Adding `safetensors` variant of this model

model-00001-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a49f817916047cbff20d60d3bff44a1d8593830f91984a9391fa3c40d07d45b
+size 4605545152

model-00002-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:040971099db6e61af1de56317e52c7ea83ab3030e436f55ce503ca70a87edda3
+size 4899212704

model-00003-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5dd4d393d69604e7de75cb2edb4b889d3f29ac50267f9788f7cb406cd86cdeed
+size 4613933848

model-00004-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f23b33f2e625e9ff7ac071edc8c346d2353a11240c63fd7ffe80c781efc6567
+size 4899212704

model-00005-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eaf21e43ffde2abbee6db42d4bcd4cdc6a14f43352e449903668fa33faada733
+size 4613933872

model-00006-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:247e9e196b7d3a07d3e460aee9f763a13df0a7c6a1abac1eb47d2ba14e95e258
+size 4899212736

model-00007-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e4aa0feeeb4f7adeebce95f43df7af8712a460bbdfcbb1a31981b2986b500e8
+size 4613933872

model-00008-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9200d515a9c60b2f2ffa6be02a7b45cd53c0d76d428d3bc0a94224f4fbfb6f23
+size 4899212736

model-00009-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:acfee3bfb81e9766cf65478eaf2f96cef3b2f5eb8d501d27157b85647df4cbfe
+size 4613933872

model-00010-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ec343d90ffe8162c1c4b94d5fbafc5272d1ee525d2198e0f41dbdbd429051c8
+size 4899212736

model-00011-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6697c3c59cc4f2a82f71026611250fe6218dd30378e72f4cb9affec71efd629d
+size 4613933872

model-00012-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a925444af6d0e4913cab237ef4809659c218f3f1b0793bd6f605fac06b3c368
+size 4899212736

model-00013-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34a353c866e0866606dc2f50f300bb5a37d37132e9e2d3ec526bea6af3caadfd
+size 4613933872

model-00014-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf1e8f76204b4e2a50e39bebdb5f9103482ef0f5db13e8009a8487fcdedf8c07
+size 4899212736

model-00015-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:473b586a9a6e896503f0ef28c02ccc231c298854493367dbb5492188540054fa
+size 4613933872

model-00016-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de0e9f35d6901ae61fb75e92e28afdc2cfc54f35808fabb90aa6ac88c5acb405
+size 4899212736

model-00017-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d176bfddb471cc984e82b61ec6be54e1f47a3b199042ad5b4eaf53aa05c3876b
+size 4613933872

model-00018-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0513dd48bf90a7d98d42311ffc0e5bc034ac6e02b94a20b7fbe8d654068901ea
+size 1895924952

