Update config.json

#1
by csy0225 - opened
Files changed (1) hide show
  1. config.json +1421 -24
config.json CHANGED
@@ -1,11 +1,8 @@
1
  {
2
  "architectures": [
3
- "MMGPTStepRoboticsForCausalLM"
4
  ],
5
- "auto_map": {
6
- "AutoConfig": "configuration_step_robotics.StepRoboticsConfig"
7
- },
8
- "model_type": "step3p5v",
9
  "im_end_token": "<im_end>",
10
  "im_patch_token": "<im_patch>",
11
  "im_start_token": "<im_start>",
@@ -32,9 +29,9 @@
32
  },
33
  "text_config": {
34
  "architectures": [
35
- "Step4ForCausalLM"
36
  ],
37
- "model_type": "step4",
38
  "att_impl_type": "GQA",
39
  "torch_dtype": "bfloat16",
40
  "norm_dtype": "float32",
@@ -685,23 +682,1423 @@
685
  "apply_to_layer_types": [
686
  "full_attention"
687
  ]
688
- },
689
- "quantization_config": {
690
- "quant_method": "fp8",
691
- "activation_scheme": "dynamic",
692
- "fmt": "e4m3",
693
- "weight_block_size": [
694
- 128,
695
- 128
696
- ],
697
- "modules_to_not_convert": [
698
- "model.layers.88.moe.experts.w13_weight",
699
- "model.layers.88.moe.experts.w2_weight",
700
- "model.layers.89.moe.experts.w13_weight",
701
- "model.layers.89.moe.experts.w2_weight",
702
- "model.layers.90.moe.experts.w13_weight",
703
- "model.layers.90.moe.experts.w2_weight"
704
- ]
705
  }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
706
  }
707
  }
 
1
  {
2
  "architectures": [
3
+ "Step5ForConditionalGeneration"
4
  ],
5
+ "model_type": "step5_vl",
 
 
 
6
  "im_end_token": "<im_end>",
7
  "im_patch_token": "<im_patch>",
8
  "im_start_token": "<im_start>",
 
29
  },
30
  "text_config": {
31
  "architectures": [
32
+ "Step5ForCausalLM"
33
  ],
34
+ "model_type": "step5",
35
  "att_impl_type": "GQA",
36
  "torch_dtype": "bfloat16",
37
  "norm_dtype": "float32",
 
682
  "apply_to_layer_types": [
683
  "full_attention"
684
  ]
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
685
  }
686
+ },
687
+ "quantization_config": {
688
+ "quant_method": "fp8",
689
+ "activation_scheme": "dynamic",
690
+ "fmt": "e4m3",
691
+ "weight_block_size": [
692
+ 128,
693
+ 128
694
+ ],
695
+ "modules_to_not_convert": [
696
+ "model.layers.88.moe.experts.w13_weight",
697
+ "model.layers.88.moe.experts.w2_weight",
698
+ "model.layers.89.moe.experts.w13_weight",
699
+ "model.layers.89.moe.experts.w2_weight",
700
+ "model.layers.90.moe.experts.w13_weight",
701
+ "model.layers.90.moe.experts.w2_weight",
702
+ "vision_model.transformer.resblocks.0.attn.qkv_proj",
703
+ "vision_model.transformer.resblocks.0.attn.q_proj",
704
+ "vision_model.transformer.resblocks.0.attn.k_proj",
705
+ "vision_model.transformer.resblocks.0.attn.v_proj",
706
+ "vision_model.transformer.resblocks.0.attn.out_proj",
707
+ "vision_model.transformer.resblocks.0.mlp.fc1",
708
+ "vision_model.transformer.resblocks.0.mlp.fc2",
709
+ "vision_model.transformer.resblocks.1.attn.qkv_proj",
710
+ "vision_model.transformer.resblocks.1.attn.q_proj",
711
+ "vision_model.transformer.resblocks.1.attn.k_proj",
712
+ "vision_model.transformer.resblocks.1.attn.v_proj",
713
+ "vision_model.transformer.resblocks.1.attn.out_proj",
714
+ "vision_model.transformer.resblocks.1.mlp.fc1",
715
+ "vision_model.transformer.resblocks.1.mlp.fc2",
716
+ "vision_model.transformer.resblocks.2.attn.qkv_proj",
717
+ "vision_model.transformer.resblocks.2.attn.q_proj",
718
+ "vision_model.transformer.resblocks.2.attn.k_proj",
719
+ "vision_model.transformer.resblocks.2.attn.v_proj",
720
+ "vision_model.transformer.resblocks.2.attn.out_proj",
721
+ "vision_model.transformer.resblocks.2.mlp.fc1",
722
+ "vision_model.transformer.resblocks.2.mlp.fc2",
723
+ "vision_model.transformer.resblocks.3.attn.qkv_proj",
724
+ "vision_model.transformer.resblocks.3.attn.q_proj",
725
+ "vision_model.transformer.resblocks.3.attn.k_proj",
726
+ "vision_model.transformer.resblocks.3.attn.v_proj",
727
+ "vision_model.transformer.resblocks.3.attn.out_proj",
728
+ "vision_model.transformer.resblocks.3.mlp.fc1",
729
+ "vision_model.transformer.resblocks.3.mlp.fc2",
730
+ "vision_model.transformer.resblocks.4.attn.qkv_proj",
731
+ "vision_model.transformer.resblocks.4.attn.q_proj",
732
+ "vision_model.transformer.resblocks.4.attn.k_proj",
733
+ "vision_model.transformer.resblocks.4.attn.v_proj",
734
+ "vision_model.transformer.resblocks.4.attn.out_proj",
735
+ "vision_model.transformer.resblocks.4.mlp.fc1",
736
+ "vision_model.transformer.resblocks.4.mlp.fc2",
737
+ "vision_model.transformer.resblocks.5.attn.qkv_proj",
738
+ "vision_model.transformer.resblocks.5.attn.q_proj",
739
+ "vision_model.transformer.resblocks.5.attn.k_proj",
740
+ "vision_model.transformer.resblocks.5.attn.v_proj",
741
+ "vision_model.transformer.resblocks.5.attn.out_proj",
742
+ "vision_model.transformer.resblocks.5.mlp.fc1",
743
+ "vision_model.transformer.resblocks.5.mlp.fc2",
744
+ "vision_model.transformer.resblocks.6.attn.qkv_proj",
745
+ "vision_model.transformer.resblocks.6.attn.q_proj",
746
+ "vision_model.transformer.resblocks.6.attn.k_proj",
747
+ "vision_model.transformer.resblocks.6.attn.v_proj",
748
+ "vision_model.transformer.resblocks.6.attn.out_proj",
749
+ "vision_model.transformer.resblocks.6.mlp.fc1",
750
+ "vision_model.transformer.resblocks.6.mlp.fc2",
751
+ "vision_model.transformer.resblocks.7.attn.qkv_proj",
752
+ "vision_model.transformer.resblocks.7.attn.q_proj",
753
+ "vision_model.transformer.resblocks.7.attn.k_proj",
754
+ "vision_model.transformer.resblocks.7.attn.v_proj",
755
+ "vision_model.transformer.resblocks.7.attn.out_proj",
756
+ "vision_model.transformer.resblocks.7.mlp.fc1",
757
+ "vision_model.transformer.resblocks.7.mlp.fc2",
758
+ "vision_model.transformer.resblocks.8.attn.qkv_proj",
759
+ "vision_model.transformer.resblocks.8.attn.q_proj",
760
+ "vision_model.transformer.resblocks.8.attn.k_proj",
761
+ "vision_model.transformer.resblocks.8.attn.v_proj",
762
+ "vision_model.transformer.resblocks.8.attn.out_proj",
763
+ "vision_model.transformer.resblocks.8.mlp.fc1",
764
+ "vision_model.transformer.resblocks.8.mlp.fc2",
765
+ "vision_model.transformer.resblocks.9.attn.qkv_proj",
766
+ "vision_model.transformer.resblocks.9.attn.q_proj",
767
+ "vision_model.transformer.resblocks.9.attn.k_proj",
768
+ "vision_model.transformer.resblocks.9.attn.v_proj",
769
+ "vision_model.transformer.resblocks.9.attn.out_proj",
770
+ "vision_model.transformer.resblocks.9.mlp.fc1",
771
+ "vision_model.transformer.resblocks.9.mlp.fc2",
772
+ "vision_model.transformer.resblocks.10.attn.qkv_proj",
773
+ "vision_model.transformer.resblocks.10.attn.q_proj",
774
+ "vision_model.transformer.resblocks.10.attn.k_proj",
775
+ "vision_model.transformer.resblocks.10.attn.v_proj",
776
+ "vision_model.transformer.resblocks.10.attn.out_proj",
777
+ "vision_model.transformer.resblocks.10.mlp.fc1",
778
+ "vision_model.transformer.resblocks.10.mlp.fc2",
779
+ "vision_model.transformer.resblocks.11.attn.qkv_proj",
780
+ "vision_model.transformer.resblocks.11.attn.q_proj",
781
+ "vision_model.transformer.resblocks.11.attn.k_proj",
782
+ "vision_model.transformer.resblocks.11.attn.v_proj",
783
+ "vision_model.transformer.resblocks.11.attn.out_proj",
784
+ "vision_model.transformer.resblocks.11.mlp.fc1",
785
+ "vision_model.transformer.resblocks.11.mlp.fc2",
786
+ "vision_model.transformer.resblocks.12.attn.qkv_proj",
787
+ "vision_model.transformer.resblocks.12.attn.q_proj",
788
+ "vision_model.transformer.resblocks.12.attn.k_proj",
789
+ "vision_model.transformer.resblocks.12.attn.v_proj",
790
+ "vision_model.transformer.resblocks.12.attn.out_proj",
791
+ "vision_model.transformer.resblocks.12.mlp.fc1",
792
+ "vision_model.transformer.resblocks.12.mlp.fc2",
793
+ "vision_model.transformer.resblocks.13.attn.qkv_proj",
794
+ "vision_model.transformer.resblocks.13.attn.q_proj",
795
+ "vision_model.transformer.resblocks.13.attn.k_proj",
796
+ "vision_model.transformer.resblocks.13.attn.v_proj",
797
+ "vision_model.transformer.resblocks.13.attn.out_proj",
798
+ "vision_model.transformer.resblocks.13.mlp.fc1",
799
+ "vision_model.transformer.resblocks.13.mlp.fc2",
800
+ "vision_model.transformer.resblocks.14.attn.qkv_proj",
801
+ "vision_model.transformer.resblocks.14.attn.q_proj",
802
+ "vision_model.transformer.resblocks.14.attn.k_proj",
803
+ "vision_model.transformer.resblocks.14.attn.v_proj",
804
+ "vision_model.transformer.resblocks.14.attn.out_proj",
805
+ "vision_model.transformer.resblocks.14.mlp.fc1",
806
+ "vision_model.transformer.resblocks.14.mlp.fc2",
807
+ "vision_model.transformer.resblocks.15.attn.qkv_proj",
808
+ "vision_model.transformer.resblocks.15.attn.q_proj",
809
+ "vision_model.transformer.resblocks.15.attn.k_proj",
810
+ "vision_model.transformer.resblocks.15.attn.v_proj",
811
