Adding `safetensors` variant of this model (#1)

- Adding `safetensors` variant of this model (be7efdf5c479e6d095757e68d0a3f892132de846)

model-00001-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4c9caca1d7740fdfbef3b7bf24fba1a01e9331f6e3ff7d29afe25f8fa36a16f2
+size 9904362872

model-00002-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da14073f1e7e6732aecd4ceb54db791de7cf7b0f917207a4c4160455b3607a4e
+size 9860447256

model-00003-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d6f27d81c95b94bd99d0c401467acd99f26af9db938aaa63c3032a9f79bdb917
+size 9854228560

model-00004-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e26fe9b6059b1cb3a0d3a64bfd2bf7554b013aeacb7436beb200bd2286d36d9e
+size 9860447304

model-00005-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4253efa3bf9db4c16fe2be558159908708432660546f6ddcf8606513d6b38f7
+size 9854228560

model-00006-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a53717ca1f6d254c022a488b7dd0b61b1d32c7abf26a466334aacf454d4139c6
+size 9860447304

model-00007-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98d944f242d729c8a4b43a2338fc54d874265a59ea39e18535e8d38760fdd73e
+size 2875714840

model.safetensors.index.json

@@ -0,0 +1,492 @@
+{
+    "metadata": {
+        "total_size": 62069825536
+    },
+    "weight_map": {
+        "lm_head.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.0.attn.c_attn.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.0.attn.c_attn.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.0.attn.c_proj.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.0.attn.c_proj.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.0.ln_1.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.0.ln_1.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.0.ln_2.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.0.ln_2.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.0.mlp.c_fc.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.0.mlp.c_fc.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.0.mlp.c_proj.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.0.mlp.c_proj.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.1.attn.c_attn.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.1.attn.c_attn.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.1.attn.c_proj.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.1.attn.c_proj.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.1.ln_1.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.1.ln_1.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.1.ln_2.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.1.ln_2.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.1.mlp.c_fc.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.1.mlp.c_fc.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.1.mlp.c_proj.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.1.mlp.c_proj.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.10.attn.c_attn.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.10.attn.c_attn.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.10.attn.c_proj.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.10.attn.c_proj.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.10.ln_1.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.10.ln_1.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.10.ln_2.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.10.ln_2.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.10.mlp.c_fc.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.10.mlp.c_fc.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.10.mlp.c_proj.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.10.mlp.c_proj.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.11.attn.c_attn.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.11.attn.c_attn.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.11.attn.c_proj.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.11.attn.c_proj.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.11.ln_1.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.11.ln_1.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.11.ln_2.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.11.ln_2.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.11.mlp.c_fc.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.11.mlp.c_fc.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.11.mlp.c_proj.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.11.mlp.c_proj.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.12.attn.c_attn.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.12.attn.c_attn.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.12.attn.c_proj.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.12.attn.c_proj.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.12.ln_1.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.12.ln_1.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.12.ln_2.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.12.ln_2.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.12.mlp.c_fc.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.12.mlp.c_fc.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.12.mlp.c_proj.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.12.mlp.c_proj.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.13.attn.c_attn.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.13.attn.c_attn.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.13.attn.c_proj.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.13.attn.c_proj.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.13.ln_1.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.13.ln_1.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.13.ln_2.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.13.ln_2.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.13.mlp.c_fc.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.13.mlp.c_fc.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.13.mlp.c_proj.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.13.mlp.c_proj.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.14.attn.c_attn.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.14.attn.c_attn.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.14.attn.c_proj.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.14.attn.c_proj.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.14.ln_1.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.14.ln_1.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.14.ln_2.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.14.ln_2.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.14.mlp.c_fc.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.14.mlp.c_fc.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.14.mlp.c_proj.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.14.mlp.c_proj.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.15.attn.c_attn.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.15.attn.c_attn.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.15.attn.c_proj.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.15.attn.c_proj.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.15.ln_1.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.15.ln_1.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.15.ln_2.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.15.ln_2.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.15.mlp.c_fc.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.15.mlp.c_fc.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.15.mlp.c_proj.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.15.mlp.c_proj.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.16.attn.c_attn.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.16.attn.c_attn.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.16.attn.c_proj.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.16.attn.c_proj.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.16.ln_1.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.16.ln_1.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.16.ln_2.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.16.ln_2.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.16.mlp.c_fc.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.16.mlp.c_fc.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.16.mlp.c_proj.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.16.mlp.c_proj.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.17.attn.c_attn.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.17.attn.c_attn.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.17.attn.c_proj.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.17.attn.c_proj.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.17.ln_1.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.17.ln_1.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.17.ln_2.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.17.ln_2.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.17.mlp.c_fc.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.17.mlp.c_fc.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.17.mlp.c_proj.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.17.mlp.c_proj.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.18.attn.c_attn.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.18.attn.c_attn.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.18.attn.c_proj.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.18.attn.c_proj.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.18.ln_1.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.18.ln_1.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.18.ln_2.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.18.ln_2.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.18.mlp.c_fc.bias": "model-00003-of-00007.safetensors",