model.safetensors.index.json

@@ -0,0 +1,491 @@
+{
+    "metadata": {
+        "total_size": 82606587904
+    },
+    "weight_map": {
+        "lm_head.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.0.ln_attn.bias": "model-00001-of-00018.safetensors",
+        "transformer.h.0.ln_attn.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.0.ln_mlp.bias": "model-00001-of-00018.safetensors",
+        "transformer.h.0.ln_mlp.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.0.mlp.dense_4h_to_h.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.0.mlp.dense_h_to_4h.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.0.self_attention.dense.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.0.self_attention.query_key_value.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.1.ln_attn.bias": "model-00001-of-00018.safetensors",
+        "transformer.h.1.ln_attn.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.1.ln_mlp.bias": "model-00001-of-00018.safetensors",
+        "transformer.h.1.ln_mlp.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.1.mlp.dense_4h_to_h.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.1.mlp.dense_h_to_4h.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.1.self_attention.dense.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.1.self_attention.query_key_value.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.10.ln_attn.bias": "model-00004-of-00018.safetensors",
+        "transformer.h.10.ln_attn.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.10.ln_mlp.bias": "model-00004-of-00018.safetensors",
+        "transformer.h.10.ln_mlp.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.10.mlp.dense_4h_to_h.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.10.mlp.dense_h_to_4h.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.10.self_attention.dense.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.10.self_attention.query_key_value.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.11.ln_attn.bias": "model-00004-of-00018.safetensors",
+        "transformer.h.11.ln_attn.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.11.ln_mlp.bias": "model-00004-of-00018.safetensors",
+        "transformer.h.11.ln_mlp.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.11.mlp.dense_4h_to_h.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.11.mlp.dense_h_to_4h.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.11.self_attention.dense.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.11.self_attention.query_key_value.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.12.ln_attn.bias": "model-00004-of-00018.safetensors",
+        "transformer.h.12.ln_attn.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.12.ln_mlp.bias": "model-00004-of-00018.safetensors",
+        "transformer.h.12.ln_mlp.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.12.mlp.dense_4h_to_h.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.12.mlp.dense_h_to_4h.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.12.self_attention.dense.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.12.self_attention.query_key_value.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.13.ln_attn.bias": "model-00004-of-00018.safetensors",
+        "transformer.h.13.ln_attn.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.13.ln_mlp.bias": "model-00004-of-00018.safetensors",
+        "transformer.h.13.ln_mlp.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.13.mlp.dense_4h_to_h.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.13.mlp.dense_h_to_4h.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.13.self_attention.dense.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.13.self_attention.query_key_value.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.14.ln_attn.bias": "model-00005-of-00018.safetensors",
+        "transformer.h.14.ln_attn.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.14.ln_mlp.bias": "model-00005-of-00018.safetensors",
+        "transformer.h.14.ln_mlp.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.14.mlp.dense_4h_to_h.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.14.mlp.dense_h_to_4h.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.14.self_attention.dense.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.14.self_attention.query_key_value.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.15.ln_attn.bias": "model-00005-of-00018.safetensors",
+        "transformer.h.15.ln_attn.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.15.ln_mlp.bias": "model-00005-of-00018.safetensors",
+        "transformer.h.15.ln_mlp.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.15.mlp.dense_4h_to_h.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.15.mlp.dense_h_to_4h.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.15.self_attention.dense.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.15.self_attention.query_key_value.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.16.ln_attn.bias": "model-00005-of-00018.safetensors",
+        "transformer.h.16.ln_attn.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.16.ln_mlp.bias": "model-00005-of-00018.safetensors",
+        "transformer.h.16.ln_mlp.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.16.mlp.dense_4h_to_h.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.16.mlp.dense_h_to_4h.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.16.self_attention.dense.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.16.self_attention.query_key_value.weight": "model-00005-of-00018.safetensors",
+        "transformer.h.17.ln_attn.bias": "model-00006-of-00018.safetensors",
+        "transformer.h.17.ln_attn.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.17.ln_mlp.bias": "model-00006-of-00018.safetensors",
+        "transformer.h.17.ln_mlp.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.17.mlp.dense_4h_to_h.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.17.mlp.dense_h_to_4h.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.17.self_attention.dense.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.17.self_attention.query_key_value.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.18.ln_attn.bias": "model-00006-of-00018.safetensors",
