model: lm_head: accum_format: SAME approximation_function: NONE input_format: SAME instance: Linear output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.drop: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.0.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.0.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.0.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.0.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.0.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.0.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.0.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.0.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.0.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.0.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.0.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.1.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.1.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.1.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.1.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.1.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.1.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.1.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.1.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.1.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.1.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.1.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.10.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.10.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.10.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.10.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.10.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.10.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.10.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.10.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.10.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.10.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.10.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.11.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.11.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.11.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.11.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.11.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.11.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.11.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.11.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.11.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.11.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.11.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.12.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.12.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.12.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.12.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.12.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.12.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.12.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.12.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.12.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.12.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.12.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.13.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.13.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.13.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.13.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.13.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.13.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.13.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.13.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.13.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.13.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.13.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.14.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.14.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.14.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.14.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.14.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.14.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.14.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.14.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.14.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.14.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.14.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.15.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.15.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.15.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.15.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.15.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.15.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.15.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.15.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.15.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.15.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.15.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.16.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.16.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.16.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.16.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.16.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.16.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.16.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.16.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.16.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.16.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.16.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.17.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.17.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.17.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.17.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.17.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.17.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.17.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.17.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.17.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.17.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.17.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.18.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.18.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.18.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.18.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.18.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.18.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.18.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.18.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.18.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.18.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.18.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.19.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.19.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.19.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.19.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.19.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.19.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.19.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.19.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.19.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.19.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.19.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.2.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.2.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.2.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.2.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.2.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.2.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.2.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.2.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.2.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.2.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.2.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.20.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.20.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.20.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.20.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.20.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.20.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.20.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.20.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.20.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.20.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.20.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.21.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.21.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.21.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.21.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.21.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.21.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.21.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.21.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.21.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.21.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.21.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.22.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.22.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.22.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.22.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.22.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.22.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.22.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.22.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.22.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.22.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.22.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.23.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.23.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.23.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.23.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.23.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.23.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.23.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.23.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.23.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.23.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.23.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.24.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.24.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.24.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.24.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.24.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.24.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.24.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.24.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.24.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.24.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.24.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.25.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.25.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.25.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.25.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.25.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.25.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.25.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.25.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.25.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.25.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.25.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.26.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.26.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.26.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.26.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.26.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.26.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.26.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.26.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.26.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.26.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.26.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.27.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.27.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.27.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.27.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.27.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.27.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.27.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.27.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.27.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.27.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.27.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.28.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.28.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.28.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.28.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.28.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.28.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.28.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.28.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.28.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.28.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.28.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.29.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.29.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.29.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.29.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.29.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.29.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.29.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.29.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.29.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.29.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.29.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.3.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.3.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.3.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.3.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.3.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.3.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.3.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.3.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.3.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.3.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.3.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.30.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.30.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.30.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.30.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.30.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.30.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.30.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.30.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.30.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.30.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.30.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.31.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.31.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.31.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.31.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.31.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.31.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.31.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.31.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.31.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.31.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.31.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.32.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.32.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.32.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.32.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.32.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.32.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.32.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.32.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.32.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.32.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.32.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.33.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.33.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.33.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.33.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.33.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.33.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.33.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.33.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.33.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.33.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.33.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.34.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.34.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.34.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.34.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.34.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.34.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.34.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.34.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.34.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.34.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.34.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.35.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.35.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.35.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.35.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.35.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.35.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.35.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.35.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.35.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.35.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.35.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.36.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.36.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.36.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.36.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.36.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.36.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.36.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.36.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.36.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.36.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.36.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.37.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.37.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.37.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.37.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.37.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.37.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.37.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.37.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.37.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.37.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.37.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.38.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.38.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.38.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.38.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.38.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.38.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.38.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.38.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.38.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.38.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.38.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.39.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.39.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.39.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.39.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.39.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.39.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.39.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.39.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.39.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.39.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.39.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.4.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.4.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.4.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.4.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.4.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.4.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.4.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.4.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.4.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.4.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.4.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.40.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.40.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.40.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.40.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.40.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.40.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.40.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.40.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.40.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.40.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.40.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.41.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.41.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.41.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.41.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.41.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.41.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.41.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.41.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.41.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.41.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.41.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.42.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.42.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.42.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.42.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.42.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.42.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.42.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.42.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.42.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.42.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.42.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.43.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.43.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.43.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.43.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.43.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.43.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.43.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.43.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.43.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.43.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.43.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.44.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.44.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.44.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.44.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.44.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.44.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.44.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.44.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.44.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.44.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.44.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.45.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.45.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.45.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.45.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.45.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.45.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.45.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.45.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.45.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.45.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.45.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.46.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.46.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.46.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.46.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.46.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.46.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.46.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.46.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.46.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.46.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.46.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.47.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.47.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.47.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.47.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.47.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.47.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.47.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.47.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.47.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.47.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.47.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.5.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.5.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.5.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.5.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.5.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.5.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.5.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.5.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.5.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.5.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.5.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.6.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.6.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.6.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.6.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.6.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.6.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.6.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.6.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.6.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.6.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.6.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.7.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.7.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.7.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.7.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.7.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.7.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.7.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.7.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.7.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.7.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.7.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.8.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.8.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.8.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.8.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.8.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.8.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.8.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.8.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.8.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.8.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.8.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.9.attn.attn_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.9.attn.c_attn: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.9.attn.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.9.attn.resid_dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.h.9.attn.softmax: approximation_function: NONE input_format: SAME instance: Softmax output_format: SAME transformer.h.9.ln_1: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.9.ln_2: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME transformer.h.9.mlp.act: approximation_function: NONE input_format: SAME instance: GELU output_format: SAME transformer.h.9.mlp.c_fc: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.9.mlp.c_proj: approximation_function: NONE bias_format: SAME input_format: SAME instance: HFTransformersConv1D output_format: SAME weight_format: SAME weight_sparseness: DENSE transformer.h.9.mlp.dropout: approximation_function: NONE input_format: SAME instance: Dropout output_format: SAME transformer.ln_f: approximation_function: NONE bias_format: SAME input_format: SAME instance: LayerNorm output_format: SAME weight_format: SAME