.k_proj.bias FLOAT[1024] 0d2a08245bce
text.transformer.layers.2.self_attn.out_proj.bias FLOAT[1024] b6cb70c54a33
text.transformer.layers.2.self_attn.q_proj.bias FLOAT[1024] 6d026578587c
text.transformer.layers.2.self_attn_layer_norm.bias FLOAT[1024] 1bf792541c90
text.transformer.layers.2.self_attn_layer_norm.weight FLOAT[1024] bd0da0b80194
text.transformer.layers.20.fc1.bias FLOAT[8192] 8851175e6b9d
text.transformer.layers.20.fc2.bias FLOAT[1024] 45a293345f1f
text.transformer.layers.20.final_layer_norm.bias FLOAT[1024] c8f88c38ef25
text.transformer.layers.20.final_layer_norm.weight FLOAT[1024] 188ebb921a3e
text.transformer.layers.20.self_attn.k_proj.bias FLOAT[1024] 91245279e2b4
text.transformer.layers.20.self_attn.out_proj.bias FLOAT[1024] 9aad750093c3
text.transformer.layers.20.self_attn.q_proj.bias FLOAT[1024] 015a19c8db9f
text.transformer.layers.20.self_attn_layer_norm.bias FLOAT[1024] 8fd1a5766340
text.transformer.layers.20.self_attn_layer_norm.weight FLOAT[1024] 81082328f822
text.transformer.layers.21.fc1.bias FLOAT[8192] 8aefdd109033
text.transformer.layers.21.fc2.bias FLOAT[1024] edc66df5277f
text.transformer.layers.21.final_layer_norm.bias FLOAT[1024] 1ab7820a9320
text.transformer.layers.21.final_layer_norm.weight FLOAT[1024] b8cd1624e57d
text.transformer.layers.21.self_attn.k_proj.bias FLOAT[1024] 23b421512c58
text.transformer.layers.21.self_attn.out_proj.bias FLOAT[1024] 6ef3350e39e2
text.transformer.layers.21.self_attn.q_proj.bias FLOAT[1024] bae9cd50a2d3
text.transformer.layers.21.self_attn_layer_norm.bias FLOAT[1024] f3e8b80aecb6
text.transformer.layers.21.self_attn_layer_norm.weight FLOAT[1024] 5b8afc88a0f6
text.transformer.layers.22.fc1.bias FLOAT[8192] 4c2ea902a330
text.transformer.layers.22.fc2.bias FLOAT[1024] a0380d9a8654
text.transformer.layers.22.final_layer_norm.bias FLOAT[1024] f63238e7bb84
text.transformer.layers.22.final_layer_norm.weight FLOAT[1024] 4d9838b4f990
text.transformer.layers.22.self_attn.k_proj.bias FLOAT[1024] 00bf9568c691
text.transformer.layers.22.self_attn.out_proj.bias FLOAT[1024] b5da530d9cb4
text.transformer.layers.22.self_attn.q_proj.bias FLOAT[1024] 0a61ee2504ea
text.transformer.layers.22.self_attn_layer_norm.bias FLOAT[1024] 0fd615d42817
text.transformer.layers.22.self_attn_layer_norm.weight FLOAT[1024] 81364ce0e716
text.transformer.layers.23.fc1.bias FLOAT[8192] 9a5705b0f66f
text.transformer.layers.23.fc2.bias FLOAT[1024] a575920a2585
text.transformer.layers.23.final_layer_norm.bias FLOAT[1024] 587338dfe7d1
text.transformer.layers.23.final_layer_norm.weight FLOAT[1024] 996b5f867e41
text.transformer.layers.23.self_attn.k_proj.bias FLOAT[1024] d4a1a0140ff4