+        "transformer.h.18.ln_attn.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.18.ln_mlp.bias": "model-00006-of-00018.safetensors",
+        "transformer.h.18.ln_mlp.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.18.mlp.dense_4h_to_h.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.18.mlp.dense_h_to_4h.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.18.self_attention.dense.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.18.self_attention.query_key_value.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.19.ln_attn.bias": "model-00006-of-00018.safetensors",
+        "transformer.h.19.ln_attn.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.19.ln_mlp.bias": "model-00006-of-00018.safetensors",
+        "transformer.h.19.ln_mlp.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.19.mlp.dense_4h_to_h.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.19.mlp.dense_h_to_4h.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.19.self_attention.dense.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.19.self_attention.query_key_value.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.2.ln_attn.bias": "model-00001-of-00018.safetensors",
+        "transformer.h.2.ln_attn.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.2.ln_mlp.bias": "model-00001-of-00018.safetensors",
+        "transformer.h.2.ln_mlp.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.2.mlp.dense_4h_to_h.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.2.mlp.dense_h_to_4h.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.2.self_attention.dense.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.2.self_attention.query_key_value.weight": "model-00001-of-00018.safetensors",
+        "transformer.h.20.ln_attn.bias": "model-00006-of-00018.safetensors",
+        "transformer.h.20.ln_attn.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.20.ln_mlp.bias": "model-00006-of-00018.safetensors",
+        "transformer.h.20.ln_mlp.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.20.mlp.dense_4h_to_h.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.20.mlp.dense_h_to_4h.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.20.self_attention.dense.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.20.self_attention.query_key_value.weight": "model-00006-of-00018.safetensors",
+        "transformer.h.21.ln_attn.bias": "model-00007-of-00018.safetensors",
+        "transformer.h.21.ln_attn.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.21.ln_mlp.bias": "model-00007-of-00018.safetensors",
+        "transformer.h.21.ln_mlp.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.21.mlp.dense_4h_to_h.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.21.mlp.dense_h_to_4h.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.21.self_attention.dense.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.21.self_attention.query_key_value.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.22.ln_attn.bias": "model-00007-of-00018.safetensors",
+        "transformer.h.22.ln_attn.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.22.ln_mlp.bias": "model-00007-of-00018.safetensors",
+        "transformer.h.22.ln_mlp.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.22.mlp.dense_4h_to_h.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.22.mlp.dense_h_to_4h.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.22.self_attention.dense.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.22.self_attention.query_key_value.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.23.ln_attn.bias": "model-00007-of-00018.safetensors",
+        "transformer.h.23.ln_attn.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.23.ln_mlp.bias": "model-00007-of-00018.safetensors",
+        "transformer.h.23.ln_mlp.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.23.mlp.dense_4h_to_h.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.23.mlp.dense_h_to_4h.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.23.self_attention.dense.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.23.self_attention.query_key_value.weight": "model-00007-of-00018.safetensors",
+        "transformer.h.24.ln_attn.bias": "model-00008-of-00018.safetensors",
+        "transformer.h.24.ln_attn.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.24.ln_mlp.bias": "model-00008-of-00018.safetensors",
+        "transformer.h.24.ln_mlp.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.24.mlp.dense_4h_to_h.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.24.mlp.dense_h_to_4h.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.24.self_attention.dense.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.24.self_attention.query_key_value.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.25.ln_attn.bias": "model-00008-of-00018.safetensors",
+        "transformer.h.25.ln_attn.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.25.ln_mlp.bias": "model-00008-of-00018.safetensors",
+        "transformer.h.25.ln_mlp.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.25.mlp.dense_4h_to_h.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.25.mlp.dense_h_to_4h.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.25.self_attention.dense.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.25.self_attention.query_key_value.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.26.ln_attn.bias": "model-00008-of-00018.safetensors",
+        "transformer.h.26.ln_attn.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.26.ln_mlp.bias": "model-00008-of-00018.safetensors",
+        "transformer.h.26.ln_mlp.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.26.mlp.dense_4h_to_h.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.26.mlp.dense_h_to_4h.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.26.self_attention.dense.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.26.self_attention.query_key_value.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.27.ln_attn.bias": "model-00008-of-00018.safetensors",