+        "transformer.h.18.mlp.c_fc.weight": "model-00003-of-00007.safetensors",
+        "transformer.h.18.mlp.c_proj.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.18.mlp.c_proj.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.19.attn.c_attn.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.19.attn.c_attn.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.19.attn.c_proj.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.19.attn.c_proj.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.19.ln_1.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.19.ln_1.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.19.ln_2.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.19.ln_2.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.19.mlp.c_fc.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.19.mlp.c_fc.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.19.mlp.c_proj.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.19.mlp.c_proj.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.2.attn.c_attn.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.2.attn.c_attn.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.2.attn.c_proj.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.2.attn.c_proj.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.2.ln_1.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.2.ln_1.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.2.ln_2.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.2.ln_2.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.2.mlp.c_fc.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.2.mlp.c_fc.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.2.mlp.c_proj.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.2.mlp.c_proj.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.20.attn.c_attn.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.20.attn.c_attn.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.20.attn.c_proj.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.20.attn.c_proj.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.20.ln_1.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.20.ln_1.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.20.ln_2.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.20.ln_2.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.20.mlp.c_fc.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.20.mlp.c_fc.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.20.mlp.c_proj.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.20.mlp.c_proj.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.21.attn.c_attn.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.21.attn.c_attn.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.21.attn.c_proj.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.21.attn.c_proj.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.21.ln_1.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.21.ln_1.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.21.ln_2.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.21.ln_2.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.21.mlp.c_fc.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.21.mlp.c_fc.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.21.mlp.c_proj.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.21.mlp.c_proj.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.22.attn.c_attn.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.22.attn.c_attn.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.22.attn.c_proj.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.22.attn.c_proj.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.22.ln_1.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.22.ln_1.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.22.ln_2.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.22.ln_2.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.22.mlp.c_fc.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.22.mlp.c_fc.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.22.mlp.c_proj.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.22.mlp.c_proj.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.23.attn.c_attn.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.23.attn.c_attn.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.23.attn.c_proj.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.23.attn.c_proj.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.23.ln_1.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.23.ln_1.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.23.ln_2.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.23.ln_2.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.23.mlp.c_fc.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.23.mlp.c_fc.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.23.mlp.c_proj.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.23.mlp.c_proj.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.24.attn.c_attn.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.24.attn.c_attn.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.24.attn.c_proj.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.24.attn.c_proj.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.24.ln_1.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.24.ln_1.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.24.ln_2.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.24.ln_2.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.24.mlp.c_fc.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.24.mlp.c_fc.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.24.mlp.c_proj.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.24.mlp.c_proj.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.25.attn.c_attn.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.25.attn.c_attn.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.25.attn.c_proj.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.25.attn.c_proj.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.25.ln_1.bias": "model-00004-of-00007.safetensors",
+        "transformer.h.25.ln_1.weight": "model-00004-of-00007.safetensors",
+        "transformer.h.25.ln_2.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.25.ln_2.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.25.mlp.c_fc.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.25.mlp.c_fc.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.25.mlp.c_proj.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.25.mlp.c_proj.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.26.attn.c_attn.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.26.attn.c_attn.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.26.attn.c_proj.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.26.attn.c_proj.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.26.ln_1.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.26.ln_1.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.26.ln_2.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.26.ln_2.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.26.mlp.c_fc.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.26.mlp.c_fc.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.26.mlp.c_proj.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.26.mlp.c_proj.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.27.attn.c_attn.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.27.attn.c_attn.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.27.attn.c_proj.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.27.attn.c_proj.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.27.ln_1.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.27.ln_1.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.27.ln_2.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.27.ln_2.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.27.mlp.c_fc.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.27.mlp.c_fc.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.27.mlp.c_proj.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.27.mlp.c_proj.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.28.attn.c_attn.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.28.attn.c_attn.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.28.attn.c_proj.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.28.attn.c_proj.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.28.ln_1.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.28.ln_1.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.28.ln_2.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.28.ln_2.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.28.mlp.c_fc.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.28.mlp.c_fc.