+ "vision_model.transformer.resblocks.15.attn.out_proj",
812
+ "vision_model.transformer.resblocks.15.mlp.fc1",
813
+ "vision_model.transformer.resblocks.15.mlp.fc2",
814
+ "vision_model.transformer.resblocks.16.attn.qkv_proj",
815
+ "vision_model.transformer.resblocks.16.attn.q_proj",
816
+ "vision_model.transformer.resblocks.16.attn.k_proj",
817
+ "vision_model.transformer.resblocks.16.attn.v_proj",
818
+ "vision_model.transformer.resblocks.16.attn.out_proj",
819
+ "vision_model.transformer.resblocks.16.mlp.fc1",
820
+ "vision_model.transformer.resblocks.16.mlp.fc2",
821
+ "vision_model.transformer.resblocks.17.attn.qkv_proj",
822
+ "vision_model.transformer.resblocks.17.attn.q_proj",
823
+ "vision_model.transformer.resblocks.17.attn.k_proj",
824
+ "vision_model.transformer.resblocks.17.attn.v_proj",
825
+ "vision_model.transformer.resblocks.17.attn.out_proj",
826
+ "vision_model.transformer.resblocks.17.mlp.fc1",
827
+ "vision_model.transformer.resblocks.17.mlp.fc2",
828
+ "vision_model.transformer.resblocks.18.attn.qkv_proj",
829
+ "vision_model.transformer.resblocks.18.attn.q_proj",
830
+ "vision_model.transformer.resblocks.18.attn.k_proj",
831
+ "vision_model.transformer.resblocks.18.attn.v_proj",
832
+ "vision_model.transformer.resblocks.18.attn.out_proj",
833
+ "vision_model.transformer.resblocks.18.mlp.fc1",
834
+ "vision_model.transformer.resblocks.18.mlp.fc2",
835
+ "vision_model.transformer.resblocks.19.attn.qkv_proj",
836
+ "vision_model.transformer.resblocks.19.attn.q_proj",
837
+ "vision_model.transformer.resblocks.19.attn.k_proj",
838
+ "vision_model.transformer.resblocks.19.attn.v_proj",
839
+ "vision_model.transformer.resblocks.19.attn.out_proj",
840
+ "vision_model.transformer.resblocks.19.mlp.fc1",
841
+ "vision_model.transformer.resblocks.19.mlp.fc2",
842
+ "vision_model.transformer.resblocks.20.attn.qkv_proj",
843
+ "vision_model.transformer.resblocks.20.attn.q_proj",
844
+ "vision_model.transformer.resblocks.20.attn.k_proj",
845
+ "vision_model.transformer.resblocks.20.attn.v_proj",
846
+ "vision_model.transformer.resblocks.20.attn.out_proj",
847
+ "vision_model.transformer.resblocks.20.mlp.fc1",
848
+ "vision_model.transformer.resblocks.20.mlp.fc2",
849
+ "vision_model.transformer.resblocks.21.attn.qkv_proj",
850
+ "vision_model.transformer.resblocks.21.attn.q_proj",
851
+ "vision_model.transformer.resblocks.21.attn.k_proj",
852
+ "vision_model.transformer.resblocks.21.attn.v_proj",
853
+ "vision_model.transformer.resblocks.21.attn.out_proj",
854
+ "vision_model.transformer.resblocks.21.mlp.fc1",
855
+ "vision_model.transformer.resblocks.21.mlp.fc2",
856
+ "vision_model.transformer.resblocks.22.attn.qkv_proj",
857
+ "vision_model.transformer.resblocks.22.attn.q_proj",
858
+ "vision_model.transformer.resblocks.22.attn.k_proj",
859
+ "vision_model.transformer.resblocks.22.attn.v_proj",
860
+ "vision_model.transformer.resblocks.22.attn.out_proj",
861
+ "vision_model.transformer.resblocks.22.mlp.fc1",
862
+ "vision_model.transformer.resblocks.22.mlp.fc2",
863
+ "vision_model.transformer.resblocks.23.attn.qkv_proj",
864
+ "vision_model.transformer.resblocks.23.attn.q_proj",
865
+ "vision_model.transformer.resblocks.23.attn.k_proj",
866
+ "vision_model.transformer.resblocks.23.attn.v_proj",
867
+ "vision_model.transformer.resblocks.23.attn.out_proj",
868
+ "vision_model.transformer.resblocks.23.mlp.fc1",
869
+ "vision_model.transformer.resblocks.23.mlp.fc2",
870
+ "vision_model.transformer.resblocks.24.attn.qkv_proj",
871
+ "vision_model.transformer.resblocks.24.attn.q_proj",
872
+ "vision_model.transformer.resblocks.24.attn.k_proj",
873
+ "vision_model.transformer.resblocks.24.attn.v_proj",
874
+ "vision_model.transformer.resblocks.24.attn.out_proj",
875
+ "vision_model.transformer.resblocks.24.mlp.fc1",
876
+ "vision_model.transformer.resblocks.24.mlp.fc2",
877
+ "vision_model.transformer.resblocks.25.attn.qkv_proj",
878
+ "vision_model.transformer.resblocks.25.attn.q_proj",
879
+ "vision_model.transformer.resblocks.25.attn.k_proj",
880
+ "vision_model.transformer.resblocks.25.attn.v_proj",
881
+ "vision_model.transformer.resblocks.25.attn.out_proj",
882
+ "vision_model.transformer.resblocks.25.mlp.fc1",
883
+ "vision_model.transformer.resblocks.25.mlp.fc2",
884
+ "vision_model.transformer.resblocks.26.attn.qkv_proj",
885
+ "vision_model.transformer.resblocks.26.attn.q_proj",
886
+ "vision_model.transformer.resblocks.26.attn.k_proj",
887
+ "vision_model.transformer.resblocks.26.attn.v_proj",
888
+ "vision_model.transformer.resblocks.26.attn.out_proj",
889
+ "vision_model.transformer.resblocks.26.mlp.fc1",
890
+ "vision_model.transformer.resblocks.26.mlp.fc2",
891
+ "vision_model.transformer.resblocks.27.attn.qkv_proj",
892
+ "vision_model.transformer.resblocks.27.attn.q_proj",
893
+ "vision_model.transformer.resblocks.27.attn.k_proj",
894
+ "vision_model.transformer.resblocks.27.attn.v_proj",
895
+ "vision_model.transformer.resblocks.27.attn.out_proj",
896
+ "vision_model.transformer.resblocks.27.mlp.fc1",
897
+ "vision_model.transformer.resblocks.27.mlp.fc2",
898
+ "vision_model.transformer.resblocks.28.attn.qkv_proj",
899
+ "vision_model.transformer.resblocks.28.attn.q_proj",
900
+ "vision_model.transformer.resblocks.28.attn.k_proj",
901
+ "vision_model.transformer.resblocks.28.attn.v_proj",
902
+ "vision_model.transformer.resblocks.28.attn.out_proj",
903
+ "vision_model.transformer.resblocks.28.mlp.fc1",
904
+ "vision_model.transformer.resblocks.28.mlp.fc2",
905
+ "vision_model.transformer.resblocks.29.attn.qkv_proj",
906
+ "vision_model.transformer.resblocks.29.attn.q_proj",
907
+ "vision_model.transformer.resblocks.29.attn.k_proj",
908
+ "vision_model.transformer.resblocks.29.attn.v_proj",
909
+ "vision_model.transformer.resblocks.29.attn.out_proj",
910
+ "vision_model.transformer.resblocks.29.mlp.fc1",
911
+ "vision_model.transformer.resblocks.29.mlp.fc2",
912
+ "vision_model.transformer.resblocks.30.attn.qkv_proj",
913
+ "vision_model.transformer.resblocks.30.attn.q_proj",
914
+ "vision_model.transformer.resblocks.30.attn.k_proj",
915
+ "vision_model.transformer.resblocks.30.attn.v_proj",
916
+ "vision_model.transformer.resblocks.30.attn.out_proj",
917
+ "vision_model.transformer.resblocks.30.mlp.fc1",
918
+ "vision_model.transformer.resblocks.30.mlp.fc2",
919
+ "vision_model.transformer.resblocks.31.attn.qkv_proj",
920
+ "vision_model.transformer.resblocks.31.attn.q_proj",
921
+ "vision_model.transformer.resblocks.31.attn.k_proj",
922
+ "vision_model.transformer.resblocks.31.attn.v_proj",
923
+ "vision_model.transformer.resblocks.31.attn.out_proj",
924
+ "vision_model.transformer.resblocks.31.mlp.fc1",
925
+ "vision_model.transformer.resblocks.31.mlp.fc2",
926
+ "vision_model.transformer.resblocks.32.attn.qkv_proj",
927
+ "vision_model.transformer.resblocks.32.attn.q_proj",
928
+ "vision_model.transformer.resblocks.32.attn.k_proj",
929
+ "vision_model.transformer.resblocks.32.attn.v_proj",
930
+ "vision_model.transformer.resblocks.32.attn.out_proj",
931
+ "vision_model.transformer.resblocks.32.mlp.fc1",
932
+ "vision_model.transformer.resblocks.32.mlp.fc2",
933
+ "vision_model.transformer.resblocks.33.attn.qkv_proj",
934
+ "vision_model.transformer.resblocks.33.attn.q_proj",
935
+ "vision_model.transformer.resblocks.33.attn.k_proj",
936
+ "vision_model.transformer.resblocks.33.attn.v_proj",
937
+ "vision_model.transformer.resblocks.33.attn.out_proj",
938
+ "vision_model.transformer.resblocks.33.mlp.fc1",
939
+ "vision_model.transformer.resblocks.33.mlp.fc2",
940
+ "vision_model.transformer.resblocks.34.attn.qkv_proj",
941
+ "vision_model.transformer.resblocks.34.attn.q_proj",
942
+ "vision_model.transformer.resblocks.34.attn.k_proj",
943
+ "vision_model.transformer.resblocks.34.attn.v_proj",
944
+ "vision_model.transformer.resblocks.34.attn.out_proj",
945
+ "vision_model.transformer.resblocks.34.mlp.fc1",
946
+ "vision_model.transformer.resblocks.34.mlp.fc2",
947
+ "vision_model.transformer.resblocks.35.attn.qkv_proj",
948
+ "vision_model.transformer.resblocks.35.attn.q_proj",
949
+ "vision_model.transformer.resblocks.35.attn.k_proj",
950
+ "vision_model.transformer.resblocks.35.attn.v_proj",
951
+ "vision_model.transformer.resblocks.35.attn.out_proj",
952
+ "vision_model.transformer.resblocks.35.mlp.fc1",
953
+ "vision_model.transformer.resblocks.35.mlp.fc2",
954
+ "vision_model.transformer.resblocks.36.attn.qkv_proj",
955
+ "vision_model.transformer.resblocks.36.attn.q_proj",
956
+ "vision_model.transformer.resblocks.36.attn.k_proj",
957
+ "vision_model.transformer.resblocks.36.attn.v_proj",
958
+ "vision_model.transformer.resblocks.36.attn.out_proj",
959
+ "vision_model.transformer.resblocks.36.mlp.fc1",