+        "transformer.h.27.ln_attn.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.27.ln_mlp.bias": "model-00008-of-00018.safetensors",
+        "transformer.h.27.ln_mlp.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.27.mlp.dense_4h_to_h.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.27.mlp.dense_h_to_4h.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.27.self_attention.dense.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.27.self_attention.query_key_value.weight": "model-00008-of-00018.safetensors",
+        "transformer.h.28.ln_attn.bias": "model-00009-of-00018.safetensors",
+        "transformer.h.28.ln_attn.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.28.ln_mlp.bias": "model-00009-of-00018.safetensors",
+        "transformer.h.28.ln_mlp.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.28.mlp.dense_4h_to_h.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.28.mlp.dense_h_to_4h.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.28.self_attention.dense.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.28.self_attention.query_key_value.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.29.ln_attn.bias": "model-00009-of-00018.safetensors",
+        "transformer.h.29.ln_attn.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.29.ln_mlp.bias": "model-00009-of-00018.safetensors",
+        "transformer.h.29.ln_mlp.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.29.mlp.dense_4h_to_h.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.29.mlp.dense_h_to_4h.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.29.self_attention.dense.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.29.self_attention.query_key_value.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.3.ln_attn.bias": "model-00002-of-00018.safetensors",
+        "transformer.h.3.ln_attn.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.3.ln_mlp.bias": "model-00002-of-00018.safetensors",
+        "transformer.h.3.ln_mlp.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.3.mlp.dense_4h_to_h.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.3.mlp.dense_h_to_4h.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.3.self_attention.dense.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.3.self_attention.query_key_value.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.30.ln_attn.bias": "model-00009-of-00018.safetensors",
+        "transformer.h.30.ln_attn.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.30.ln_mlp.bias": "model-00009-of-00018.safetensors",
+        "transformer.h.30.ln_mlp.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.30.mlp.dense_4h_to_h.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.30.mlp.dense_h_to_4h.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.30.self_attention.dense.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.30.self_attention.query_key_value.weight": "model-00009-of-00018.safetensors",
+        "transformer.h.31.ln_attn.bias": "model-00010-of-00018.safetensors",
+        "transformer.h.31.ln_attn.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.31.ln_mlp.bias": "model-00010-of-00018.safetensors",
+        "transformer.h.31.ln_mlp.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.31.mlp.dense_4h_to_h.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.31.mlp.dense_h_to_4h.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.31.self_attention.dense.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.31.self_attention.query_key_value.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.32.ln_attn.bias": "model-00010-of-00018.safetensors",
+        "transformer.h.32.ln_attn.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.32.ln_mlp.bias": "model-00010-of-00018.safetensors",
+        "transformer.h.32.ln_mlp.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.32.mlp.dense_4h_to_h.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.32.mlp.dense_h_to_4h.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.32.self_attention.dense.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.32.self_attention.query_key_value.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.33.ln_attn.bias": "model-00010-of-00018.safetensors",
+        "transformer.h.33.ln_attn.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.33.ln_mlp.bias": "model-00010-of-00018.safetensors",
+        "transformer.h.33.ln_mlp.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.33.mlp.dense_4h_to_h.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.33.mlp.dense_h_to_4h.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.33.self_attention.dense.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.33.self_attention.query_key_value.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.34.ln_attn.bias": "model-00010-of-00018.safetensors",
+        "transformer.h.34.ln_attn.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.34.ln_mlp.bias": "model-00010-of-00018.safetensors",
+        "transformer.h.34.ln_mlp.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.34.mlp.dense_4h_to_h.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.34.mlp.dense_h_to_4h.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.34.self_attention.dense.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.34.self_attention.query_key_value.weight": "model-00010-of-00018.safetensors",
+        "transformer.h.35.ln_attn.bias": "model-00011-of-00018.safetensors",
+        "transformer.h.35.ln_attn.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.35.ln_mlp.bias": "model-00011-of-00018.safetensors",
+        "transformer.h.35.ln_mlp.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.35.mlp.dense_4h_to_h.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.35.mlp.dense_h_to_4h.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.35.self_attention.dense.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.35.self_attention.query_key_value.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.36.ln_attn.bias": "model-00011-of-00018.safetensors",
