inference-optimization/DSV4-tiny-empty

1{2  "model.embed_tokens.weight": "embed.weight",3  "model.hc_head.hc_base": "hc_head_base",4  "model.hc_head.hc_fn": "hc_head_fn",5  "model.hc_head.hc_scale": "hc_head_scale",6  "model.layers.0.attn_hc.base": "layers.0.hc_attn_base",7  "model.layers.0.attn_hc.fn": "layers.0.hc_attn_fn",8  "model.layers.0.attn_hc.scale": "layers.0.hc_attn_scale",9  "model.layers.0.ffn_hc.base": "layers.0.hc_ffn_base",10  "model.layers.0.ffn_hc.fn": "layers.0.hc_ffn_fn",11  "model.layers.0.ffn_hc.scale": "layers.0.hc_ffn_scale",12  "model.layers.0.input_layernorm.weight": "layers.0.attn_norm.weight",13  "model.layers.0.mlp.experts.0.w1.weight": "layers.0.ffn.experts.0.w1.weight",14  "model.layers.0.mlp.experts.0.w2.weight": "layers.0.ffn.experts.0.w2.weight",15  "model.layers.0.mlp.experts.0.w3.weight": "layers.0.ffn.experts.0.w3.weight",16  "model.layers.0.mlp.experts.1.w1.weight": "layers.0.ffn.experts.1.w1.weight",17  "model.layers.0.mlp.experts.1.w2.weight": "layers.0.ffn.experts.1.w2.weight",18  "model.layers.0.mlp.experts.1.w3.weight": "layers.0.ffn.experts.1.w3.weight",19  "model.layers.0.mlp.experts.10.w1.weight": "layers.0.ffn.experts.10.w1.weight",20  "model.layers.0.mlp.experts.10.w2.weight": "layers.0.ffn.experts.10.w2.weight",21  "model.layers.0.mlp.experts.10.w3.weight": "layers.0.ffn.experts.10.w3.weight",22  "model.layers.0.mlp.experts.11.w1.weight": "layers.0.ffn.experts.11.w1.weight",23  "model.layers.0.mlp.experts.11.w2.weight": "layers.0.ffn.experts.11.w2.weight",24  "model.layers.0.mlp.experts.11.w3.weight": "layers.0.ffn.experts.11.w3.weight",25  "model.layers.0.mlp.experts.12.w1.weight": "layers.0.ffn.experts.12.w1.weight",26  "model.layers.0.mlp.experts.12.w2.weight": "layers.0.ffn.experts.12.w2.weight",27  "model.layers.0.mlp.experts.12.w3.weight": "layers.0.ffn.experts.12.w3.weight",28  "model.layers.0.mlp.experts.13.w1.weight": "layers.0.ffn.experts.13.w1.weight",29  "model.layers.0.mlp.experts.13.w2.weight": "layers.0.ffn.experts.13.w2.weight",30  "model.layers.0.mlp.experts.13.w3.weight": "layers.0.ffn.experts.13.w3.weight",31  "model.layers.0.mlp.experts.14.w1.weight": "layers.0.ffn.experts.14.w1.weight",32  "model.layers.0.mlp.experts.14.w2.weight": "layers.0.ffn.experts.14.w2.weight",33  "model.layers.0.mlp.experts.14.w3.weight": "layers.0.ffn.experts.14.w3.weight",34  "model.layers.0.mlp.experts.15.w1.weight": "layers.0.ffn.experts.15.w1.weight",35  "model.layers.0.mlp.experts.15.w2.weight": "layers.0.ffn.experts.15.w2.weight",36  "model.layers.0.mlp.experts.15.w3.weight": "layers.0.ffn.experts.15.w3.weight",37  "model.layers.0.mlp.experts.2.w1.weight": "layers.0.ffn.experts.2.w1.weight",38  "model.layers.0.mlp.experts.2.w2.weight": "layers.0.ffn.experts.2.w2.weight",39  "model.layers.0.mlp.experts.2.w3.weight": "layers.0.ffn.experts.2.w3.weight",40  "model.layers.0.mlp.experts.3.w1.weight": "layers.0.ffn.experts.3.w1.weight",41  "model.layers.0.mlp.experts.3.w2.weight": "layers.0.ffn.experts.3.w2.weight",42  "model.layers.0.mlp.experts.3.w3.weight": "layers.0.ffn.experts.3.w3.weight",43  "model.layers.0.mlp.experts.4.w1.weight": "layers.0.ffn.experts.4.w1.weight",44  "model.layers.0.mlp.experts.4.w2.weight": "layers.0.ffn.experts.4.w2.weight",45  "model.layers.0.mlp.experts.4.w3.weight": "layers.0.ffn.experts.4.w3.weight",46  "model.layers.0.mlp.experts.5.w1.weight": "layers.0.ffn.experts.5.w1.weight",47  "model.layers.0.mlp.experts.5.w2.weight": "layers.0.ffn.experts.5.w2.weight",48  "model.layers.0.mlp.experts.5.w3.weight": "layers.0.ffn.experts.5.w3.weight",49  "model.layers.0.mlp.experts.6.w1.weight": "layers.0.ffn.experts.6.w1.weight",50  "model.layers.0.mlp.experts.6.w2.weight": "layers.0.ffn.experts.6.w2.weight",51  "model.layers.0.mlp.experts.6.w3.weight": "layers.0.ffn.experts.6.w3.weight",52  "model.layers.0.mlp.experts.7.w1.weight": "layers.0.ffn.experts.7.w1.weight",53  "model.layers.0.mlp.experts.7.w2.weight": "layers.0.ffn.experts.7.w2.weight",54  "model.layers.0.mlp.experts.7.w3.weight": "layers.0.ffn.experts.7.w3.weight",55  "model.layers.0.mlp.experts.8.w1.weight": "layers.0.ffn.experts.8.w1.weight",56  "model.layers.0.mlp.experts.8.w2.weight": "layers.0.ffn.experts.8.w2.weight",57  "model.layers.0.mlp.experts.8.w3.weight": "layers.0.ffn.experts.8.w3.weight",58  "model.layers.0.mlp.experts.9.w1.weight": "layers.0.ffn.experts.9.w1.weight",59  "model.layers.0.mlp.experts.9.w2.weight": "layers.0.ffn.experts.9.w2.weight",60  "model.layers.0.mlp.experts.9.w3.weight": "layers.0.ffn.experts.9.w3.weight",61  "model.layers.0.mlp.gate.tid2eid": "layers.0.ffn.gate.tid2eid",62  "model.layers.0.mlp.gate.weight": "layers.0.ffn.gate.weight",63  "model.layers.0.mlp.shared_experts.down_proj.weight": "layers.0.ffn.shared_experts.w2.weight",64  "model.layers.0.mlp.shared_experts.gate_proj.weight": "layers.0.ffn.shared_experts.w1.weight",65  "model.layers.0.mlp.shared_experts.up_proj.weight": "layers.0.ffn.shared_experts.w3.weight",66  "model.layers.0.post_attention_layernorm.weight": "layers.0.ffn_norm.weight",67  "model.layers.0.self_attn.compressor.gate_proj.weight": "layers.0.attn.compressor.gate_proj.weight",68  "model.layers.0.self_attn.compressor.kv_norm.weight": "layers.0.attn.compressor.kv_norm.weight",69  "model.layers.0.self_attn.compressor.kv_proj.weight": "layers.0.attn.compressor.kv_proj.weight",70  "model.layers.0.self_attn.compressor.position_bias": "layers.0.attn.compressor.position_bias",71  "model.layers.0.self_attn.kv_norm.weight": "layers.0.attn.kv_norm.weight",72  "model.layers.0.self_attn.kv_proj.weight": "layers.0.attn.wkv.weight",73  "model.layers.0.self_attn.o_a_proj.weight": "layers.0.attn.wo_a.weight",74  "model.layers.0.self_attn.o_b_proj.weight": "layers.0.attn.wo_b.weight",75  "model.layers.0.self_attn.q_a_norm.weight": "layers.0.attn.q_norm.weight",76  "model.layers.0.self_attn.q_a_proj.weight": "layers.0.attn.wq_a.weight",77  "model.layers.0.self_attn.q_b_proj.weight": "layers.0.attn.wq_b.weight",78  "model.layers.0.self_attn.sinks": "layers.0.attn.attn_sink",79  "model.layers.1.attn_hc.base": "layers.1.hc_attn_base",80  "model.layers.1.attn_hc.fn": "layers.1.hc_attn_fn",81  "model.layers.1.attn_hc.scale": "layers.1.hc_attn_scale",82  "model.layers.1.ffn_hc.base": "layers.1.hc_ffn_base",83  "model.layers.1.ffn_hc.fn": "layers.1.hc_ffn_fn",84  "model.layers.1.ffn_hc.scale": "layers.1.hc_ffn_scale",85  "model.layers.1.input_layernorm.weight": "layers.1.attn_norm.weight",86  "model.layers.1.mlp.experts.0.w1.weight": "layers.1.ffn.experts.0.w1.weight",87  "model.layers.1.mlp.experts.0.w2.weight": "layers.1.ffn.experts.0.w2.weight",88  "model.layers.1.mlp.experts.0.w3.weight": "layers.1.ffn.experts.0.w3.weight",89  "model.layers.1.mlp.experts.1.w1.weight": "layers.1.ffn.experts.1.w1.weight",90  "model.layers.1.mlp.experts.1.w2.weight": "layers.1.ffn.experts.1.w2.weight",91  "model.layers.1.mlp.experts.1.w3.weight": "layers.1.ffn.experts.1.w3.weight",92  "model.layers.1.mlp.experts.10.w1.weight": "layers.1.ffn.experts.10.w1.weight",93  "model.layers.1.mlp.experts.10.w2.weight": "layers.1.ffn.experts.10.w2.weight",94  "model.layers.1.mlp.experts.10.w3.weight": "layers.1.ffn.experts.10.w3.weight",95  "model.layers.1.mlp.experts.11.w1.weight": "layers.1.ffn.experts.11.w1.weight",96  "model.layers.1.mlp.experts.11.w2.weight": "layers.1.ffn.experts.11.w2.weight",97  "model.layers.1.mlp.experts.11.w3.weight": "layers.1.ffn.experts.11.w3.weight",98  "model.layers.1.mlp.experts.12.w1.weight": "layers.1.ffn.experts.12.w1.weight",99  "model.layers.1.mlp.experts.12.w2.weight": "layers.1.ffn.experts.12.w2.weight",100  "model.layers.1.mlp.experts.12.w3.weight": "layers.1.ffn.experts.12.w3.weight",101  "model.layers.1.mlp.experts.13.w1.weight": "layers.1.ffn.experts.13.w1.weight",102  "model.layers.1.mlp.experts.13.w2.weight": "layers.1.ffn.experts.13.w2.weight",103  "model.layers.1.mlp.experts.13.w3.weight": "layers.1.ffn.experts.13.w3.weight",104  "model.layers.1.mlp.experts.14.w1.weight": "layers.1.ffn.experts.14.w1.weight",105  "model.layers.1.mlp.experts.14.w2.weight": "layers.1.ffn.experts.14.w2.weight",106  "model.layers.1.mlp.experts.14.w3.weight": "layers.1.ffn.experts.14.w3.weight",107  "model.layers.1.mlp.experts.15.w1.weight": "layers.1.ffn.experts.15.w1.weight",108  "model.layers.1.mlp.experts.15.w2.weight": "layers.1.ffn.experts.15.w2.weight",109  "model.layers.1.mlp.experts.15.w3.weight": "layers.1.ffn.experts.15.w3.weight",110  "model.layers.1.mlp.experts.2.w1.weight": "layers.1.ffn.experts.2.w1.weight",111  "model.layers.1.mlp.experts.2.w2.weight": "layers.1.ffn.experts.2.w2.weight",112  "model.layers.1.mlp.experts.2.w3.weight": "layers.1.ffn.experts.2.w3.weight",113  "model.layers.1.mlp.experts.3.w1.weight": "layers.1.ffn.experts.3.w1.weight",114  "model.layers.1.mlp.experts.3.w2.weight": "layers.1.ffn.experts.3.w2.weight",115  "model.layers.1.mlp.experts.3.w3.weight": "layers.1.ffn.experts.3.w3.weight",116  "model.layers.1.mlp.experts.4.w1.weight": "layers.1.ffn.experts.4.w1.weight",117  "model.layers.1.mlp.experts.4.w2.weight": "layers.1.ffn.experts.4.w2.weight",118  "model.layers.1.mlp.experts.4.w3.weight": "layers.1.ffn.experts.4.w3.weight",119  "model.layers.1.mlp.experts.5.w1.weight": "layers.1.ffn.experts.5.w1.weight",120  "model.layers.1.mlp.experts.5.w2.weight": "layers.1.ffn.experts.5.w2.weight",121  "model.layers.1.mlp.experts.5.w3.weight": "layers.1.ffn.experts.5.w3.weight",122  "model.layers.1.mlp.experts.6.w1.weight": "layers.1.ffn.experts.6.w1.weight",123  "model.layers.1.mlp.experts.6.w2.weight": "layers.1.ffn.experts.6.w2.weight",124  "model.layers.1.mlp.experts.6.w3.weight": "layers.1.ffn.experts.6.w3.weight",125  "model.layers.1.mlp.experts.7.w1.weight": "layers.1.ffn.experts.7.w1.weight",126  "model.layers.1.mlp.experts.7.w2.weight": "layers.1.ffn.experts.7.w2.weight",127  "model.layers.1.mlp.experts.7.w3.weight": "layers.1.ffn.experts.7.w3.weight",128  "model.layers.1.mlp.experts.8.w1.weight": "layers.1.ffn.experts.8.w1.weight",129  "model.layers.1.mlp.experts.8.w2.weight": "layers.1.ffn.experts.8.w2.weight",130  "model.layers.1.mlp.experts.8.w3.weight": "layers.1.ffn.experts.8.w3.weight",131  "model.layers.1.mlp.experts.9.w1.weight": "layers.1.ffn.experts.9.w1.weight",132  "model.layers.1.mlp.experts.9.w2.weight": "layers.1.ffn.experts.9.w2.weight",133  "model.layers.1.mlp.experts.9.w3.weight": "layers.1.ffn.experts.9.w3.weight",134  "model.layers.1.mlp.gate.e_score_correction_bias": "layers.1.ffn.gate.e_score_correction_bias",135  "model.layers.1.mlp.gate.weight": "layers.1.ffn.gate.weight",136  "model.layers.1.mlp.shared_experts.down_proj.weight": "layers.1.ffn.shared_experts.w2.weight",137  "model.layers.1.mlp.shared_experts.gate_proj.weight": "layers.1.ffn.shared_experts.w1.weight",138  "model.layers.1.mlp.shared_experts.up_proj.weight": "layers.1.ffn.shared_experts.w3.weight",139  "model.layers.1.post_attention_layernorm.weight": "layers.1.ffn_norm.weight",140  "model.layers.1.self_attn.compressor.gate_proj.weight": "layers.1.attn.compressor.gate_proj.weight",141  "model.layers.1.self_attn.compressor.indexer.gate_proj.weight": "layers.1.attn.compressor.indexer.gate_proj.weight",142  "model.layers.1.self_attn.compressor.indexer.kv_norm.weight": "layers.1.attn.compressor.indexer.kv_norm.weight",143  "model.layers.1.self_attn.compressor.indexer.kv_proj.weight": "layers.1.attn.compressor.indexer.kv_proj.weight",144  "model.layers.1.self_attn.compressor.indexer.position_bias": "layers.1.attn.compressor.indexer.position_bias",145  "model.layers.1.self_attn.compressor.indexer.q_b_proj.weight": "layers.1.attn.compressor.indexer.q_b_proj.weight",146  "model.layers.1.self_attn.compressor.indexer.weights_proj.weight": "layers.1.attn.compressor.indexer.weights_proj.weight",147  "model.layers.1.self_attn.compressor.kv_norm.weight": "layers.1.attn.compressor.kv_norm.weight",148  "model.layers.1.self_attn.compressor.kv_proj.weight": "layers.1.attn.compressor.kv_proj.weight",149  "model.layers.1.self_attn.compressor.position_bias": "layers.1.attn.compressor.position_bias",150  "model.layers.1.self_attn.kv_norm.weight": "layers.1.attn.kv_norm.weight",151  "model.layers.1.self_attn.kv_proj.weight": "layers.1.attn.wkv.weight",152  "model.layers.1.self_attn.o_a_proj.weight": "layers.1.attn.wo_a.weight",153  "model.layers.1.self_attn.o_b_proj.weight": "layers.1.attn.wo_b.weight",154  "model.layers.1.self_attn.q_a_norm.weight": "layers.1.attn.q_norm.weight",155  "model.layers.1.self_attn.q_a_proj.weight": "layers.1.attn.wq_a.weight",156  "model.layers.1.self_attn.q_b_proj.weight": "layers.1.attn.wq_b.weight",157  "model.layers.1.self_attn.sinks": "layers.1.attn.attn_sink",158  "model.layers.2.attn_hc.base": "layers.2.hc_attn_base",159  "model.layers.2.attn_hc.fn": "layers.2.hc_attn_fn",160  "model.layers.2.attn_hc.scale": "layers.2.hc_attn_scale",161  "model.layers.2.ffn_hc.base": "layers.2.hc_ffn_base",162  "model.layers.2.ffn_hc.fn": "layers.2.hc_ffn_fn",163  "model.layers.2.ffn_hc.scale": "layers.2.hc_ffn_scale",164  "model.layers.2.input_layernorm.weight": "layers.2.attn_norm.weight",165  "model.layers.2.mlp.experts.0.w1.weight": "layers.2.ffn.experts.0.w1.weight",166  "model.layers.2.mlp.experts.0.w2.weight": "layers.2.ffn.experts.0.w2.weight",167  "model.layers.2.mlp.experts.0.w3.weight": "layers.2.ffn.experts.0.w3.weight",168  "model.layers.2.mlp.experts.1.w1.weight": "layers.2.ffn.experts.1.w1.weight",169  "model.layers.2.mlp.experts.1.w2.weight": "layers.2.ffn.experts.1.w2.weight",170  "model.layers.2.mlp.experts.1.w3.weight": "layers.2.ffn.experts.1.w3.weight",171  "model.layers.2.mlp.experts.10.w1.weight": "layers.2.ffn.experts.10.w1.weight",172  "model.layers.2.mlp.experts.10.w2.weight": "layers.2.ffn.experts.10.w2.weight",173  "model.layers.2.mlp.experts.10.w3.weight": "layers.2.ffn.experts.10.w3.weight",174  "model.layers.2.mlp.experts.11.w1.weight": "layers.2.ffn.experts.11.w1.weight",175  "model.layers.2.mlp.experts.11.w2.weight": "layers.2.ffn.experts.11.w2.weight",176  "model.layers.2.mlp.experts.11.w3.weight": "layers.2.ffn.experts.11.w3.weight",177  "model.layers.2.mlp.experts.12.w1.weight": "layers.2.ffn.experts.12.w1.weight",178  "model.layers.2.mlp.experts.12.w2.weight": "layers.2.ffn.experts.12.w2.weight",179  "model.layers.2.mlp.experts.12.w3.weight": "layers.2.ffn.experts.12.w3.weight",180  "model.layers.2.mlp.experts.13.w1.weight": "layers.2.ffn.experts.13.w1.weight",181  "model.layers.2.mlp.experts.13.w2.weight": "layers.2.ffn.experts.13.w2.weight",182  "model.layers.2.mlp.experts.13.w3.weight": "layers.2.ffn.experts.13.w3.weight",183  "model.layers.2.mlp.experts.14.w1.weight": "layers.2.ffn.experts.14.w1.weight",184  "model.layers.2.mlp.experts.14.w2.weight": "layers.2.ffn.experts.14.w2.weight",185  "model.layers.2.mlp.experts.14.w3.weight": "layers.2.ffn.experts.14.w3.weight",186  "model.layers.2.mlp.experts.15.w1.weight": "layers.2.ffn.experts.15.w1.weight",187  "model.layers.2.mlp.experts.15.w2.weight": "layers.2.ffn.experts.15.w2.weight",188  "model.layers.2.mlp.experts.15.w3.weight": "layers.2.ffn.experts.15.w3.weight",189  "model.layers.2.mlp.experts.2.w1.weight": "layers.2.ffn.experts.2.w1.weight",190  "model.layers.2.mlp.experts.2.w2.weight": "layers.2.ffn.experts.2.w2.weight",191  "model.layers.2.mlp.experts.2.w3.weight": "layers.2.ffn.experts.2.w3.weight",192  "model.layers.2.mlp.experts.3.w1.weight": "layers.2.ffn.experts.3.w1.weight",193  "model.layers.2.mlp.experts.3.w2.weight": "layers.2.ffn.experts.3.w2.weight",194  "model.layers.2.mlp.experts.3.w3.weight": "layers.2.ffn.experts.3.w3.weight",195  "model.layers.2.mlp.experts.4.w1.weight": "layers.2.ffn.experts.4.w1.weight",196  "model.layers.2.mlp.experts.4.w2.weight": "layers.2.ffn.experts.4.w2.weight",197  "model.layers.2.mlp.experts.4.w3.weight": "layers.2.ffn.experts.4.w3.weight",198  "model.layers.2.mlp.experts.5.w1.weight": "layers.2.ffn.experts.5.w1.weight",199  "model.layers.2.mlp.experts.5.w2.weight": "layers.2.ffn.experts.5.w2.weight",200  "model.layers.2.mlp.experts.5.w3.weight": "layers.2.ffn.experts.5.w3.weight",201  "model.layers.2.mlp.experts.6.w1.weight": "layers.2.ffn.experts.6.w1.weight",202  "model.layers.2.mlp.experts.6.w2.weight": "layers.2.ffn.experts.6.w2.weight",203  "model.layers.2.mlp.experts.6.w3.weight": "layers.2.ffn.experts.6.w3.weight",204  "model.layers.2.mlp.experts.7.w1.weight": "layers.2.ffn.experts.7.w1.weight",205  "model.layers.2.mlp.experts.7.w2.weight": "layers.2.ffn.experts.7.w2.weight",206  "model.layers.2.mlp.experts.7.w3.weight": "layers.2.ffn.experts.7.w3.weight",207  "model.layers.2.mlp.experts.8.w1.weight": "layers.2.ffn.experts.8.w1.weight",208  "model.layers.2.mlp.experts.8.w2.weight": "layers.2.ffn.experts.8.w2.weight",209  "model.layers.2.mlp.experts.8.w3.weight": "layers.2.ffn.experts.8.w3.weight",210  "model.layers.2.mlp.experts.9.w1.weight": "layers.2.ffn.experts.9.w1.weight",211  "model.layers.2.mlp.experts.9.w2.weight": "layers.2.ffn.experts.9.w2.weight",212  "model.layers.2.mlp.experts.9.w3.weight": "layers.2.ffn.experts.9.w3.weight",213  "model.layers.2.mlp.gate.e_score_correction_bias": "layers.2.ffn.gate.e_score_correction_bias",214  "model.layers.2.mlp.gate.weight": "layers.2.ffn.gate.weight",215  "model.layers.2.mlp.shared_experts.down_proj.weight": "layers.2.ffn.shared_experts.w2.weight",216  "model.layers.2.mlp.shared_experts.gate_proj.weight": "layers.2.ffn.shared_experts.w1.weight",217  "model.layers.2.mlp.shared_experts.up_proj.weight": "layers.2.ffn.shared_experts.w3.weight",218  "model.layers.2.post_attention_layernorm.weight": "layers.2.ffn_norm.weight",219  "model.layers.2.self_attn.kv_norm.weight": "layers.2.attn.kv_norm.weight",220  "model.layers.2.self_attn.kv_proj.weight": "layers.2.attn.wkv.weight",221  "model.layers.2.self_attn.o_a_proj.weight": "layers.2.attn.wo_a.weight",222  "model.layers.2.self_attn.o_b_proj.weight": "layers.2.attn.wo_b.weight",223  "model.layers.2.self_attn.q_a_norm.weight": "layers.2.attn.q_norm.weight",224  "model.layers.2.self_attn.q_a_proj.weight": "layers.2.attn.wq_a.weight",225  "model.layers.2.self_attn.q_b_proj.weight": "layers.2.attn.wq_b.weight",226  "model.layers.2.self_attn.sinks": "layers.2.attn.attn_sink",227  "model.norm.weight": "norm.weight"228}