960
+ "vision_model.transformer.resblocks.36.mlp.fc2",
961
+ "vision_model.transformer.resblocks.37.attn.qkv_proj",
962
+ "vision_model.transformer.resblocks.37.attn.q_proj",
963
+ "vision_model.transformer.resblocks.37.attn.k_proj",
964
+ "vision_model.transformer.resblocks.37.attn.v_proj",
965
+ "vision_model.transformer.resblocks.37.attn.out_proj",
966
+ "vision_model.transformer.resblocks.37.mlp.fc1",
967
+ "vision_model.transformer.resblocks.37.mlp.fc2",
968
+ "vision_model.transformer.resblocks.38.attn.qkv_proj",
969
+ "vision_model.transformer.resblocks.38.attn.q_proj",
970
+ "vision_model.transformer.resblocks.38.attn.k_proj",
971
+ "vision_model.transformer.resblocks.38.attn.v_proj",
972
+ "vision_model.transformer.resblocks.38.attn.out_proj",
973
+ "vision_model.transformer.resblocks.38.mlp.fc1",
974
+ "vision_model.transformer.resblocks.38.mlp.fc2",
975
+ "vision_model.transformer.resblocks.39.attn.qkv_proj",
976
+ "vision_model.transformer.resblocks.39.attn.q_proj",
977
+ "vision_model.transformer.resblocks.39.attn.k_proj",
978
+ "vision_model.transformer.resblocks.39.attn.v_proj",
979
+ "vision_model.transformer.resblocks.39.attn.out_proj",
980
+ "vision_model.transformer.resblocks.39.mlp.fc1",
981
+ "vision_model.transformer.resblocks.39.mlp.fc2",
982
+ "vision_model.transformer.resblocks.40.attn.qkv_proj",
983
+ "vision_model.transformer.resblocks.40.attn.q_proj",
984
+ "vision_model.transformer.resblocks.40.attn.k_proj",
985
+ "vision_model.transformer.resblocks.40.attn.v_proj",
986
+ "vision_model.transformer.resblocks.40.attn.out_proj",
987
+ "vision_model.transformer.resblocks.40.mlp.fc1",
988
+ "vision_model.transformer.resblocks.40.mlp.fc2",
989
+ "vision_model.transformer.resblocks.41.attn.qkv_proj",
990
+ "vision_model.transformer.resblocks.41.attn.q_proj",
991
+ "vision_model.transformer.resblocks.41.attn.k_proj",
992
+ "vision_model.transformer.resblocks.41.attn.v_proj",
993
+ "vision_model.transformer.resblocks.41.attn.out_proj",
994
+ "vision_model.transformer.resblocks.41.mlp.fc1",
995
+ "vision_model.transformer.resblocks.41.mlp.fc2",
996
+ "vision_model.transformer.resblocks.42.attn.qkv_proj",
997
+ "vision_model.transformer.resblocks.42.attn.q_proj",
998
+ "vision_model.transformer.resblocks.42.attn.k_proj",
999
+ "vision_model.transformer.resblocks.42.attn.v_proj",
1000
+ "vision_model.transformer.resblocks.42.attn.out_proj",
1001
+ "vision_model.transformer.resblocks.42.mlp.fc1",
1002
+ "vision_model.transformer.resblocks.42.mlp.fc2",
1003
+ "vision_model.transformer.resblocks.43.attn.qkv_proj",
1004
+ "vision_model.transformer.resblocks.43.attn.q_proj",
1005
+ "vision_model.transformer.resblocks.43.attn.k_proj",
1006
+ "vision_model.transformer.resblocks.43.attn.v_proj",
1007
+ "vision_model.transformer.resblocks.43.attn.out_proj",
1008
+ "vision_model.transformer.resblocks.43.mlp.fc1",
1009
+ "vision_model.transformer.resblocks.43.mlp.fc2",
1010
+ "vision_model.transformer.resblocks.44.attn.qkv_proj",
1011
+ "vision_model.transformer.resblocks.44.attn.q_proj",
1012
+ "vision_model.transformer.resblocks.44.attn.k_proj",
1013
+ "vision_model.transformer.resblocks.44.attn.v_proj",
1014
+ "vision_model.transformer.resblocks.44.attn.out_proj",
1015
+ "vision_model.transformer.resblocks.44.mlp.fc1",
1016
+ "vision_model.transformer.resblocks.44.mlp.fc2",
1017
+ "vision_model.transformer.resblocks.45.attn.qkv_proj",
1018
+ "vision_model.transformer.resblocks.45.attn.q_proj",
1019
+ "vision_model.transformer.resblocks.45.attn.k_proj",
1020
+ "vision_model.transformer.resblocks.45.attn.v_proj",
1021
+ "vision_model.transformer.resblocks.45.attn.out_proj",
1022
+ "vision_model.transformer.resblocks.45.mlp.fc1",
1023
+ "vision_model.transformer.resblocks.45.mlp.fc2",
1024
+ "vision_model.transformer.resblocks.46.attn.qkv_proj",
1025
+ "vision_model.transformer.resblocks.46.attn.q_proj",
1026
+ "vision_model.transformer.resblocks.46.attn.k_proj",
1027
+ "vision_model.transformer.resblocks.46.attn.v_proj",
1028
+ "vision_model.transformer.resblocks.46.attn.out_proj",
1029
+ "vision_model.transformer.resblocks.46.mlp.fc1",
1030
+ "vision_model.transformer.resblocks.46.mlp.fc2",
1031
+ "vit_large_projector",
1032
+ "lm_head",
1033
+ "model.embed_tokens",
1034
+ "model.norm",
1035
+ "model.layers.0.self_attn.g_proj",
1036
+ "model.layers.0.self_attn.qkv_proj",
1037
+ "model.layers.0.self_attn.q_proj",
1038
+ "model.layers.0.self_attn.k_proj",
1039
+ "model.layers.0.self_attn.v_proj",
1040
+ "model.layers.0.self_attn.o_proj",
1041
+ "model.layers.0.mlp.gate_up_proj",
1042
+ "model.layers.0.mlp.gate_proj",
1043
+ "model.layers.0.mlp.up_proj",
1044
+ "model.layers.0.mlp.down_proj",
1045
+ "model.layers.1.self_attn.g_proj",
1046
+ "model.layers.1.self_attn.qkv_proj",
1047
+ "model.layers.1.self_attn.q_proj",
1048
+ "model.layers.1.self_attn.k_proj",
1049
+ "model.layers.1.self_attn.v_proj",
1050
+ "model.layers.1.self_attn.o_proj",
1051
+ "model.layers.1.mlp.gate_up_proj",
1052
+ "model.layers.1.mlp.gate_proj",
1053
+ "model.layers.1.mlp.up_proj",
1054
+ "model.layers.1.mlp.down_proj",
1055
+ "model.layers.2.self_attn.g_proj",
1056
+ "model.layers.2.self_attn.qkv_proj",
1057
+ "model.layers.2.self_attn.q_proj",
1058
+ "model.layers.2.self_attn.k_proj",
1059
+ "model.layers.2.self_attn.v_proj",
1060
+ "model.layers.2.self_attn.o_proj",
1061
+ "model.layers.2.mlp.gate_up_proj",
1062
+ "model.layers.2.mlp.gate_proj",
1063
+ "model.layers.2.mlp.up_proj",
1064
+ "model.layers.2.mlp.down_proj",
1065
+ "model.layers.3.self_attn.g_proj",
1066
+ "model.layers.3.self_attn.qkv_proj",
1067
+ "model.layers.3.self_attn.q_proj",
1068
+ "model.layers.3.self_attn.k_proj",
1069
+ "model.layers.3.self_attn.v_proj",
1070
+ "model.layers.3.self_attn.o_proj",
1071
+ "model.layers.3.self_attn.qkv_indexer_proj",
1072
+ "model.layers.3.moe.gate",
1073
+ "model.layers.3.share_expert.gate_up_proj",
1074
+ "model.layers.3.share_expert.gate_proj",
1075
+ "model.layers.3.share_expert.up_proj",
1076
+ "model.layers.3.share_expert.down_proj",
1077
+ "model.layers.4.self_attn.g_proj",
1078
+ "model.layers.4.self_attn.qkv_proj",
1079
+ "model.layers.4.self_attn.q_proj",
1080
+ "model.layers.4.self_attn.k_proj",
1081
+ "model.layers.4.self_attn.v_proj",
1082
+ "model.layers.4.self_attn.o_proj",
1083
+ "model.layers.4.moe.gate",
1084
+ "model.layers.4.share_expert.gate_up_proj",
1085
+ "model.layers.4.share_expert.gate_proj",
1086
+ "model.layers.4.share_expert.up_proj",
1087
+ "model.layers.4.share_expert.down_proj",
1088
+ "model.layers.5.self_attn.g_proj",
1089
+ "model.layers.5.self_attn.qkv_proj",
1090
+ "model.layers.5.self_attn.q_proj",
1091
+ "model.layers.5.self_attn.k_proj",
1092
+ "model.layers.5.self_attn.v_proj",
1093
+ "model.layers.5.self_attn.o_proj",
1094
+ "model.layers.5.moe.gate",
1095
+ "model.layers.5.share_expert.gate_up_proj",
1096
+ "model.layers.5.share_expert.gate_proj",
1097
+ "model.layers.5.share_expert.up_proj",
1098
+ "model.layers.5.share_expert.down_proj",
1099
+ "model.layers.6.self_attn.g_proj",
1100
+ "model.layers.6.self_attn.qkv_proj",
1101
+ "model.layers.6.self_attn.q_proj",
1102
+ "model.layers.6.self_attn.k_proj",
1103
+ "model.layers.6.self_attn.v_proj",
1104
+ "model.layers.6.self_attn.o_proj",
1105
+ "model.layers.6.moe.gate",
1106
+ "model.layers.6.share_expert.gate_up_proj",
1107
+ "model.layers.6.share_expert.gate_proj",
1108
+ "model.layers.6.share_expert.up_proj",
1109
+ "model.layers.6.share_expert.down_proj",
1110
+ "model.layers.7.self_attn.g_proj",
1111
+ "model.layers.7.self_attn.qkv_proj",
1112
+ "model.layers.7.self_attn.q_proj",
1113
+ "model.layers.7.self_attn.k_proj",
1114
+ "model.layers.7.self_attn.v_proj",
1115
+ "model.layers.7.self_attn.o_proj",
1116
+ "model.layers.7.self_attn.qkv_indexer_proj",
1117
+ "model.layers.7.moe.gate",
1118
+ "model.layers.7.share_expert.gate_up_proj",
1119
+ "model.layers.7.share_expert.gate_proj",
1120
+ "model.layers.7.share_expert.up_proj",
1121
+ "model.layers.7.share_expert.down_proj",
1122
+ "model.layers.8.self_attn.g_proj",
1123
+ "model.layers.8.self_attn.qkv_proj",
1124
+ "model.layers.8.self_attn.q_proj",
1125
+ "model.layers.8.self_attn.k_proj",
1126
+ "model.layers.8.self_attn.v_proj",
1127
+ "model.layers.8.self_attn.o_proj",
1128
+ "model.layers.8.moe.gate",
1129
+ "model.layers.8.share_expert.gate_up_proj",
1130
+ "model.layers.8.share_expert.gate_proj",
1131
+ "model.layers.8.share_expert.up_proj",
1132
+ "model.layers.8.share_expert.down_proj",
1133
+ "model.layers.9.self_attn.g_proj",
1134
+ "model.layers.9.self_attn.qkv_proj",
1135
+ "model.layers.9.self_attn.q_proj",
1136
+ "model.layers.9.self_attn.k_proj",
1137
+ "model.layers.9.self_attn.v_proj",