+        "transformer.h.36.ln_attn.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.36.ln_mlp.bias": "model-00011-of-00018.safetensors",
+        "transformer.h.36.ln_mlp.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.36.mlp.dense_4h_to_h.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.36.mlp.dense_h_to_4h.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.36.self_attention.dense.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.36.self_attention.query_key_value.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.37.ln_attn.bias": "model-00011-of-00018.safetensors",
+        "transformer.h.37.ln_attn.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.37.ln_mlp.bias": "model-00011-of-00018.safetensors",
+        "transformer.h.37.ln_mlp.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.37.mlp.dense_4h_to_h.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.37.mlp.dense_h_to_4h.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.37.self_attention.dense.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.37.self_attention.query_key_value.weight": "model-00011-of-00018.safetensors",
+        "transformer.h.38.ln_attn.bias": "model-00012-of-00018.safetensors",
+        "transformer.h.38.ln_attn.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.38.ln_mlp.bias": "model-00012-of-00018.safetensors",
+        "transformer.h.38.ln_mlp.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.38.mlp.dense_4h_to_h.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.38.mlp.dense_h_to_4h.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.38.self_attention.dense.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.38.self_attention.query_key_value.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.39.ln_attn.bias": "model-00012-of-00018.safetensors",
+        "transformer.h.39.ln_attn.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.39.ln_mlp.bias": "model-00012-of-00018.safetensors",
+        "transformer.h.39.ln_mlp.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.39.mlp.dense_4h_to_h.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.39.mlp.dense_h_to_4h.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.39.self_attention.dense.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.39.self_attention.query_key_value.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.4.ln_attn.bias": "model-00002-of-00018.safetensors",
+        "transformer.h.4.ln_attn.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.4.ln_mlp.bias": "model-00002-of-00018.safetensors",
+        "transformer.h.4.ln_mlp.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.4.mlp.dense_4h_to_h.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.4.mlp.dense_h_to_4h.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.4.self_attention.dense.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.4.self_attention.query_key_value.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.40.ln_attn.bias": "model-00012-of-00018.safetensors",
+        "transformer.h.40.ln_attn.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.40.ln_mlp.bias": "model-00012-of-00018.safetensors",
+        "transformer.h.40.ln_mlp.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.40.mlp.dense_4h_to_h.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.40.mlp.dense_h_to_4h.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.40.self_attention.dense.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.40.self_attention.query_key_value.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.41.ln_attn.bias": "model-00012-of-00018.safetensors",
+        "transformer.h.41.ln_attn.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.41.ln_mlp.bias": "model-00012-of-00018.safetensors",
+        "transformer.h.41.ln_mlp.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.41.mlp.dense_4h_to_h.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.41.mlp.dense_h_to_4h.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.41.self_attention.dense.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.41.self_attention.query_key_value.weight": "model-00012-of-00018.safetensors",
+        "transformer.h.42.ln_attn.bias": "model-00013-of-00018.safetensors",
+        "transformer.h.42.ln_attn.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.42.ln_mlp.bias": "model-00013-of-00018.safetensors",
+        "transformer.h.42.ln_mlp.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.42.mlp.dense_4h_to_h.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.42.mlp.dense_h_to_4h.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.42.self_attention.dense.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.42.self_attention.query_key_value.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.43.ln_attn.bias": "model-00013-of-00018.safetensors",
+        "transformer.h.43.ln_attn.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.43.ln_mlp.bias": "model-00013-of-00018.safetensors",
+        "transformer.h.43.ln_mlp.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.43.mlp.dense_4h_to_h.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.43.mlp.dense_h_to_4h.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.43.self_attention.dense.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.43.self_attention.query_key_value.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.44.ln_attn.bias": "model-00013-of-00018.safetensors",
+        "transformer.h.44.ln_attn.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.44.ln_mlp.bias": "model-00013-of-00018.safetensors",
+        "transformer.h.44.ln_mlp.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.44.mlp.dense_4h_to_h.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.44.mlp.dense_h_to_4h.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.44.self_attention.dense.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.44.self_attention.query_key_value.weight": "model-00013-of-00018.safetensors",
+        "transformer.h.45.ln_attn.bias": "model-00014-of-00018.safetensors",
+        "transformer.h.45.ln_attn.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.45.ln_mlp.bias": "model-00014-of-00018.safetensors",
