llama-7b-hf-c0.4 / config.json
李登杰
Init
dcf55ea
{
"_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": 2318
},
"model.layers.0.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1267
},
"model.layers.0.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1251
},
"model.layers.0.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 212
},
"model.layers.0.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 670
},
"model.layers.0.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 178
},
"model.layers.0.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1109
},
"model.layers.1.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2370
},
"model.layers.1.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1712
},
"model.layers.1.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1748
},
"model.layers.1.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 375
},
"model.layers.1.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1250
},
"model.layers.1.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 376
},
"model.layers.1.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1195
},
"model.layers.10.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 1980
},
"model.layers.10.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1607
},
"model.layers.10.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2033
},
"model.layers.10.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 631
},
"model.layers.10.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1141
},
"model.layers.10.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 686
},
"model.layers.10.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1153
},
"model.layers.11.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2249
},
"model.layers.11.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1764
},
"model.layers.11.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2106
},
"model.layers.11.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 551
},
"model.layers.11.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1329
},
"model.layers.11.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 568
},
"model.layers.11.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1569
},
"model.layers.12.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2298
},
"model.layers.12.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1699
},
"model.layers.12.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2100
},
"model.layers.12.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 604
},
"model.layers.12.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1252
},
"model.layers.12.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 696
},
"model.layers.12.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1318
},
"model.layers.13.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2335
},
"model.layers.13.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1761
},
"model.layers.13.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2126
},
"model.layers.13.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 534
},
"model.layers.13.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 867
},
"model.layers.13.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 611
},
"model.layers.13.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1152
},
"model.layers.14.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2344
},
"model.layers.14.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1835
},
"model.layers.14.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2151
},
"model.layers.14.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 658
},
"model.layers.14.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1202
},
"model.layers.14.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 702
},
"model.layers.14.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1406
},
"model.layers.15.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2409
},
"model.layers.15.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1903
},
"model.layers.15.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2155
},
"model.layers.15.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 572
},
"model.layers.15.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1327
},
"model.layers.15.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 688
},
"model.layers.15.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1472
},
"model.layers.16.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 0
},
"model.layers.16.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1985
},
"model.layers.16.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2214
},
"model.layers.16.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 613
},
"model.layers.16.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1246
},
"model.layers.16.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 665
},
"model.layers.16.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1421
},
"model.layers.17.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 0
},
"model.layers.17.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1962
},
"model.layers.17.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2190
},
"model.layers.17.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 612
},
"model.layers.17.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1292
},
"model.layers.17.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 684
},
"model.layers.17.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1394
},
"model.layers.18.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 0
},
"model.layers.18.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1990
},
"model.layers.18.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2188
},
"model.layers.18.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 520
},
"model.layers.18.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1370
},
"model.layers.18.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 555
},
"model.layers.18.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1391
},
"model.layers.19.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 0
},
"model.layers.19.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1969
},
"model.layers.19.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2140
},
"model.layers.19.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 480
},
"model.layers.19.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 956
},
"model.layers.19.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 517
},
"model.layers.19.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1170
},
"model.layers.2.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 0
},
"model.layers.2.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1771
},
"model.layers.2.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1985
},
"model.layers.2.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 571
},
"model.layers.2.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1601
},
"model.layers.2.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 512
},
"model.layers.2.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1451
},
"model.layers.20.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 0
},
"model.layers.20.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1915
},
"model.layers.20.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2126
},
"model.layers.20.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 513
},
"model.layers.20.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1027
},
"model.layers.20.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 562
},
"model.layers.20.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1215
},
"model.layers.21.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2460
},
"model.layers.21.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1922
},
"model.layers.21.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2074
},
"model.layers.21.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 368
},
"model.layers.21.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 900
},
"model.layers.21.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 374
},
"model.layers.21.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1093
},
"model.layers.22.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2339
},
"model.layers.22.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1911
},
"model.layers.22.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2050
},
"model.layers.22.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 335
},
"model.layers.22.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 533
},
"model.layers.22.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 352
},
"model.layers.22.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 766
},
"model.layers.23.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2237