1138
+ "model.layers.9.self_attn.o_proj",
1139
+ "model.layers.9.moe.gate",
1140
+ "model.layers.9.share_expert.gate_up_proj",
1141
+ "model.layers.9.share_expert.gate_proj",
1142
+ "model.layers.9.share_expert.up_proj",
1143
+ "model.layers.9.share_expert.down_proj",
1144
+ "model.layers.10.self_attn.g_proj",
1145
+ "model.layers.10.self_attn.qkv_proj",
1146
+ "model.layers.10.self_attn.q_proj",
1147
+ "model.layers.10.self_attn.k_proj",
1148
+ "model.layers.10.self_attn.v_proj",
1149
+ "model.layers.10.self_attn.o_proj",
1150
+ "model.layers.10.moe.gate",
1151
+ "model.layers.10.share_expert.gate_up_proj",
1152
+ "model.layers.10.share_expert.gate_proj",
1153
+ "model.layers.10.share_expert.up_proj",
1154
+ "model.layers.10.share_expert.down_proj",
1155
+ "model.layers.11.self_attn.g_proj",
1156
+ "model.layers.11.self_attn.qkv_proj",
1157
+ "model.layers.11.self_attn.q_proj",
1158
+ "model.layers.11.self_attn.k_proj",
1159
+ "model.layers.11.self_attn.v_proj",
1160
+ "model.layers.11.self_attn.o_proj",
1161
+ "model.layers.11.self_attn.qkv_indexer_proj",
1162
+ "model.layers.11.moe.gate",
1163
+ "model.layers.11.share_expert.gate_up_proj",
1164
+ "model.layers.11.share_expert.gate_proj",
1165
+ "model.layers.11.share_expert.up_proj",
1166
+ "model.layers.11.share_expert.down_proj",
1167
+ "model.layers.12.self_attn.g_proj",
1168
+ "model.layers.12.self_attn.qkv_proj",
1169
+ "model.layers.12.self_attn.q_proj",
1170
+ "model.layers.12.self_attn.k_proj",
1171
+ "model.layers.12.self_attn.v_proj",
1172
+ "model.layers.12.self_attn.o_proj",
1173
+ "model.layers.12.moe.gate",
1174
+ "model.layers.12.share_expert.gate_up_proj",
1175
+ "model.layers.12.share_expert.gate_proj",
1176
+ "model.layers.12.share_expert.up_proj",
1177
+ "model.layers.12.share_expert.down_proj",
1178
+ "model.layers.13.self_attn.g_proj",
1179
+ "model.layers.13.self_attn.qkv_proj",
1180
+ "model.layers.13.self_attn.q_proj",
1181
+ "model.layers.13.self_attn.k_proj",
1182
+ "model.layers.13.self_attn.v_proj",
1183
+ "model.layers.13.self_attn.o_proj",
1184
+ "model.layers.13.moe.gate",
1185
+ "model.layers.13.share_expert.gate_up_proj",
1186
+ "model.layers.13.share_expert.gate_proj",
1187
+ "model.layers.13.share_expert.up_proj",
1188
+ "model.layers.13.share_expert.down_proj",
1189
+ "model.layers.14.self_attn.g_proj",
1190
+ "model.layers.14.self_attn.qkv_proj",
1191
+ "model.layers.14.self_attn.q_proj",
1192
+ "model.layers.14.self_attn.k_proj",
1193
+ "model.layers.14.self_attn.v_proj",
1194
+ "model.layers.14.self_attn.o_proj",
1195
+ "model.layers.14.moe.gate",
1196
+ "model.layers.14.share_expert.gate_up_proj",
1197
+ "model.layers.14.share_expert.gate_proj",
1198
+ "model.layers.14.share_expert.up_proj",
1199
+ "model.layers.14.share_expert.down_proj",
1200
+ "model.layers.15.self_attn.g_proj",
1201
+ "model.layers.15.self_attn.qkv_proj",
1202
+ "model.layers.15.self_attn.q_proj",
1203
+ "model.layers.15.self_attn.k_proj",
1204
+ "model.layers.15.self_attn.v_proj",
1205
+ "model.layers.15.self_attn.o_proj",
1206
+ "model.layers.15.self_attn.qkv_indexer_proj",
1207
+ "model.layers.15.moe.gate",
1208
+ "model.layers.15.share_expert.gate_up_proj",
1209
+ "model.layers.15.share_expert.gate_proj",
1210
+ "model.layers.15.share_expert.up_proj",
1211
+ "model.layers.15.share_expert.down_proj",
1212
+ "model.layers.16.self_attn.g_proj",
1213
+ "model.layers.16.self_attn.qkv_proj",
1214
+ "model.layers.16.self_attn.q_proj",
1215
+ "model.layers.16.self_attn.k_proj",
1216
+ "model.layers.16.self_attn.v_proj",
1217
+ "model.layers.16.self_attn.o_proj",
1218
+ "model.layers.16.moe.gate",
1219
+ "model.layers.16.share_expert.gate_up_proj",
1220
+ "model.layers.16.share_expert.gate_proj",
1221
+ "model.layers.16.share_expert.up_proj",
1222
+ "model.layers.16.share_expert.down_proj",
1223
+ "model.layers.17.self_attn.g_proj",
1224
+ "model.layers.17.self_attn.qkv_proj",
1225
+ "model.layers.17.self_attn.q_proj",
1226
+ "model.layers.17.self_attn.k_proj",
1227
+ "model.layers.17.self_attn.v_proj",
1228
+ "model.layers.17.self_attn.o_proj",
1229
+ "model.layers.17.moe.gate",
1230
+ "model.layers.17.share_expert.gate_up_proj",
1231
+ "model.layers.17.share_expert.gate_proj",
1232
+ "model.layers.17.share_expert.up_proj",
1233
+ "model.layers.17.share_expert.down_proj",
1234
+ "model.layers.18.self_attn.g_proj",
1235
+ "model.layers.18.self_attn.qkv_proj",
1236
+ "model.layers.18.self_attn.q_proj",
1237
+ "model.layers.18.self_attn.k_proj",
1238
+ "model.layers.18.self_attn.v_proj",
1239
+ "model.layers.18.self_attn.o_proj",
1240
+ "model.layers.18.moe.gate",
1241
+ "model.layers.18.share_expert.gate_up_proj",
1242
+ "model.layers.18.share_expert.gate_proj",
1243
+ "model.layers.18.share_expert.up_proj",
1244
+ "model.layers.18.share_expert.down_proj",
1245
+ "model.layers.19.self_attn.g_proj",
1246
+ "model.layers.19.self_attn.qkv_proj",
1247
+ "model.layers.19.self_attn.q_proj",
1248
+ "model.layers.19.self_attn.k_proj",
1249
+ "model.layers.19.self_attn.v_proj",
1250
+ "model.layers.19.self_attn.o_proj",
1251
+ "model.layers.19.self_attn.qkv_indexer_proj",
1252
+ "model.layers.19.moe.gate",
1253
+ "model.layers.19.share_expert.gate_up_proj",
1254
+ "model.layers.19.share_expert.gate_proj",
1255
+ "model.layers.19.share_expert.up_proj",
1256
+ "model.layers.19.share_expert.down_proj",
1257
+ "model.layers.20.self_attn.g_proj",
1258
+ "model.layers.20.self_attn.qkv_proj",
1259
+ "model.layers.20.self_attn.q_proj",
1260
+ "model.layers.20.self_attn.k_proj",
1261
+ "model.layers.20.self_attn.v_proj",
1262
+ "model.layers.20.self_attn.o_proj",
1263
+ "model.layers.20.moe.gate",
1264
+ "model.layers.20.share_expert.gate_up_proj",
1265
+ "model.layers.20.share_expert.gate_proj",
1266
+ "model.layers.20.share_expert.up_proj",
1267
+ "model.layers.20.share_expert.down_proj",
1268
+ "model.layers.21.self_attn.g_proj",
1269
+ "model.layers.21.self_attn.qkv_proj",
1270
+ "model.layers.21.self_attn.q_proj",
1271
+ "model.layers.21.self_attn.k_proj",
1272
+ "model.layers.21.self_attn.v_proj",
1273
+ "model.layers.21.self_attn.o_proj",
1274
+ "model.layers.21.moe.gate",
1275
+ "model.layers.21.share_expert.gate_up_proj",
1276
+ "model.layers.21.share_expert.gate_proj",
1277
+ "model.layers.21.share_expert.up_proj",
1278
+ "model.layers.21.share_expert.down_proj",
1279
+ "model.layers.22.self_attn.g_proj",
1280
+ "model.layers.22.self_attn.qkv_proj",
1281
+ "model.layers.22.self_attn.q_proj",
1282
+ "model.layers.22.self_attn.k_proj",
1283
+ "model.layers.22.self_attn.v_proj",
1284
+ "model.layers.22.self_attn.o_proj",
1285
+ "model.layers.22.moe.gate",
1286
+ "model.layers.22.share_expert.gate_up_proj",
1287
+ "model.layers.22.share_expert.gate_proj",
1288
+ "model.layers.22.share_expert.up_proj",
1289
+ "model.layers.22.share_expert.down_proj",
1290
+ "model.layers.23.self_attn.g_proj",
1291
+ "model.layers.23.self_attn.qkv_proj",
1292
+ "model.layers.23.self_attn.q_proj",
1293
+ "model.layers.23.self_attn.k_proj",
1294
+ "model.layers.23.self_attn.v_proj",
1295
+ "model.layers.23.self_attn.o_proj",
1296
+ "model.layers.23.self_attn.qkv_indexer_proj",
1297
+ "model.layers.23.moe.gate",
1298
+ "model.layers.23.share_expert.gate_up_proj",
1299
+ "model.layers.23.share_expert.gate_proj",
1300
+ "model.layers.23.share_expert.up_proj",
1301
+ "model.layers.23.share_expert.down_proj",
1302
+ "model.layers.24.self_attn.g_proj",
1303
+ "model.layers.24.self_attn.qkv_proj",
1304
+ "model.layers.24.self_attn.q_proj",
1305
+ "model.layers.24.self_attn.k_proj",
1306
+ "model.layers.24.self_attn.v_proj",
1307
+ "model.layers.24.self_attn.o_proj",
1308
+ "model.layers.24.moe.gate",
1309
+ "model.layers.24.share_expert.gate_up_proj",
1310
+ "model.layers.24.share_expert.gate_proj",
1311
+ "model.layers.24.share_expert.up_proj",
1312
+ "model.layers.24.share_expert.down_proj",
1313
+ "model.layers.25.self_attn.g_proj",
1314
+ "model.layers.25.self_attn.qkv_proj",
1315
+ "model.layers.25.self_attn.q_proj",
1316
+ "model.layers.25.self_attn.k_proj",
1317
+ "model.layers.25.self_attn.v_proj",
1318
+ "model.layers.25.self_attn.o_proj",
1319
+ "model.layers.25.moe.gate",
1320
+ "model.layers.25.share_expert.gate_up_proj",
1321
+ "model.layers.25.share_expert.gate_proj",
1322
+ "model.layers.25.share_expert.up_proj",
1323
+ "model.layers.25.share_expert.down_proj",
1324
+ "model.layers.26.self_attn.g_proj",
1325
+ "model.layers.26.self_attn.qkv_proj",
1326
+ "model.layers.26.self_attn.q_proj",
1327
+ "model.layers.26.self_attn.k_proj",
1328
+ "model.layers.26.self_attn.v_proj",
1329
+ "model.layers.26.self_attn.o_proj",
1330
+ "model.layers.26.moe.gate",
1331
+ "model.layers.26.share_expert.gate_up_proj",