+        "transformer.h.45.ln_mlp.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.45.mlp.dense_4h_to_h.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.45.mlp.dense_h_to_4h.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.45.self_attention.dense.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.45.self_attention.query_key_value.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.46.ln_attn.bias": "model-00014-of-00018.safetensors",
+        "transformer.h.46.ln_attn.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.46.ln_mlp.bias": "model-00014-of-00018.safetensors",
+        "transformer.h.46.ln_mlp.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.46.mlp.dense_4h_to_h.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.46.mlp.dense_h_to_4h.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.46.self_attention.dense.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.46.self_attention.query_key_value.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.47.ln_attn.bias": "model-00014-of-00018.safetensors",
+        "transformer.h.47.ln_attn.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.47.ln_mlp.bias": "model-00014-of-00018.safetensors",
+        "transformer.h.47.ln_mlp.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.47.mlp.dense_4h_to_h.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.47.mlp.dense_h_to_4h.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.47.self_attention.dense.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.47.self_attention.query_key_value.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.48.ln_attn.bias": "model-00014-of-00018.safetensors",
+        "transformer.h.48.ln_attn.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.48.ln_mlp.bias": "model-00014-of-00018.safetensors",
+        "transformer.h.48.ln_mlp.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.48.mlp.dense_4h_to_h.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.48.mlp.dense_h_to_4h.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.48.self_attention.dense.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.48.self_attention.query_key_value.weight": "model-00014-of-00018.safetensors",
+        "transformer.h.49.ln_attn.bias": "model-00015-of-00018.safetensors",
+        "transformer.h.49.ln_attn.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.49.ln_mlp.bias": "model-00015-of-00018.safetensors",
+        "transformer.h.49.ln_mlp.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.49.mlp.dense_4h_to_h.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.49.mlp.dense_h_to_4h.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.49.self_attention.dense.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.49.self_attention.query_key_value.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.5.ln_attn.bias": "model-00002-of-00018.safetensors",
+        "transformer.h.5.ln_attn.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.5.ln_mlp.bias": "model-00002-of-00018.safetensors",
+        "transformer.h.5.ln_mlp.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.5.mlp.dense_4h_to_h.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.5.mlp.dense_h_to_4h.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.5.self_attention.dense.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.5.self_attention.query_key_value.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.50.ln_attn.bias": "model-00015-of-00018.safetensors",
+        "transformer.h.50.ln_attn.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.50.ln_mlp.bias": "model-00015-of-00018.safetensors",
+        "transformer.h.50.ln_mlp.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.50.mlp.dense_4h_to_h.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.50.mlp.dense_h_to_4h.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.50.self_attention.dense.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.50.self_attention.query_key_value.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.51.ln_attn.bias": "model-00015-of-00018.safetensors",
+        "transformer.h.51.ln_attn.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.51.ln_mlp.bias": "model-00015-of-00018.safetensors",
+        "transformer.h.51.ln_mlp.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.51.mlp.dense_4h_to_h.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.51.mlp.dense_h_to_4h.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.51.self_attention.dense.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.51.self_attention.query_key_value.weight": "model-00015-of-00018.safetensors",
+        "transformer.h.52.ln_attn.bias": "model-00016-of-00018.safetensors",
+        "transformer.h.52.ln_attn.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.52.ln_mlp.bias": "model-00016-of-00018.safetensors",
+        "transformer.h.52.ln_mlp.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.52.mlp.dense_4h_to_h.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.52.mlp.dense_h_to_4h.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.52.self_attention.dense.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.52.self_attention.query_key_value.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.53.ln_attn.bias": "model-00016-of-00018.safetensors",
+        "transformer.h.53.ln_attn.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.53.ln_mlp.bias": "model-00016-of-00018.safetensors",
+        "transformer.h.53.ln_mlp.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.53.mlp.dense_4h_to_h.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.53.mlp.dense_h_to_4h.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.53.self_attention.dense.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.53.self_attention.query_key_value.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.54.ln_attn.bias": "model-00016-of-00018.safetensors",
+        "transformer.h.54.ln_attn.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.54.ln_mlp.bias": "model-00016-of-00018.safetensors",