text.transformer.layers.23.self_attn.out_proj.bias FLOAT[1024] c206743c1351
text.transformer.layers.23.self_attn.q_proj.bias FLOAT[1024] 7333c1dd810d
text.transformer.layers.23.self_attn_layer_norm.bias FLOAT[1024] 03f0e4d84e89
text.transformer.layers.23.self_attn_layer_norm.weight FLOAT[1024] 6e1386d82001
text.transformer.layers.3.fc1.bias FLOAT[8192] aba1a5898432
text.transformer.layers.3.fc2.bias FLOAT[1024] cce345c43f69
text.transformer.layers.3.final_layer_norm.bias FLOAT[1024] ede02fba7d2d
text.transformer.layers.3.final_layer_norm.weight FLOAT[1024] 7efc4027887e
text.transformer.layers.3.self_attn.k_proj.bias FLOAT[1024] 3186b2ee84d1
text.transformer.layers.3.self_attn.out_proj.bias FLOAT[1024] e910c0369ba7
text.transformer.layers.3.self_attn.q_proj.bias FLOAT[1024] ae4611f210a2
text.transformer.layers.3.self_attn_layer_norm.bias FLOAT[1024] 8f8825fd622d
text.transformer.layers.3.self_attn_layer_norm.weight FLOAT[1024] 88f562e2bd54
text.transformer.layers.4.fc1.bias FLOAT[8192] ab66009e5368
text.transformer.layers.4.fc2.bias FLOAT[1024] f3dfd7c086e8
text.transformer.layers.4.final_layer_norm.bias FLOAT[1024] c030c8f41ad3
text.transformer.layers.4.final_layer_norm.weight FLOAT[1024] ff12398f3ac9
text.transformer.layers.4.self_attn.k_proj.bias FLOAT[1024] ceed4e539627
text.transformer.layers.4.self_attn.out_proj.bias FLOAT[1024] 1808078da290
text.transformer.layers.4.self_attn.q_proj.bias FLOAT[1024] f44a8a5dfb5d
text.transformer.layers.4.self_attn_layer_norm.bias FLOAT[1024] 3597e42f16c7
text.transformer.layers.4.self_attn_layer_norm.weight FLOAT[1024] 45397992ddf5
text.transformer.layers.5.fc1.bias FLOAT[8192] 2b7e961b036a
text.transformer.layers.5.fc2.bias FLOAT[1024] ed5c2d6e8f0b
text.transformer.layers.5.final_layer_norm.bias FLOAT[1024] dba0c5335c7b
text.transformer.layers.5.final_layer_norm.weight FLOAT[1024] 12b4aa8d4d2e
text.transformer.layers.5.self_attn.k_proj.bias FLOAT[1024] ebd6eec24aef
text.transformer.layers.5.self_attn.out_proj.bias FLOAT[1024] 86b62625211a
text.transformer.layers.5.self_attn.q_proj.bias FLOAT[1024] ad12abd8af2e
text.transformer.layers.5.self_attn_layer_norm.bias FLOAT[1024] 3b904325bdd3
text.transformer.layers.5.self_attn_layer_norm.weight FLOAT[1024] b90bea20410a
text.transformer.layers.6.fc1.bias FLOAT[8192] 82ae7418b25d
text.transformer.layers.6.fc2.bias FLOAT[1024] 64070be6ec11
text.transformer.layers.6.final_layer_norm.bias FLOAT[1024] 90d10243693a
text.transformer.layers.6.final_layer_norm.weight FLOAT[1024] 856cf108b699
text.transformer.layers.6.self_attn.k_proj.bias FLOAT[1024] bf07ac91ccba
text.transformer.layers.6.self_attn.out_proj.bias FLOAT[1024] 49a67a219829