1332
+ "model.layers.26.share_expert.gate_proj",
1333
+ "model.layers.26.share_expert.up_proj",
1334
+ "model.layers.26.share_expert.down_proj",
1335
+ "model.layers.27.self_attn.g_proj",
1336
+ "model.layers.27.self_attn.qkv_proj",
1337
+ "model.layers.27.self_attn.q_proj",
1338
+ "model.layers.27.self_attn.k_proj",
1339
+ "model.layers.27.self_attn.v_proj",
1340
+ "model.layers.27.self_attn.o_proj",
1341
+ "model.layers.27.self_attn.qkv_indexer_proj",
1342
+ "model.layers.27.moe.gate",
1343
+ "model.layers.27.share_expert.gate_up_proj",
1344
+ "model.layers.27.share_expert.gate_proj",
1345
+ "model.layers.27.share_expert.up_proj",
1346
+ "model.layers.27.share_expert.down_proj",
1347
+ "model.layers.28.self_attn.g_proj",
1348
+ "model.layers.28.self_attn.qkv_proj",
1349
+ "model.layers.28.self_attn.q_proj",
1350
+ "model.layers.28.self_attn.k_proj",
1351
+ "model.layers.28.self_attn.v_proj",
1352
+ "model.layers.28.self_attn.o_proj",
1353
+ "model.layers.28.moe.gate",
1354
+ "model.layers.28.share_expert.gate_up_proj",
1355
+ "model.layers.28.share_expert.gate_proj",
1356
+ "model.layers.28.share_expert.up_proj",
1357
+ "model.layers.28.share_expert.down_proj",
1358
+ "model.layers.29.self_attn.g_proj",
1359
+ "model.layers.29.self_attn.qkv_proj",
1360
+ "model.layers.29.self_attn.q_proj",
1361
+ "model.layers.29.self_attn.k_proj",
1362
+ "model.layers.29.self_attn.v_proj",
1363
+ "model.layers.29.self_attn.o_proj",
1364
+ "model.layers.29.moe.gate",
1365
+ "model.layers.29.share_expert.gate_up_proj",
1366
+ "model.layers.29.share_expert.gate_proj",
1367
+ "model.layers.29.share_expert.up_proj",
1368
+ "model.layers.29.share_expert.down_proj",
1369
+ "model.layers.30.self_attn.g_proj",
1370
+ "model.layers.30.self_attn.qkv_proj",
1371
+ "model.layers.30.self_attn.q_proj",
1372
+ "model.layers.30.self_attn.k_proj",
1373
+ "model.layers.30.self_attn.v_proj",
1374
+ "model.layers.30.self_attn.o_proj",
1375
+ "model.layers.30.moe.gate",
1376
+ "model.layers.30.share_expert.gate_up_proj",
1377
+ "model.layers.30.share_expert.gate_proj",
1378
+ "model.layers.30.share_expert.up_proj",
1379
+ "model.layers.30.share_expert.down_proj",
1380
+ "model.layers.31.self_attn.g_proj",
1381
+ "model.layers.31.self_attn.qkv_proj",
1382
+ "model.layers.31.self_attn.q_proj",
1383
+ "model.layers.31.self_attn.k_proj",
1384
+ "model.layers.31.self_attn.v_proj",
1385
+ "model.layers.31.self_attn.o_proj",
1386
+ "model.layers.31.self_attn.qkv_indexer_proj",
1387
+ "model.layers.31.moe.gate",
1388
+ "model.layers.31.share_expert.gate_up_proj",
1389
+ "model.layers.31.share_expert.gate_proj",
1390
+ "model.layers.31.share_expert.up_proj",
1391
+ "model.layers.31.share_expert.down_proj",
1392
+ "model.layers.32.self_attn.g_proj",
1393
+ "model.layers.32.self_attn.qkv_proj",
1394
+ "model.layers.32.self_attn.q_proj",
1395
+ "model.layers.32.self_attn.k_proj",
1396
+ "model.layers.32.self_attn.v_proj",
1397
+ "model.layers.32.self_attn.o_proj",
1398
+ "model.layers.32.moe.gate",
1399
+ "model.layers.32.share_expert.gate_up_proj",
1400
+ "model.layers.32.share_expert.gate_proj",
1401
+ "model.layers.32.share_expert.up_proj",
1402
+ "model.layers.32.share_expert.down_proj",
1403
+ "model.layers.33.self_attn.g_proj",
1404
+ "model.layers.33.self_attn.qkv_proj",
1405
+ "model.layers.33.self_attn.q_proj",
1406
+ "model.layers.33.self_attn.k_proj",
1407
+ "model.layers.33.self_attn.v_proj",
1408
+ "model.layers.33.self_attn.o_proj",
1409
+ "model.layers.33.moe.gate",
1410
+ "model.layers.33.share_expert.gate_up_proj",
1411
+ "model.layers.33.share_expert.gate_proj",
1412
+ "model.layers.33.share_expert.up_proj",
1413
+ "model.layers.33.share_expert.down_proj",
1414
+ "model.layers.34.self_attn.g_proj",
1415
+ "model.layers.34.self_attn.qkv_proj",
1416
+ "model.layers.34.self_attn.q_proj",
1417
+ "model.layers.34.self_attn.k_proj",
1418
+ "model.layers.34.self_attn.v_proj",
1419
+ "model.layers.34.self_attn.o_proj",
1420
+ "model.layers.34.moe.gate",
1421
+ "model.layers.34.share_expert.gate_up_proj",
1422
+ "model.layers.34.share_expert.gate_proj",
1423
+ "model.layers.34.share_expert.up_proj",
1424
+ "model.layers.34.share_expert.down_proj",
1425
+ "model.layers.35.self_attn.g_proj",
1426
+ "model.layers.35.self_attn.qkv_proj",
1427
+ "model.layers.35.self_attn.q_proj",
1428
+ "model.layers.35.self_attn.k_proj",
1429
+ "model.layers.35.self_attn.v_proj",
1430
+ "model.layers.35.self_attn.o_proj",
1431
+ "model.layers.35.self_attn.qkv_indexer_proj",
1432
+ "model.layers.35.moe.gate",
1433
+ "model.layers.35.share_expert.gate_up_proj",
1434
+ "model.layers.35.share_expert.gate_proj",
1435
+ "model.layers.35.share_expert.up_proj",
1436
+ "model.layers.35.share_expert.down_proj",
1437
+ "model.layers.36.self_attn.g_proj",
1438
+ "model.layers.36.self_attn.qkv_proj",
1439
+ "model.layers.36.self_attn.q_proj",
1440
+ "model.layers.36.self_attn.k_proj",
1441
+ "model.layers.36.self_attn.v_proj",
1442
+ "model.layers.36.self_attn.o_proj",
1443
+ "model.layers.36.moe.gate",
1444
+ "model.layers.36.share_expert.gate_up_proj",
1445
+ "model.layers.36.share_expert.gate_proj",
1446
+ "model.layers.36.share_expert.up_proj",
1447
+ "model.layers.36.share_expert.down_proj",
1448
+ "model.layers.37.self_attn.g_proj",
1449
+ "model.layers.37.self_attn.qkv_proj",
1450
+ "model.layers.37.self_attn.q_proj",
1451
+ "model.layers.37.self_attn.k_proj",
1452
+ "model.layers.37.self_attn.v_proj",
1453
+ "model.layers.37.self_attn.o_proj",
1454
+ "model.layers.37.moe.gate",
1455
+ "model.layers.37.share_expert.gate_up_proj",
1456
+ "model.layers.37.share_expert.gate_proj",
1457
+ "model.layers.37.share_expert.up_proj",
1458
+ "model.layers.37.share_expert.down_proj",
1459
+ "model.layers.38.self_attn.g_proj",
1460
+ "model.layers.38.self_attn.qkv_proj",
1461
+ "model.layers.38.self_attn.q_proj",
1462
+ "model.layers.38.self_attn.k_proj",
1463
+ "model.layers.38.self_attn.v_proj",
1464
+ "model.layers.38.self_attn.o_proj",
1465
+ "model.layers.38.moe.gate",
1466
+ "model.layers.38.share_expert.gate_up_proj",
1467
+ "model.layers.38.share_expert.gate_proj",
1468
+ "model.layers.38.share_expert.up_proj",
1469
+ "model.layers.38.share_expert.down_proj",
1470
+ "model.layers.39.self_attn.g_proj",
1471
+ "model.layers.39.self_attn.qkv_proj",
1472
+ "model.layers.39.self_attn.q_proj",
1473
+ "model.layers.39.self_attn.k_proj",
1474
+ "model.layers.39.self_attn.v_proj",
1475
+ "model.layers.39.self_attn.o_proj",
1476
+ "model.layers.39.self_attn.qkv_indexer_proj",
1477
+ "model.layers.39.moe.gate",
1478
+ "model.layers.39.share_expert.gate_up_proj",
1479
+ "model.layers.39.share_expert.gate_proj",
1480
+ "model.layers.39.share_expert.up_proj",
1481
+ "model.layers.39.share_expert.down_proj",
1482
+ "model.layers.40.self_attn.g_proj",
1483
+ "model.layers.40.self_attn.qkv_proj",
1484
+ "model.layers.40.self_attn.q_proj",
1485
+ "model.layers.40.self_attn.k_proj",
1486
+ "model.layers.40.self_attn.v_proj",
1487
+ "model.layers.40.self_attn.o_proj",
1488
+ "model.layers.40.moe.gate",
1489
+ "model.layers.40.share_expert.gate_up_proj",
1490
+ "model.layers.40.share_expert.gate_proj",
1491
+ "model.layers.40.share_expert.up_proj",
1492
+ "model.layers.40.share_expert.down_proj",
1493
+ "model.layers.41.self_attn.g_proj",
1494
+ "model.layers.41.self_attn.qkv_proj",
1495
+ "model.layers.41.self_attn.q_proj",
1496
+ "model.layers.41.self_attn.k_proj",
1497
+ "model.layers.41.self_attn.v_proj",
1498
+ "model.layers.41.self_attn.o_proj",
1499
+ "model.layers.41.moe.gate",
1500
+ "model.layers.41.share_expert.gate_up_proj",
1501
+ "model.layers.41.share_expert.gate_proj",
1502
+ "model.layers.41.share_expert.up_proj",
1503
+ "model.layers.41.share_expert.down_proj",
1504
+ "model.layers.42.self_attn.g_proj",
1505
+ "model.layers.42.self_attn.qkv_proj",
1506
+ "model.layers.42.self_attn.q_proj",
1507
+ "model.layers.42.self_attn.k_proj",
1508
+ "model.layers.42.self_attn.v_proj",
1509
+ "model.layers.42.self_attn.o_proj",
1510
+ "model.layers.42.moe.gate",
1511
+ "model.layers.42.share_expert.gate_up_proj",
1512
+ "model.layers.42.share_expert.gate_proj",
1513
+ "model.layers.42.share_expert.up_proj",
1514
+ "model.layers.42.share_expert.down_proj",
1515
+ "model.layers.43.self_attn.g_proj",
1516
+ "model.layers.43.self_attn.qkv_proj",
1517
+ "model.layers.43.self_attn.q_proj",
1518
+ "model.layers.43.self_attn.k_proj",
1519
+ "model.layers.43.self_attn.v_proj",
1520
+ "model.layers.43.self_attn.o_proj",
1521
+ "model.layers.43.self_attn.qkv_indexer_proj",
1522
+ "model.layers.43.moe.gate",
1523
+ "model.layers.43.share_expert.gate_up_proj",
1524
+ "model.layers.43.share_expert.gate_proj",
1525