+        "transformer.h.54.ln_mlp.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.54.mlp.dense_4h_to_h.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.54.mlp.dense_h_to_4h.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.54.self_attention.dense.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.54.self_attention.query_key_value.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.55.ln_attn.bias": "model-00016-of-00018.safetensors",
+        "transformer.h.55.ln_attn.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.55.ln_mlp.bias": "model-00016-of-00018.safetensors",
+        "transformer.h.55.ln_mlp.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.55.mlp.dense_4h_to_h.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.55.mlp.dense_h_to_4h.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.55.self_attention.dense.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.55.self_attention.query_key_value.weight": "model-00016-of-00018.safetensors",
+        "transformer.h.56.ln_attn.bias": "model-00017-of-00018.safetensors",
+        "transformer.h.56.ln_attn.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.56.ln_mlp.bias": "model-00017-of-00018.safetensors",
+        "transformer.h.56.ln_mlp.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.56.mlp.dense_4h_to_h.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.56.mlp.dense_h_to_4h.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.56.self_attention.dense.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.56.self_attention.query_key_value.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.57.ln_attn.bias": "model-00017-of-00018.safetensors",
+        "transformer.h.57.ln_attn.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.57.ln_mlp.bias": "model-00017-of-00018.safetensors",
+        "transformer.h.57.ln_mlp.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.57.mlp.dense_4h_to_h.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.57.mlp.dense_h_to_4h.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.57.self_attention.dense.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.57.self_attention.query_key_value.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.58.ln_attn.bias": "model-00017-of-00018.safetensors",
+        "transformer.h.58.ln_attn.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.58.ln_mlp.bias": "model-00017-of-00018.safetensors",
+        "transformer.h.58.ln_mlp.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.58.mlp.dense_4h_to_h.weight": "model-00018-of-00018.safetensors",
+        "transformer.h.58.mlp.dense_h_to_4h.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.58.self_attention.dense.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.58.self_attention.query_key_value.weight": "model-00017-of-00018.safetensors",
+        "transformer.h.59.ln_attn.bias": "model-00018-of-00018.safetensors",
+        "transformer.h.59.ln_attn.weight": "model-00018-of-00018.safetensors",
+        "transformer.h.59.ln_mlp.bias": "model-00018-of-00018.safetensors",
+        "transformer.h.59.ln_mlp.weight": "model-00018-of-00018.safetensors",
+        "transformer.h.59.mlp.dense_4h_to_h.weight": "model-00018-of-00018.safetensors",
+        "transformer.h.59.mlp.dense_h_to_4h.weight": "model-00018-of-00018.safetensors",
+        "transformer.h.59.self_attention.dense.weight": "model-00018-of-00018.safetensors",
+        "transformer.h.59.self_attention.query_key_value.weight": "model-00018-of-00018.safetensors",
+        "transformer.h.6.ln_attn.bias": "model-00002-of-00018.safetensors",
+        "transformer.h.6.ln_attn.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.6.ln_mlp.bias": "model-00002-of-00018.safetensors",
+        "transformer.h.6.ln_mlp.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.6.mlp.dense_4h_to_h.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.6.mlp.dense_h_to_4h.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.6.self_attention.dense.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.6.self_attention.query_key_value.weight": "model-00002-of-00018.safetensors",
+        "transformer.h.7.ln_attn.bias": "model-00003-of-00018.safetensors",
+        "transformer.h.7.ln_attn.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.7.ln_mlp.bias": "model-00003-of-00018.safetensors",
+        "transformer.h.7.ln_mlp.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.7.mlp.dense_4h_to_h.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.7.mlp.dense_h_to_4h.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.7.self_attention.dense.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.7.self_attention.query_key_value.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.8.ln_attn.bias": "model-00003-of-00018.safetensors",
+        "transformer.h.8.ln_attn.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.8.ln_mlp.bias": "model-00003-of-00018.safetensors",
+        "transformer.h.8.ln_mlp.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.8.mlp.dense_4h_to_h.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.8.mlp.dense_h_to_4h.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.8.self_attention.dense.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.8.self_attention.query_key_value.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.9.ln_attn.bias": "model-00003-of-00018.safetensors",
+        "transformer.h.9.ln_attn.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.9.ln_mlp.bias": "model-00003-of-00018.safetensors",
+        "transformer.h.9.ln_mlp.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.9.mlp.dense_4h_to_h.weight": "model-00004-of-00018.safetensors",
+        "transformer.h.9.mlp.dense_h_to_4h.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.9.self_attention.dense.weight": "model-00003-of-00018.safetensors",
+        "transformer.h.9.self_attention.query_key_value.weight": "model-00003-of-00018.safetensors",
+        "transformer.ln_f.bias": "model-00018-of-00018.safetensors",
+        "transformer.ln_f.weight": "model-00018-of-00018.safetensors",
+        "transformer.word_embeddings.weight": "model-00001-of-00018.safetensors"
+    }
+}