},
"model.layers.23.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1903
},
"model.layers.23.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2040
},
"model.layers.23.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 372
},
"model.layers.23.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 647
},
"model.layers.23.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 343
},
"model.layers.23.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 827
},
"model.layers.24.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2129
},
"model.layers.24.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1912
},
"model.layers.24.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2049
},
"model.layers.24.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 253
},
"model.layers.24.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 447
},
"model.layers.24.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 250
},
"model.layers.24.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 753
},
"model.layers.25.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2121
},
"model.layers.25.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1976
},
"model.layers.25.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2068
},
"model.layers.25.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 206
},
"model.layers.25.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 375
},
"model.layers.25.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 247
},
"model.layers.25.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 700
},
"model.layers.26.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2077
},
"model.layers.26.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1994
},
"model.layers.26.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2075
},
"model.layers.26.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 194
},
"model.layers.26.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 387
},
"model.layers.26.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 227
},
"model.layers.26.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 710
},
"model.layers.27.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2056
},
"model.layers.27.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2029
},
"model.layers.27.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2098
},
"model.layers.27.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 172
},
"model.layers.27.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 279
},
"model.layers.27.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 190
},
"model.layers.27.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 579
},
"model.layers.28.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2102
},
"model.layers.28.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2085
},
"model.layers.28.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2179
},
"model.layers.28.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 281
},
"model.layers.28.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 529
},
"model.layers.28.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 291
},
"model.layers.28.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 773
},
"model.layers.29.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2235
},
"model.layers.29.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2205
},
"model.layers.29.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2324
},
"model.layers.29.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 157
},
"model.layers.29.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 381
},
"model.layers.29.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 218
},
"model.layers.29.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 643
},
"model.layers.3.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2278
},
"model.layers.3.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2094
},
"model.layers.3.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2251
},
"model.layers.3.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 542
},
"model.layers.3.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 920
},
"model.layers.3.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 534
},
"model.layers.3.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1283
},
"model.layers.30.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2449
},
"model.layers.30.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2307
},
"model.layers.30.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2394
},
"model.layers.30.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 143
},
"model.layers.30.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 438
},
"model.layers.30.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 166
},
"model.layers.30.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 652
},
"model.layers.31.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 0
},
"model.layers.31.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2427
},
"model.layers.31.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 0
},
"model.layers.31.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 211
},
"model.layers.31.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 516
},
"model.layers.31.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 205
},
"model.layers.31.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 713
},
"model.layers.4.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2198
},
"model.layers.4.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1964
},
"model.layers.4.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2219
},
"model.layers.4.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 658
},
"model.layers.4.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1302
},
"model.layers.4.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 640
},
"model.layers.4.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1316
},
"model.layers.5.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2294
},
"model.layers.5.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1899
},
"model.layers.5.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2332
},
"model.layers.5.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 638
},
"model.layers.5.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1002
},
"model.layers.5.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 654
},
"model.layers.5.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1228
},
"model.layers.6.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2214
},
"model.layers.6.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1834
},
"model.layers.6.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2277
},
"model.layers.6.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 642
},
"model.layers.6.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1225
},
"model.layers.6.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 717
},
"model.layers.6.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1232
},
"model.layers.7.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2072
},
"model.layers.7.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1678
},
"model.layers.7.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2130
},
"model.layers.7.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 564
},
"model.layers.7.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1167
},
"model.layers.7.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 557
},
"model.layers.7.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1256
},
"model.layers.8.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2066
},
"model.layers.8.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1599
},
"model.layers.8.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2113
},
"model.layers.8.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 659
},
"model.layers.8.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1130
},
"model.layers.8.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 650
},
"model.layers.8.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1292
},
"model.layers.9.mlp.down_proj": {
"bias": true,
"in_features": 11008,
"out_features": 4096,
"rank": 2057
},
"model.layers.9.mlp.gate_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 1606
},
"model.layers.9.mlp.up_proj": {
"bias": true,
"in_features": 4096,
"out_features": 11008,
"rank": 2054
},
"model.layers.9.self_attn.k_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 633
},
"model.layers.9.self_attn.o_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1003
},
"model.layers.9.self_attn.q_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 662
},
"model.layers.9.self_attn.v_proj": {
"bias": true,
"in_features": 4096,
"out_features": 4096,
"rank": 1049
}
},
"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
}