+ "model.layers.43.share_expert.up_proj",
1526
+ "model.layers.43.share_expert.down_proj",
1527
+ "model.layers.44.self_attn.g_proj",
1528
+ "model.layers.44.self_attn.qkv_proj",
1529
+ "model.layers.44.self_attn.q_proj",
1530
+ "model.layers.44.self_attn.k_proj",
1531
+ "model.layers.44.self_attn.v_proj",
1532
+ "model.layers.44.self_attn.o_proj",
1533
+ "model.layers.44.moe.gate",
1534
+ "model.layers.44.share_expert.gate_up_proj",
1535
+ "model.layers.44.share_expert.gate_proj",
1536
+ "model.layers.44.share_expert.up_proj",
1537
+ "model.layers.44.share_expert.down_proj",
1538
+ "model.layers.45.self_attn.g_proj",
1539
+ "model.layers.45.self_attn.qkv_proj",
1540
+ "model.layers.45.self_attn.q_proj",
1541
+ "model.layers.45.self_attn.k_proj",
1542
+ "model.layers.45.self_attn.v_proj",
1543
+ "model.layers.45.self_attn.o_proj",
1544
+ "model.layers.45.moe.gate",
1545
+ "model.layers.45.share_expert.gate_up_proj",
1546
+ "model.layers.45.share_expert.gate_proj",
1547
+ "model.layers.45.share_expert.up_proj",
1548
+ "model.layers.45.share_expert.down_proj",
1549
+ "model.layers.46.self_attn.g_proj",
1550
+ "model.layers.46.self_attn.qkv_proj",
1551
+ "model.layers.46.self_attn.q_proj",
1552
+ "model.layers.46.self_attn.k_proj",
1553
+ "model.layers.46.self_attn.v_proj",
1554
+ "model.layers.46.self_attn.o_proj",
1555
+ "model.layers.46.moe.gate",
1556
+ "model.layers.46.share_expert.gate_up_proj",
1557
+ "model.layers.46.share_expert.gate_proj",
1558
+ "model.layers.46.share_expert.up_proj",
1559
+ "model.layers.46.share_expert.down_proj",
1560
+ "model.layers.47.self_attn.g_proj",
1561
+ "model.layers.47.self_attn.qkv_proj",
1562
+ "model.layers.47.self_attn.q_proj",
1563
+ "model.layers.47.self_attn.k_proj",
1564
+ "model.layers.47.self_attn.v_proj",
1565
+ "model.layers.47.self_attn.o_proj",
1566
+ "model.layers.47.self_attn.qkv_indexer_proj",
1567
+ "model.layers.47.moe.gate",
1568
+ "model.layers.47.share_expert.gate_up_proj",
1569
+ "model.layers.47.share_expert.gate_proj",
1570
+ "model.layers.47.share_expert.up_proj",
1571
+ "model.layers.47.share_expert.down_proj",
1572
+ "model.layers.48.self_attn.g_proj",
1573
+ "model.layers.48.self_attn.qkv_proj",
1574
+ "model.layers.48.self_attn.q_proj",
1575
+ "model.layers.48.self_attn.k_proj",
1576
+ "model.layers.48.self_attn.v_proj",
1577
+ "model.layers.48.self_attn.o_proj",
1578
+ "model.layers.48.moe.gate",
1579
+ "model.layers.48.share_expert.gate_up_proj",
1580
+ "model.layers.48.share_expert.gate_proj",
1581
+ "model.layers.48.share_expert.up_proj",
1582
+ "model.layers.48.share_expert.down_proj",
1583
+ "model.layers.49.self_attn.g_proj",
1584
+ "model.layers.49.self_attn.qkv_proj",
1585
+ "model.layers.49.self_attn.q_proj",
1586
+ "model.layers.49.self_attn.k_proj",
1587
+ "model.layers.49.self_attn.v_proj",
1588
+ "model.layers.49.self_attn.o_proj",
1589
+ "model.layers.49.moe.gate",
1590
+ "model.layers.49.share_expert.gate_up_proj",
1591
+ "model.layers.49.share_expert.gate_proj",
1592
+ "model.layers.49.share_expert.up_proj",
1593
+ "model.layers.49.share_expert.down_proj",
1594
+ "model.layers.50.self_attn.g_proj",
1595
+ "model.layers.50.self_attn.qkv_proj",
1596
+ "model.layers.50.self_attn.q_proj",
1597
+ "model.layers.50.self_attn.k_proj",
1598
+ "model.layers.50.self_attn.v_proj",
1599
+ "model.layers.50.self_attn.o_proj",
1600
+ "model.layers.50.moe.gate",
1601
+ "model.layers.50.share_expert.gate_up_proj",
1602
+ "model.layers.50.share_expert.gate_proj",
1603
+ "model.layers.50.share_expert.up_proj",
1604
+ "model.layers.50.share_expert.down_proj",
1605
+ "model.layers.51.self_attn.g_proj",
1606
+ "model.layers.51.self_attn.qkv_proj",
1607
+ "model.layers.51.self_attn.q_proj",
1608
+ "model.layers.51.self_attn.k_proj",
1609
+ "model.layers.51.self_attn.v_proj",
1610
+ "model.layers.51.self_attn.o_proj",
1611
+ "model.layers.51.self_attn.qkv_indexer_proj",
1612
+ "model.layers.51.moe.gate",
1613
+ "model.layers.51.share_expert.gate_up_proj",
1614
+ "model.layers.51.share_expert.gate_proj",
1615
+ "model.layers.51.share_expert.up_proj",
1616
+ "model.layers.51.share_expert.down_proj",
1617
+ "model.layers.52.self_attn.g_proj",
1618
+ "model.layers.52.self_attn.qkv_proj",
1619
+ "model.layers.52.self_attn.q_proj",
1620
+ "model.layers.52.self_attn.k_proj",
1621
+ "model.layers.52.self_attn.v_proj",
1622
+ "model.layers.52.self_attn.o_proj",
1623
+ "model.layers.52.moe.gate",
1624
+ "model.layers.52.share_expert.gate_up_proj",
1625
+ "model.layers.52.share_expert.gate_proj",
1626
+ "model.layers.52.share_expert.up_proj",
1627
+ "model.layers.52.share_expert.down_proj",
1628
+ "model.layers.53.self_attn.g_proj",
1629
+ "model.layers.53.self_attn.qkv_proj",
1630
+ "model.layers.53.self_attn.q_proj",
1631
+ "model.layers.53.self_attn.k_proj",
1632
+ "model.layers.53.self_attn.v_proj",
1633
+ "model.layers.53.self_attn.o_proj",
1634
+ "model.layers.53.moe.gate",
1635
+ "model.layers.53.share_expert.gate_up_proj",
1636
+ "model.layers.53.share_expert.gate_proj",
1637
+ "model.layers.53.share_expert.up_proj",
1638
+ "model.layers.53.share_expert.down_proj",
1639
+ "model.layers.54.self_attn.g_proj",
1640
+ "model.layers.54.self_attn.qkv_proj",
1641
+ "model.layers.54.self_attn.q_proj",
1642
+ "model.layers.54.self_attn.k_proj",
1643
+ "model.layers.54.self_attn.v_proj",
1644
+ "model.layers.54.self_attn.o_proj",
1645
+ "model.layers.54.moe.gate",
1646
+ "model.layers.54.share_expert.gate_up_proj",
1647
+ "model.layers.54.share_expert.gate_proj",
1648
+ "model.layers.54.share_expert.up_proj",
1649
+ "model.layers.54.share_expert.down_proj",
1650
+ "model.layers.55.self_attn.g_proj",
1651
+ "model.layers.55.self_attn.qkv_proj",
1652
+ "model.layers.55.self_attn.q_proj",
1653
+ "model.layers.55.self_attn.k_proj",
1654
+ "model.layers.55.self_attn.v_proj",
1655
+ "model.layers.55.self_attn.o_proj",
1656
+ "model.layers.55.self_attn.qkv_indexer_proj",
1657
+ "model.layers.55.moe.gate",
1658
+ "model.layers.55.share_expert.gate_up_proj",
1659
+ "model.layers.55.share_expert.gate_proj",
1660
+ "model.layers.55.share_expert.up_proj",
1661
+ "model.layers.55.share_expert.down_proj",
1662
+ "model.layers.56.self_attn.g_proj",
1663
+ "model.layers.56.self_attn.qkv_proj",
1664
+ "model.layers.56.self_attn.q_proj",
1665
+ "model.layers.56.self_attn.k_proj",
1666
+ "model.layers.56.self_attn.v_proj",
1667
+ "model.layers.56.self_attn.o_proj",
1668
+ "model.layers.56.moe.gate",
1669
+ "model.layers.56.share_expert.gate_up_proj",
1670
+ "model.layers.56.share_expert.gate_proj",
1671
+ "model.layers.56.share_expert.up_proj",
1672
+ "model.layers.56.share_expert.down_proj",
1673
+ "model.layers.57.self_attn.g_proj",
1674
+ "model.layers.57.self_attn.qkv_proj",
1675
+ "model.layers.57.self_attn.q_proj",
1676
+ "model.layers.57.self_attn.k_proj",
1677
+ "model.layers.57.self_attn.v_proj",
1678
+ "model.layers.57.self_attn.o_proj",
1679
+ "model.layers.57.moe.gate",
1680
+ "model.layers.57.share_expert.gate_up_proj",
1681
+ "model.layers.57.share_expert.gate_proj",
1682
+ "model.layers.57.share_expert.up_proj",
1683
+ "model.layers.57.share_expert.down_proj",
1684
+ "model.layers.58.self_attn.g_proj",
1685
+ "model.layers.58.self_attn.qkv_proj",
1686
+ "model.layers.58.self_attn.q_proj",
1687
+ "model.layers.58.self_attn.k_proj",
1688
+ "model.layers.58.self_attn.v_proj",
1689
+ "model.layers.58.self_attn.o_proj",
1690
+ "model.layers.58.moe.gate",
1691
+ "model.layers.58.share_expert.gate_up_proj",
1692
+ "model.layers.58.share_expert.gate_proj",
1693
+ "model.layers.58.share_expert.up_proj",
1694
+ "model.layers.58.share_expert.down_proj",
1695
+ "model.layers.59.self_attn.g_proj",
1696
+ "model.layers.59.self_attn.qkv_proj",
1697
+ "model.layers.59.self_attn.q_proj",
1698
+ "model.layers.59.self_attn.k_proj",
1699
+ "model.layers.59.self_attn.v_proj",
1700
+ "model.layers.59.self_attn.o_proj",
1701
+ "model.layers.59.self_attn.qkv_indexer_proj",
1702
+ "model.layers.59.moe.gate",
1703
+ "model.layers.59.share_expert.gate_up_proj",
1704
+ "model.layers.59.share_expert.gate_proj",
1705
+ "model.layers.59.share_expert.up_proj",
1706
+ "model.layers.59.share_expert.down_proj",
1707
+ "model.layers.60.self_attn.g_proj",
1708
+ "model.layers.60.self_attn.qkv_proj",
1709
+ "model.layers.60.self_attn.q_proj",
1710
+ "model.layers.60.self_attn.k_proj",
1711
+ "model.layers.60.self_attn.v_proj",
1712
+ "model.layers.60.self_attn.o_proj",
1713
+ "model.layers.60.moe.gate",
1714
+ "model.layers.60.share_expert.gate_up_proj",
1715
+ "model.layers.60.share_expert.gate_proj",
1716
+ "model.layers.60.share_expert.up_proj",
1717
+ "model.layers.60.share_expert.down_proj",
1718
+ "model.layers.61.self_attn.g_proj",
