llama-7b-hf-c0.8 / config.json
李登杰
init
1abeac5
{
"_name_or_path": "/mnt/dolphinfs/ssd_pool/docker/user/hadoop-vacv/lidengjie/projects/fsvd-llm/cache/llama-7b-hf/models/stage1-bf16-usv",
"architectures": [
"LlamaForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": 1,
"eos_token_id": 2,
"hidden_act": "silu",
"hidden_size": 4096,
"initializer_range": 0.02,
"intermediate_size": 11008,
"linear_info": {
"model.layers.0.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2023
},
"model.layers.0.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1124
},
"model.layers.0.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1139
},
"model.layers.0.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 94
},
"model.layers.0.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 538
},
"model.layers.0.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 83
},
"model.layers.0.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1063
},
"model.layers.1.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 1644
},
"model.layers.1.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1288
},
"model.layers.1.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1319
},
"model.layers.1.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 207
},
"model.layers.1.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 945
},
"model.layers.1.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 211
},
"model.layers.1.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 762
},
"model.layers.10.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 193
},
"model.layers.10.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 162
},
"model.layers.10.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 196
},
"model.layers.10.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 94
},
"model.layers.10.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 32
},
"model.layers.10.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 113
},
"model.layers.10.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 99
},
"model.layers.11.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 273
},
"model.layers.11.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 219
},
"model.layers.11.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 268
},
"model.layers.11.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 54
},
"model.layers.11.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 34
},
"model.layers.11.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 62
},
"model.layers.11.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 133
},
"model.layers.12.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 46
},
"model.layers.12.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 84
},
"model.layers.12.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 66
},
"model.layers.12.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 100
},
"model.layers.12.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 63
},
"model.layers.12.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 115
},
"model.layers.12.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 156
},
"model.layers.13.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 270
},
"model.layers.13.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 247
},
"model.layers.13.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 278
},
"model.layers.13.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 40
},
"model.layers.13.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1
},
"model.layers.13.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 42
},
"model.layers.13.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 55
},
"model.layers.14.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 678
},
"model.layers.14.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 467
},
"model.layers.14.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 613
},
"model.layers.14.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 271
},
"model.layers.14.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 702
},
"model.layers.14.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 265
},
"model.layers.14.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 682
},
"model.layers.15.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 996
},
"model.layers.15.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 665
},
"model.layers.15.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 826
},
"model.layers.15.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 126
},
"model.layers.15.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 359
},
"model.layers.15.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 152
},
"model.layers.15.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 455
},
"model.layers.16.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 1673
},
"model.layers.16.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 978
},
"model.layers.16.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1109
},
"model.layers.16.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 128
},
"model.layers.16.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 504
},
"model.layers.16.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 159
},
"model.layers.16.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 469
},
"model.layers.17.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 1545
},
"model.layers.17.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 905
},
"model.layers.17.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1087
},
"model.layers.17.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 262
},
"model.layers.17.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1038
},
"model.layers.17.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 327
},
"model.layers.17.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 785
},
"model.layers.18.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 1414
},
"model.layers.18.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 851
},
"model.layers.18.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 984
},
"model.layers.18.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 298
},
"model.layers.18.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1130
},
"model.layers.18.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 285
},
"model.layers.18.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 857
},
"model.layers.19.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 1348
},
"model.layers.19.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 819
},
"model.layers.19.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 999
},
"model.layers.19.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 160
},
"model.layers.19.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 648
},
"model.layers.19.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 151
},
"model.layers.19.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 497
},
"model.layers.2.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 1450
},
"model.layers.2.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1014
},
"model.layers.2.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1167
},
"model.layers.2.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 257
},
"model.layers.2.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1176
},
"model.layers.2.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 205
},
"model.layers.2.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 862
},
"model.layers.20.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 1412
},
"model.layers.20.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 844
},
"model.layers.20.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 965
},
"model.layers.20.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 151
},
"model.layers.20.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 685
},
"model.layers.20.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 166
},
"model.layers.20.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 587
},
"model.layers.21.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 1355
},
"model.layers.21.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 833
},
"model.layers.21.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 908
},
"model.layers.21.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 137
},
"model.layers.21.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 472
},
"model.layers.21.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 124
},
"model.layers.21.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 393
},
"model.layers.22.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 1113
},
"model.layers.22.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 667
},
"model.layers.22.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 737
},
"model.layers.22.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 92
},
"model.layers.22.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 130
},
"model.layers.22.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 89
},