text.transformer.layers.6.self_attn.q_proj.bias FLOAT[1024] c6253bbf974f
text.transformer.layers.6.self_attn_layer_norm.bias FLOAT[1024] 25efcaa12eb1
text.transformer.layers.6.self_attn_layer_norm.weight FLOAT[1024] e37c886cc88d
text.transformer.layers.7.fc1.bias FLOAT[8192] e8c7d04faafc
text.transformer.layers.7.fc2.bias FLOAT[1024] 37d31aba74fc
text.transformer.layers.7.final_layer_norm.bias FLOAT[1024] b3bd909f3da1
text.transformer.layers.7.final_layer_norm.weight FLOAT[1024] 5bd21c2f8894
text.transformer.layers.7.self_attn.k_proj.bias FLOAT[1024] 3f91c022804b
text.transformer.layers.7.self_attn.out_proj.bias FLOAT[1024] 48f97f95ff68
text.transformer.layers.7.self_attn.q_proj.bias FLOAT[1024] 0d4f82347209
text.transformer.layers.7.self_attn_layer_norm.bias FLOAT[1024] 4fb86fad9ba0
text.transformer.layers.7.self_attn_layer_norm.weight FLOAT[1024] c0e129cdc6ef
text.transformer.layers.8.fc1.bias FLOAT[8192] 0600c4959c6b
text.transformer.layers.8.fc2.bias FLOAT[1024] bd0f49cbf912
text.transformer.layers.8.final_layer_norm.bias FLOAT[1024] d80990b88123
text.transformer.layers.8.final_layer_norm.weight FLOAT[1024] f09bd58a1f4c
text.transformer.layers.8.self_attn.k_proj.bias FLOAT[1024] bfb71f84c010
text.transformer.layers.8.self_attn.out_proj.bias FLOAT[1024] 1ceaf7175b61
text.transformer.layers.8.self_attn.q_proj.bias FLOAT[1024] b1473ce74a27
text.transformer.layers.8.self_attn_layer_norm.bias FLOAT[1024] a6b82c5dbfee
text.transformer.layers.8.self_attn_layer_norm.weight FLOAT[1024] 3f8fa335a73a
text.transformer.layers.9.fc1.bias FLOAT[8192] ef170a48a222
text.transformer.layers.9.fc2.bias FLOAT[1024] 9a85c801d0ae
text.transformer.layers.9.final_layer_norm.bias FLOAT[1024] 4864135639dc
text.transformer.layers.9.final_layer_norm.weight FLOAT[1024] eb052aa5bbc5
text.transformer.layers.9.self_attn.k_proj.bias FLOAT[1024] 6f4901948cf4
text.transformer.layers.9.self_attn.out_proj.bias FLOAT[1024] 18bd2708a8ba
text.transformer.layers.9.self_attn.q_proj.bias FLOAT[1024] b61d87d49838
text.transformer.layers.9.self_attn_layer_norm.bias FLOAT[1024] 3e86f6f3762d
text.transformer.layers.9.self_attn_layer_norm.weight FLOAT[1024] 275ee6452f24
text_one FLOAT[] e00e5eb94441
text_pad_keep FLOAT[256206] 47daba355159
text_q_axis FLOAT[77,1] 9575b2125169
text_row_axes INT64[2] 0c730b69905c
text_scale FLOAT[] e401200e5808
val_0 INT64[1] 12a3ae445661
val_1 FLOAT[] 6708d9be4956
val_10 FLOAT[8192,1024] 420803442f13
val_100 FLOAT[8192,1024] 2663b9446204
val_101 FLOAT[1024,1024] 340713d9715b
val_102 FLOAT[1024,1024] fecebf087891
val_103 FLOAT[1024,1024] 30285600db2e
val_104 FLOAT[1024,1024] 214bc90bd462