1719
+ "model.layers.61.self_attn.qkv_proj",
1720
+ "model.layers.61.self_attn.q_proj",
1721
+ "model.layers.61.self_attn.k_proj",
1722
+ "model.layers.61.self_attn.v_proj",
1723
+ "model.layers.61.self_attn.o_proj",
1724
+ "model.layers.61.moe.gate",
1725
+ "model.layers.61.share_expert.gate_up_proj",
1726
+ "model.layers.61.share_expert.gate_proj",
1727
+ "model.layers.61.share_expert.up_proj",
1728
+ "model.layers.61.share_expert.down_proj",
1729
+ "model.layers.62.self_attn.g_proj",
1730
+ "model.layers.62.self_attn.qkv_proj",
1731
+ "model.layers.62.self_attn.q_proj",
1732
+ "model.layers.62.self_attn.k_proj",
1733
+ "model.layers.62.self_attn.v_proj",
1734
+ "model.layers.62.self_attn.o_proj",
1735
+ "model.layers.62.moe.gate",
1736
+ "model.layers.62.share_expert.gate_up_proj",
1737
+ "model.layers.62.share_expert.gate_proj",
1738
+ "model.layers.62.share_expert.up_proj",
1739
+ "model.layers.62.share_expert.down_proj",
1740
+ "model.layers.63.self_attn.g_proj",
1741
+ "model.layers.63.self_attn.qkv_proj",
1742
+ "model.layers.63.self_attn.q_proj",
1743
+ "model.layers.63.self_attn.k_proj",
1744
+ "model.layers.63.self_attn.v_proj",
1745
+ "model.layers.63.self_attn.o_proj",
1746
+ "model.layers.63.self_attn.qkv_indexer_proj",
1747
+ "model.layers.63.moe.gate",
1748
+ "model.layers.63.share_expert.gate_up_proj",
1749
+ "model.layers.63.share_expert.gate_proj",
1750
+ "model.layers.63.share_expert.up_proj",
1751
+ "model.layers.63.share_expert.down_proj",
1752
+ "model.layers.64.self_attn.g_proj",
1753
+ "model.layers.64.self_attn.qkv_proj",
1754
+ "model.layers.64.self_attn.q_proj",
1755
+ "model.layers.64.self_attn.k_proj",
1756
+ "model.layers.64.self_attn.v_proj",
1757
+ "model.layers.64.self_attn.o_proj",
1758
+ "model.layers.64.moe.gate",
1759
+ "model.layers.64.share_expert.gate_up_proj",
1760
+ "model.layers.64.share_expert.gate_proj",
1761
+ "model.layers.64.share_expert.up_proj",
1762
+ "model.layers.64.share_expert.down_proj",
1763
+ "model.layers.65.self_attn.g_proj",
1764
+ "model.layers.65.self_attn.qkv_proj",
1765
+ "model.layers.65.self_attn.q_proj",
1766
+ "model.layers.65.self_attn.k_proj",
1767
+ "model.layers.65.self_attn.v_proj",
1768
+ "model.layers.65.self_attn.o_proj",
1769
+ "model.layers.65.moe.gate",
1770
+ "model.layers.65.share_expert.gate_up_proj",
1771
+ "model.layers.65.share_expert.gate_proj",
1772
+ "model.layers.65.share_expert.up_proj",
1773
+ "model.layers.65.share_expert.down_proj",
1774
+ "model.layers.66.self_attn.g_proj",
1775
+ "model.layers.66.self_attn.qkv_proj",
1776
+ "model.layers.66.self_attn.q_proj",
1777
+ "model.layers.66.self_attn.k_proj",
1778
+ "model.layers.66.self_attn.v_proj",
1779
+ "model.layers.66.self_attn.o_proj",
1780
+ "model.layers.66.moe.gate",
1781
+ "model.layers.66.share_expert.gate_up_proj",
1782
+ "model.layers.66.share_expert.gate_proj",
1783
+ "model.layers.66.share_expert.up_proj",
1784
+ "model.layers.66.share_expert.down_proj",
1785
+ "model.layers.67.self_attn.g_proj",
1786
+ "model.layers.67.self_attn.qkv_proj",
1787
+ "model.layers.67.self_attn.q_proj",
1788
+ "model.layers.67.self_attn.k_proj",
1789
+ "model.layers.67.self_attn.v_proj",
1790
+ "model.layers.67.self_attn.o_proj",
1791
+ "model.layers.67.self_attn.qkv_indexer_proj",
1792
+ "model.layers.67.moe.gate",
1793
+ "model.layers.67.share_expert.gate_up_proj",
1794
+ "model.layers.67.share_expert.gate_proj",
1795
+ "model.layers.67.share_expert.up_proj",
1796
+ "model.layers.67.share_expert.down_proj",
1797
+ "model.layers.68.self_attn.g_proj",
1798
+ "model.layers.68.self_attn.qkv_proj",
1799
+ "model.layers.68.self_attn.q_proj",
1800
+ "model.layers.68.self_attn.k_proj",
1801
+ "model.layers.68.self_attn.v_proj",
1802
+ "model.layers.68.self_attn.o_proj",
1803
+ "model.layers.68.moe.gate",
1804
+ "model.layers.68.share_expert.gate_up_proj",
1805
+ "model.layers.68.share_expert.gate_proj",
1806
+ "model.layers.68.share_expert.up_proj",
1807
+ "model.layers.68.share_expert.down_proj",
1808
+ "model.layers.69.self_attn.g_proj",
1809
+ "model.layers.69.self_attn.qkv_proj",
1810
+ "model.layers.69.self_attn.q_proj",
1811
+ "model.layers.69.self_attn.k_proj",
1812
+ "model.layers.69.self_attn.v_proj",
1813
+ "model.layers.69.self_attn.o_proj",
1814
+ "model.layers.69.moe.gate",
1815
+ "model.layers.69.share_expert.gate_up_proj",
1816
+ "model.layers.69.share_expert.gate_proj",
1817
+ "model.layers.69.share_expert.up_proj",
1818
+ "model.layers.69.share_expert.down_proj",
1819
+ "model.layers.70.self_attn.g_proj",
1820
+ "model.layers.70.self_attn.qkv_proj",
1821
+ "model.layers.70.self_attn.q_proj",
1822
+ "model.layers.70.self_attn.k_proj",
1823
+ "model.layers.70.self_attn.v_proj",
1824
+ "model.layers.70.self_attn.o_proj",
1825
+ "model.layers.70.moe.gate",
1826
+ "model.layers.70.share_expert.gate_up_proj",
1827
+ "model.layers.70.share_expert.gate_proj",
1828
+ "model.layers.70.share_expert.up_proj",
1829
+ "model.layers.70.share_expert.down_proj",
1830
+ "model.layers.71.self_attn.g_proj",
1831
+ "model.layers.71.self_attn.qkv_proj",
1832
+ "model.layers.71.self_attn.q_proj",
1833
+ "model.layers.71.self_attn.k_proj",
1834
+ "model.layers.71.self_attn.v_proj",
1835
+ "model.layers.71.self_attn.o_proj",
1836
+ "model.layers.71.self_attn.qkv_indexer_proj",
1837
+ "model.layers.71.moe.gate",
1838
+ "model.layers.71.share_expert.gate_up_proj",
1839
+ "model.layers.71.share_expert.gate_proj",
1840
+ "model.layers.71.share_expert.up_proj",
1841
+ "model.layers.71.share_expert.down_proj",
1842
+ "model.layers.72.self_attn.g_proj",
1843
+ "model.layers.72.self_attn.qkv_proj",
1844
+ "model.layers.72.self_attn.q_proj",
1845
+ "model.layers.72.self_attn.k_proj",
1846
+ "model.layers.72.self_attn.v_proj",
1847
+ "model.layers.72.self_attn.o_proj",
1848
+ "model.layers.72.moe.gate",
1849
+ "model.layers.72.share_expert.gate_up_proj",
1850
+ "model.layers.72.share_expert.gate_proj",
1851
+ "model.layers.72.share_expert.up_proj",
1852
+ "model.layers.72.share_expert.down_proj",
1853
+ "model.layers.73.self_attn.g_proj",
1854
+ "model.layers.73.self_attn.qkv_proj",
1855
+ "model.layers.73.self_attn.q_proj",
1856
+ "model.layers.73.self_attn.k_proj",
1857
+ "model.layers.73.self_attn.v_proj",
1858
+ "model.layers.73.self_attn.o_proj",
1859
+ "model.layers.73.moe.gate",
1860
+ "model.layers.73.share_expert.gate_up_proj",
1861
+ "model.layers.73.share_expert.gate_proj",
1862
+ "model.layers.73.share_expert.up_proj",
1863
+ "model.layers.73.share_expert.down_proj",
1864
+ "model.layers.74.self_attn.g_proj",
1865
+ "model.layers.74.self_attn.qkv_proj",
1866
+ "model.layers.74.self_attn.q_proj",
1867
+ "model.layers.74.self_attn.k_proj",
1868
+ "model.layers.74.self_attn.v_proj",
1869
+ "model.layers.74.self_attn.o_proj",
1870
+ "model.layers.74.moe.gate",
1871
+ "model.layers.74.share_expert.gate_up_proj",
1872
+ "model.layers.74.share_expert.gate_proj",
1873
+ "model.layers.74.share_expert.up_proj",
1874
+ "model.layers.74.share_expert.down_proj",
1875
+ "model.layers.75.self_attn.g_proj",
1876
+ "model.layers.75.self_attn.qkv_proj",
1877
+ "model.layers.75.self_attn.q_proj",
1878
+ "model.layers.75.self_attn.k_proj",
1879
+ "model.layers.75.self_attn.v_proj",
1880
+ "model.layers.75.self_attn.o_proj",
1881
+ "model.layers.75.self_attn.qkv_indexer_proj",
1882
+ "model.layers.75.moe.gate",
1883
+ "model.layers.75.share_expert.gate_up_proj",
1884
+ "model.layers.75.share_expert.gate_proj",
1885
+ "model.layers.75.share_expert.up_proj",
1886
+ "model.layers.75.share_expert.down_proj",
1887
+ "model.layers.76.self_attn.g_proj",
1888
+ "model.layers.76.self_attn.qkv_proj",
1889
+ "model.layers.76.self_attn.q_proj",
1890
+ "model.layers.76.self_attn.k_proj",
1891
+ "model.layers.76.self_attn.v_proj",
1892
+ "model.layers.76.self_attn.o_proj",
1893
+ "model.layers.76.moe.gate",
1894
+ "model.layers.76.share_expert.gate_up_proj",
1895
+ "model.layers.76.share_expert.gate_proj",
1896
+ "model.layers.76.share_expert.up_proj",
1897
+ "model.layers.76.share_expert.down_proj",
1898
+ "model.layers.77.self_attn.g_proj",
1899
+ "model.layers.77.self_attn.qkv_proj",
1900
+ "model.layers.77.self_attn.q_proj",
1901
+ "model.layers.77.self_attn.k_proj",
1902
+ "model.layers.77.self_attn.v_proj",
1903
+ "model.layers.77.self_attn.o_proj",
1904
+ "model.layers.77.moe.gate",
1905
+ "model.layers.77.share_expert.gate_up_proj",
1906
+ "model.layers.77.share_expert.gate_proj",
1907
+ "model.layers.77.share_expert.up_proj",
1908
+ "model.layers.77.share_expert.down_proj",
1909
+ "model.layers.78.self_attn.g_proj",
1910
+ "model.layers.78.self_attn.qkv_proj",
1911
+ "model.layers.78.self_attn.q_proj",
1912
+ "model.layers.78.self_attn.k_proj",