"model.layers.22.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 150
},
"model.layers.23.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 856
},
"model.layers.23.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 478
},
"model.layers.23.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 586
},
"model.layers.23.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 168
},
"model.layers.23.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 516
},
"model.layers.23.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 144
},
"model.layers.23.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 356
},
"model.layers.24.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 435
},
"model.layers.24.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 242
},
"model.layers.24.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 319
},
"model.layers.24.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 96
},
"model.layers.24.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 440
},
"model.layers.24.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 93
},
"model.layers.24.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 297
},
"model.layers.25.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 147
},
"model.layers.25.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 121
},
"model.layers.25.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 122
},
"model.layers.25.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 11
},
"model.layers.25.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 7
},
"model.layers.25.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 14
},
"model.layers.25.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 13
},
"model.layers.26.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 17
},
"model.layers.26.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 36
},
"model.layers.26.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 20
},
"model.layers.26.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 2
},
"model.layers.26.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1
},
"model.layers.26.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 4
},
"model.layers.26.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1
},
"model.layers.27.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 1
},
"model.layers.27.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 15
},
"model.layers.27.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 6
},
"model.layers.27.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 5
},
"model.layers.27.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1
},
"model.layers.27.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 4
},
"model.layers.27.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 8
},
"model.layers.28.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 304
},
"model.layers.28.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 248
},
"model.layers.28.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 213
},
"model.layers.28.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 65
},
"model.layers.28.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 174
},
"model.layers.28.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 80
},
"model.layers.28.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 238
},
"model.layers.29.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 436
},
"model.layers.29.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 430
},
"model.layers.29.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 334
},
"model.layers.29.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1
},
"model.layers.29.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1
},
"model.layers.29.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 2
},
"model.layers.29.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 2
},
"model.layers.3.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 1216
},
"model.layers.3.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1042
},
"model.layers.3.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1217
},
"model.layers.3.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 228
},
"model.layers.3.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 627
},
"model.layers.3.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 203
},
"model.layers.3.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 574
},
"model.layers.30.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 973
},
"model.layers.30.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 747
},
"model.layers.30.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 683
},
"model.layers.30.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 14
},
"model.layers.30.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 10
},
"model.layers.30.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 14
},
"model.layers.30.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 22
},
"model.layers.31.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2035
},
"model.layers.31.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1270
},
"model.layers.31.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1335
},
"model.layers.31.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 59
},
"model.layers.31.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 214
},
"model.layers.31.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 60
},
"model.layers.31.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 284
},
"model.layers.4.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 960
},
"model.layers.4.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 730
},
"model.layers.4.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 962
},
"model.layers.4.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 271
},
"model.layers.4.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 966
},
"model.layers.4.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 266
},
"model.layers.4.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 699
},
"model.layers.5.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 906
},
"model.layers.5.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 654
},
"model.layers.5.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1019
},
"model.layers.5.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 243
},
"model.layers.5.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 503
},
"model.layers.5.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 233
},
"model.layers.5.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 470
},
"model.layers.6.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 740
},
"model.layers.6.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 535
},
"model.layers.6.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 826
},
"model.layers.6.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 248
},
"model.layers.6.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 456
},
"model.layers.6.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 237
},
"model.layers.6.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 370
},
"model.layers.7.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 437
},
"model.layers.7.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 317
},
"model.layers.7.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 494
},
"model.layers.7.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 244
},
"model.layers.7.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 625
},
"model.layers.7.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 251
},
"model.layers.7.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 545
},
"model.layers.8.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 201
},
"model.layers.8.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 164
},
"model.layers.8.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 203
},
"model.layers.8.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 167
},
"model.layers.8.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 195
},
"model.layers.8.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 149
},
"model.layers.8.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 287
},
"model.layers.9.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 93
},
"model.layers.9.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 118
},
"model.layers.9.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 89
},
"model.layers.9.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 212
},
"model.layers.9.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 284
},
"model.layers.9.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 199
},
"model.layers.9.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 235
}
},
"max_position_embeddings": 2048,
"mlp_bias": false,
"model_type": "llama",
"num_attention_heads": 32,
"num_hidden_layers": 32,
"num_key_value_heads": 32,
"pad_token_id": 0,
"pretraining_tp": 1,
"rms_norm_eps": 1e-06,
"rope_scaling": null,
"rope_theta": 10000.0,
"tie_word_embeddings": false,
"torch_dtype": "bfloat16",
"transformers_version": "4.43.2",
"use_cache": true,
"vocab_size": 32000
}