val_105 FLOAT[1024,8192] 671b03e642f0
val_106 FLOAT[8192,1024] 380aced66fc8
val_107 FLOAT[1024,1024] 61a6c57c3e90
val_108 FLOAT[1024,1024] 69e333090691
val_109 FLOAT[1024,1024] ddfa9dffc5cf
val_11 FLOAT[1024,1024] 79c5962d7c77
val_110 FLOAT[1024,1024] a1c38bafcc9e
val_111 FLOAT[1024,8192] 16e96c405802
val_112 FLOAT[8192,1024] e0b5a9c3803b
val_113 FLOAT[1024,1024] 51b5d6fe83a5
val_114 FLOAT[1024,1024] 4917884358da
val_115 FLOAT[1024,1024] 7217192c0362
val_116 FLOAT[1024,1024] 503f7c212bd5
val_117 FLOAT[1024,8192] 0271131ccd08
val_118 FLOAT[8192,1024] e4002c20784f
val_119 FLOAT[1024,1024] 4f94f6aee019
val_12 FLOAT[1024,1024] b08f8c2266f1
val_120 FLOAT[1024,1024] b008c5981a25
val_121 FLOAT[1024,1024] 3003a0c7acbc
val_122 FLOAT[1024,1024] 53868aa25345
val_123 FLOAT[1024,8192] a04c7281beb1
val_124 FLOAT[8192,1024] 0b1d6526d7f2
val_125 FLOAT[1024,1024] 83e19ed22a68
val_126 FLOAT[1024,1024] ab8c1c497a52
val_127 FLOAT[1024,1024] f7ef8fedef30
val_128 FLOAT[1024,1024] 3ef177e89c00
val_129 FLOAT[1024,8192] 89433893b9ce
val_13 FLOAT[1024,1024] 5d064a98712e
val_130 FLOAT[8192,1024] b9c4f84b6689
val_131 FLOAT[1024,1024] 87556b1949fd
val_132 FLOAT[1024,1024] 4c8b98894cec
val_133 FLOAT[1024,1024] 0b262143dc82
val_134 FLOAT[1024,1024] 5cadf0b41a1d
val_135 FLOAT[1024,8192] f2de67c15958
val_136 FLOAT[8192,1024] 8d170ef0c3dd
val_137 FLOAT[1024,1024] a55607a7fe9e
val_138 FLOAT[1024,1024] a60de05424ef
val_139 FLOAT[1024,1024] 09e2ad3b1539
val_14 FLOAT[1024,1024] 5efb65f67c36
val_140 FLOAT[1024,1024] bc34b58983b9
val_141 FLOAT[1024,8192] af155575c681
val_142 FLOAT[8192,1024] e2a598132040
val_143 FLOAT[1024,1024] 4c1aced4bab4
val_144 FLOAT[1024,1024] 37981311dfaf
val_145 FLOAT[1024,1024] 6beab35332f4
val_146 FLOAT[1024,1024] ef662cb7979f
val_147 FLOAT[1024,8192] 23fd926b3824
val_148 FLOAT[8192,1024] 4e28ac987c82
val_15 FLOAT[1024,8192] ac7a021cefe8
val_16 FLOAT[8192,1024] 1792230f7f1d
val_17 FLOAT[1024,1024] 38915fb8c18f
val_18 FLOAT[1024,1024] 30bf769a0a4e
val_19 FLOAT[1024,1024] c0e7a2407b1c
val_2 FLOAT[] 825ac1bb838d
val_20 FLOAT[1024,1024] 651f870f0598
val_21 FLOAT[1024,8192] e16d226fc992
val_22 FLOAT[8192,1024] aac72f1afb62
val_23 FLOAT[1024,1024] 492579e2aa0e
val_24 FLOAT[1024,1024] 55e048f618ba
val_25 FLOAT[1024,1024] c63ea16033b7
val_26 FLOAT[1024,1024] e15e486e5496
val_27 FLOAT[1024,8192] d974898e1775
val_28 FLOAT[8192,1024] 1dabe9387b79
val_29 FLOAT[1024,1024] cc7fe19a64f4
val_3 INT64[1] 7c9fa136d441
val_30 FLOAT[1024,1024] 8a532ccd2677
val_31 FLOAT[1024,1024] 84e6c0802336
val_32 FLOAT[1024,1024] 00e92676ef9e
val_33 FLOAT[1024,8192] 4c84b5dec0d3
val_34 FLOAT[8192,1024] 96406c21a589