1913
+ "model.layers.78.self_attn.v_proj",
1914
+ "model.layers.78.self_attn.o_proj",
1915
+ "model.layers.78.moe.gate",
1916
+ "model.layers.78.share_expert.gate_up_proj",
1917
+ "model.layers.78.share_expert.gate_proj",
1918
+ "model.layers.78.share_expert.up_proj",
1919
+ "model.layers.78.share_expert.down_proj",
1920
+ "model.layers.79.self_attn.g_proj",
1921
+ "model.layers.79.self_attn.qkv_proj",
1922
+ "model.layers.79.self_attn.q_proj",
1923
+ "model.layers.79.self_attn.k_proj",
1924
+ "model.layers.79.self_attn.v_proj",
1925
+ "model.layers.79.self_attn.o_proj",
1926
+ "model.layers.79.self_attn.qkv_indexer_proj",
1927
+ "model.layers.79.moe.gate",
1928
+ "model.layers.79.share_expert.gate_up_proj",
1929
+ "model.layers.79.share_expert.gate_proj",
1930
+ "model.layers.79.share_expert.up_proj",
1931
+ "model.layers.79.share_expert.down_proj",
1932
+ "model.layers.80.self_attn.g_proj",
1933
+ "model.layers.80.self_attn.qkv_proj",
1934
+ "model.layers.80.self_attn.q_proj",
1935
+ "model.layers.80.self_attn.k_proj",
1936
+ "model.layers.80.self_attn.v_proj",
1937
+ "model.layers.80.self_attn.o_proj",
1938
+ "model.layers.80.moe.gate",
1939
+ "model.layers.80.share_expert.gate_up_proj",
1940
+ "model.layers.80.share_expert.gate_proj",
1941
+ "model.layers.80.share_expert.up_proj",
1942
+ "model.layers.80.share_expert.down_proj",
1943
+ "model.layers.81.self_attn.g_proj",
1944
+ "model.layers.81.self_attn.qkv_proj",
1945
+ "model.layers.81.self_attn.q_proj",
1946
+ "model.layers.81.self_attn.k_proj",
1947
+ "model.layers.81.self_attn.v_proj",
1948
+ "model.layers.81.self_attn.o_proj",
1949
+ "model.layers.81.moe.gate",
1950
+ "model.layers.81.share_expert.gate_up_proj",
1951
+ "model.layers.81.share_expert.gate_proj",
1952
+ "model.layers.81.share_expert.up_proj",
1953
+ "model.layers.81.share_expert.down_proj",
1954
+ "model.layers.82.self_attn.g_proj",
1955
+ "model.layers.82.self_attn.qkv_proj",
1956
+ "model.layers.82.self_attn.q_proj",
1957
+ "model.layers.82.self_attn.k_proj",
1958
+ "model.layers.82.self_attn.v_proj",
1959
+ "model.layers.82.self_attn.o_proj",
1960
+ "model.layers.82.moe.gate",
1961
+ "model.layers.82.share_expert.gate_up_proj",
1962
+ "model.layers.82.share_expert.gate_proj",
1963
+ "model.layers.82.share_expert.up_proj",
1964
+ "model.layers.82.share_expert.down_proj",
1965
+ "model.layers.83.self_attn.g_proj",
1966
+ "model.layers.83.self_attn.qkv_proj",
1967
+ "model.layers.83.self_attn.q_proj",
1968
+ "model.layers.83.self_attn.k_proj",
1969
+ "model.layers.83.self_attn.v_proj",
1970
+ "model.layers.83.self_attn.o_proj",
1971
+ "model.layers.83.self_attn.qkv_indexer_proj",
1972
+ "model.layers.83.moe.gate",
1973
+ "model.layers.83.share_expert.gate_up_proj",
1974
+ "model.layers.83.share_expert.gate_proj",
1975
+ "model.layers.83.share_expert.up_proj",
1976
+ "model.layers.83.share_expert.down_proj",
1977
+ "model.layers.84.self_attn.g_proj",
1978
+ "model.layers.84.self_attn.qkv_proj",
1979
+ "model.layers.84.self_attn.q_proj",
1980
+ "model.layers.84.self_attn.k_proj",
1981
+ "model.layers.84.self_attn.v_proj",
1982
+ "model.layers.84.self_attn.o_proj",
1983
+ "model.layers.84.moe.gate",
1984
+ "model.layers.84.share_expert.gate_up_proj",
1985
+ "model.layers.84.share_expert.gate_proj",
1986
+ "model.layers.84.share_expert.up_proj",
1987
+ "model.layers.84.share_expert.down_proj",
1988
+ "model.layers.85.self_attn.g_proj",
1989
+ "model.layers.85.self_attn.qkv_proj",
1990
+ "model.layers.85.self_attn.q_proj",
1991
+ "model.layers.85.self_attn.k_proj",
1992
+ "model.layers.85.self_attn.v_proj",
1993
+ "model.layers.85.self_attn.o_proj",
1994
+ "model.layers.85.moe.gate",
1995
+ "model.layers.85.share_expert.gate_up_proj",
1996
+ "model.layers.85.share_expert.gate_proj",
1997
+ "model.layers.85.share_expert.up_proj",
1998
+ "model.layers.85.share_expert.down_proj",
1999
+ "model.layers.86.self_attn.g_proj",
2000
+ "model.layers.86.self_attn.qkv_proj",
2001
+ "model.layers.86.self_attn.q_proj",
2002
+ "model.layers.86.self_attn.k_proj",
2003
+ "model.layers.86.self_attn.v_proj",
2004
+ "model.layers.86.self_attn.o_proj",
2005
+ "model.layers.86.moe.gate",
2006
+ "model.layers.86.share_expert.gate_up_proj",
2007
+ "model.layers.86.share_expert.gate_proj",
2008
+ "model.layers.86.share_expert.up_proj",
2009
+ "model.layers.86.share_expert.down_proj",
2010
+ "model.layers.87.self_attn.g_proj",
2011
+ "model.layers.87.self_attn.qkv_proj",
2012
+ "model.layers.87.self_attn.q_proj",
2013
+ "model.layers.87.self_attn.k_proj",
2014
+ "model.layers.87.self_attn.v_proj",
2015
+ "model.layers.87.self_attn.o_proj",
2016
+ "model.layers.87.self_attn.qkv_indexer_proj",
2017
+ "model.layers.87.moe.gate",
2018
+ "model.layers.87.share_expert.gate_up_proj",
2019
+ "model.layers.87.share_expert.gate_proj",
2020
+ "model.layers.87.share_expert.up_proj",
2021
+ "model.layers.87.share_expert.down_proj",
2022
+ "model.layers.88.self_attn.g_proj",
2023
+ "model.layers.88.self_attn.qkv_proj",
2024
+ "model.layers.88.self_attn.q_proj",
2025
+ "model.layers.88.self_attn.k_proj",
2026
+ "model.layers.88.self_attn.v_proj",
2027
+ "model.layers.88.self_attn.o_proj",
2028
+ "model.layers.88.moe.gate",
2029
+ "model.layers.88.share_expert.gate_up_proj",
2030
+ "model.layers.88.share_expert.gate_proj",
2031
+ "model.layers.88.share_expert.up_proj",
2032
+ "model.layers.88.share_expert.down_proj",
2033
+ "model.layers.89.self_attn.g_proj",
2034
+ "model.layers.89.self_attn.qkv_proj",
2035
+ "model.layers.89.self_attn.q_proj",
2036
+ "model.layers.89.self_attn.k_proj",
2037
+ "model.layers.89.self_attn.v_proj",
2038
+ "model.layers.89.self_attn.o_proj",
2039
+ "model.layers.89.moe.gate",
2040
+ "model.layers.89.share_expert.gate_up_proj",
2041
+ "model.layers.89.share_expert.gate_proj",
2042
+ "model.layers.89.share_expert.up_proj",
2043
+ "model.layers.89.share_expert.down_proj",
2044
+ "model.layers.90.self_attn.g_proj",
2045
+ "model.layers.90.self_attn.qkv_proj",
2046
+ "model.layers.90.self_attn.q_proj",
2047
+ "model.layers.90.self_attn.k_proj",
2048
+ "model.layers.90.self_attn.v_proj",
2049
+ "model.layers.90.self_attn.o_proj",
2050
+ "model.layers.90.moe.gate",
2051
+ "model.layers.90.share_expert.gate_up_proj",
2052
+ "model.layers.90.share_expert.gate_proj",
2053
+ "model.layers.90.share_expert.up_proj",
2054
+ "model.layers.90.share_expert.down_proj",
2055
+ "model.layers.91.self_attn.g_proj",
2056
+ "model.layers.91.self_attn.qkv_proj",
2057
+ "model.layers.91.self_attn.q_proj",
2058
+ "model.layers.91.self_attn.k_proj",
2059
+ "model.layers.91.self_attn.v_proj",
2060
+ "model.layers.91.self_attn.o_proj",
2061
+ "model.layers.91.self_attn.qkv_indexer_proj",
2062
+ "model.layers.91.mlp.gate_up_proj",
2063
+ "model.layers.91.mlp.gate_proj",
2064
+ "model.layers.91.mlp.up_proj",
2065
+ "model.layers.91.mlp.down_proj",
2066
+ "model.layers.92.mtp_block.self_attn.g_proj",
2067
+ "model.layers.92.mtp_block.self_attn.qkv_proj",
2068
+ "model.layers.92.mtp_block.self_attn.q_proj",
2069
+ "model.layers.92.mtp_block.self_attn.k_proj",
2070
+ "model.layers.92.mtp_block.self_attn.v_proj",
2071
+ "model.layers.92.mtp_block.self_attn.o_proj",
2072
+ "model.layers.92.mtp_block.mlp.gate_up_proj",
2073
+ "model.layers.92.mtp_block.mlp.gate_proj",
2074
+ "model.layers.92.mtp_block.mlp.up_proj",
2075
+ "model.layers.92.mtp_block.mlp.down_proj",
2076
+ "model.layers.92.shared_head",
2077
+ "model.layers.92.shared_head.head",
2078
+ "model.layers.93.mtp_block.self_attn.g_proj",
2079
+ "model.layers.93.mtp_block.self_attn.qkv_proj",
2080
+ "model.layers.93.mtp_block.self_attn.q_proj",
2081
+ "model.layers.93.mtp_block.self_attn.k_proj",
2082
+ "model.layers.93.mtp_block.self_attn.v_proj",
2083
+ "model.layers.93.mtp_block.self_attn.o_proj",
2084
+ "model.layers.93.mtp_block.mlp.gate_up_proj",
2085
+ "model.layers.93.mtp_block.mlp.gate_proj",
2086
+ "model.layers.93.mtp_block.mlp.up_proj",
2087
+ "model.layers.93.mtp_block.mlp.down_proj",
2088
+ "model.layers.93.shared_head",
2089
+ "model.layers.93.shared_head.head",
2090
+ "model.layers.94.mtp_block.self_attn.g_proj",
2091
+ "model.layers.94.mtp_block.self_attn.qkv_proj",
2092
+ "model.layers.94.mtp_block.self_attn.q_proj",
2093
+ "model.layers.94.mtp_block.self_attn.k_proj",
2094
+ "model.layers.94.mtp_block.self_attn.v_proj",
2095
+ "model.layers.94.mtp_block.self_attn.o_proj",
2096
+ "model.layers.94.mtp_block.mlp.gate_up_proj",
2097
+ "model.layers.94.mtp_block.mlp.gate_proj",
2098
+ "model.layers.94.mtp_block.mlp.up_proj",
2099
+ "model.layers.94.mtp_block.mlp.down_proj",
2100
+ "model.layers.94.shared_head",
2101
+ "model.layers.94.shared_head.head"
2102
+ ]
2103
  }
2104
  }