val_35 FLOAT[1024,1024] bce4be7cbc51
val_36 FLOAT[1024,1024] 3d3b80ccae41
val_37 FLOAT[1024,1024] 1caeda91c18b
val_38 FLOAT[1024,1024] c16b74accd2b
val_39 FLOAT[1024,8192] e8e197b9aaca
val_4 INT64[1] af5570f5a181
val_40 FLOAT[8192,1024] 5c2c4f7a3378
val_41 FLOAT[1024,1024] 40880d253da2
val_42 FLOAT[1024,1024] 39774d53d598
val_43 FLOAT[1024,1024] 504da6e36a8d
val_44 FLOAT[1024,1024] 09a627d9fee7
val_45 FLOAT[1024,8192] 5041d22915fd
val_46 FLOAT[8192,1024] 718b9fa09ccb
val_47 FLOAT[1024,1024] d8c259fa3801
val_48 FLOAT[1024,1024] aa816e583046
val_49 FLOAT[1024,1024] 15866f8e6967
val_5 FLOAT[1024,1024] 617487bf6b9f
val_50 FLOAT[1024,1024] 7d0dd917d7c8
val_51 FLOAT[1024,8192] 0fbf421a9ab2
val_52 FLOAT[8192,1024] 5f6cd4651325
val_53 FLOAT[1024,1024] e3b84c293793
val_54 FLOAT[1024,1024] 90f4d569c2ab
val_55 FLOAT[1024,1024] 78bcfbc1673e
val_56 FLOAT[1024,1024] fe0892974081
val_57 FLOAT[1024,8192] 03b18a514791
val_58 FLOAT[8192,1024] 00ee352b5285
val_59 FLOAT[1024,1024] 98468466e54d
val_6 FLOAT[1024,1024] 2cf3b13364a1
val_60 FLOAT[1024,1024] 4baecbab6564
val_61 FLOAT[1024,1024] 8703c9c87bc7
val_62 FLOAT[1024,1024] db2c707d5792
val_63 FLOAT[1024,8192] fe1466af1167
val_64 FLOAT[8192,1024] f5b1980867bf
val_65 FLOAT[1024,1024] ec53a398e0b0
val_66 FLOAT[1024,1024] d3a2b0673e13
val_67 FLOAT[1024,1024] f93f65cc7fed
val_68 FLOAT[1024,1024] ce0cf258c2f6
val_69 FLOAT[1024,8192] fe344df412eb
val_7 FLOAT[1024,1024] 013f859ee8d6
val_70 FLOAT[8192,1024] e156d7460be8
val_71 FLOAT[1024,1024] a090ad0e8d0a
val_72 FLOAT[1024,1024] 62e7b5b87af7
val_73 FLOAT[1024,1024] c8488ae303fe
val_74 FLOAT[1024,1024] 604078593db3
val_75 FLOAT[1024,8192] ca25c3399cae
val_76 FLOAT[8192,1024] e61d49fbc819
val_77 FLOAT[1024,1024] dfb018264816
val_78 FLOAT[1024,1024] 7354704f652a
val_79 FLOAT[1024,1024] d4f7b2756d34
val_8 FLOAT[1024,1024] b3c26c42b3d1
val_80 FLOAT[1024,1024] bba02e4ffb51
val_81 FLOAT[1024,8192] e1776282d379
val_82 FLOAT[8192,1024] 72ac47dd12f0
val_83 FLOAT[1024,1024] 12835ff146ca
val_84 FLOAT[1024,1024] cdcda639fb80
val_85 FLOAT[1024,1024] efa54129149e
val_86 FLOAT[1024,1024] f43818ebcbb9
val_87 FLOAT[1024,8192] f41f4061db41
val_88 FLOAT[8192,1024] 849668b36f23
val_89 FLOAT[1024,1024] 4b585eb605b7
val_9 FLOAT[1024,8192] f6b91f9d3803
val_90 FLOAT[1024,1024] 3959c3d0e164
val_91 FLOAT[1024,1024] 119cdde28808
val_92 FLOAT[1024,1024] b2fd0b8bc1ac
val_93 FLOAT[1024,8192] 87a6aa869010
val_94 FLOAT[8192,1024] 51c21c0bcc83
val_95 FLOAT[1024,1024] 004835ffe6e6
val_96 FLOAT[1024,1024] fa8709396f1b
val_97 FLOAT[1024,1024] adb98de60864
val_98 FLOAT[1024,1024] a80885967ee3
val_99 FLOAT[1024,8192] 68a9c556fcd4
view_1_target INT64[3] 6091ae349ac5