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.28.mlp.c_proj.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.28.mlp.c_proj.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.29.attn.c_attn.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.29.attn.c_attn.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.29.attn.c_proj.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.29.attn.c_proj.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.29.ln_1.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.29.ln_1.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.29.ln_2.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.29.ln_2.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.29.mlp.c_fc.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.29.mlp.c_fc.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.29.mlp.c_proj.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.29.mlp.c_proj.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.3.attn.c_attn.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.3.attn.c_attn.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.3.attn.c_proj.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.3.attn.c_proj.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.3.ln_1.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.3.ln_1.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.3.ln_2.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.3.ln_2.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.3.mlp.c_fc.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.3.mlp.c_fc.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.3.mlp.c_proj.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.3.mlp.c_proj.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.30.attn.c_attn.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.30.attn.c_attn.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.30.attn.c_proj.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.30.attn.c_proj.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.30.ln_1.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.30.ln_1.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.30.ln_2.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.30.ln_2.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.30.mlp.c_fc.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.30.mlp.c_fc.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.30.mlp.c_proj.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.30.mlp.c_proj.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.31.attn.c_attn.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.31.attn.c_attn.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.31.attn.c_proj.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.31.attn.c_proj.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.31.ln_1.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.31.ln_1.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.31.ln_2.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.31.ln_2.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.31.mlp.c_fc.bias": "model-00005-of-00007.safetensors",
+        "transformer.h.31.mlp.c_fc.weight": "model-00005-of-00007.safetensors",
+        "transformer.h.31.mlp.c_proj.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.31.mlp.c_proj.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.32.attn.c_attn.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.32.attn.c_attn.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.32.attn.c_proj.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.32.attn.c_proj.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.32.ln_1.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.32.ln_1.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.32.ln_2.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.32.ln_2.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.32.mlp.c_fc.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.32.mlp.c_fc.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.32.mlp.c_proj.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.32.mlp.c_proj.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.33.attn.c_attn.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.33.attn.c_attn.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.33.attn.c_proj.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.33.attn.c_proj.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.33.ln_1.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.33.ln_1.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.33.ln_2.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.33.ln_2.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.33.mlp.c_fc.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.33.mlp.c_fc.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.33.mlp.c_proj.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.33.mlp.c_proj.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.34.attn.c_attn.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.34.attn.c_attn.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.34.attn.c_proj.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.34.attn.c_proj.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.34.ln_1.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.34.ln_1.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.34.ln_2.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.34.ln_2.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.34.mlp.c_fc.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.34.mlp.c_fc.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.34.mlp.c_proj.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.34.mlp.c_proj.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.35.attn.c_attn.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.35.attn.c_attn.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.35.attn.c_proj.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.35.attn.c_proj.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.35.ln_1.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.35.ln_1.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.35.ln_2.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.35.ln_2.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.35.mlp.c_fc.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.35.mlp.c_fc.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.35.mlp.c_proj.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.35.mlp.c_proj.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.36.attn.c_attn.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.36.attn.c_attn.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.36.attn.c_proj.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.36.attn.c_proj.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.36.ln_1.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.36.ln_1.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.36.ln_2.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.36.ln_2.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.36.mlp.c_fc.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.36.mlp.c_fc.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.36.mlp.c_proj.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.36.mlp.c_proj.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.37.attn.c_attn.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.37.attn.c_attn.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.37.attn.c_proj.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.37.attn.c_proj.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.37.ln_1.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.37.ln_1.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.37.ln_2.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.37.ln_2.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.37.mlp.c_fc.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.37.mlp.c_fc.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.37.mlp.c_proj.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.37.mlp.c_proj.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.38.attn.c_attn.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.38.attn.c_attn.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.38.attn.c_proj.bias": "model-00007-of-00007.safetensors",
+        "transformer.h.38.attn.c_proj.weight": "model-00007-of-00007.safetensors",
+        "transformer.h.38.ln_1.bias": "model-00006-of-00007.safetensors",
+        "transformer.h.38.ln_1.weight": "model-00006-of-00007.safetensors",
+        "transformer.h.38.ln_2.bias": "model-00007-of-00007.safetensors",
+        "transformer.h.38.ln_2.weight": "model-00007-of-00007.safetensors",
+        "transformer.h.38.mlp.c_fc.bias": "model-00007-of-00007.safetensors",
+        "transformer.h.38.mlp.c_fc.weight": "model-00007-of-00007.safetensors",
+        "transformer.h.38.mlp.c_proj.bias": "model-00007-of-00007.safetensors",
+        "transformer.h.38.mlp.c_proj.weight": "model-00007-of-00007.safetensors",
+        "transformer.h.39.attn.c_attn.bias": "model-00007-of-00007.safetensors",
+        "transformer.h.39.attn.c_attn.weight": "model-00007-of-00007.safetensors",
+        "transformer.h.39.attn.c_proj.bias": "model-00007-of-00007.safetensors",
+        "transformer.h.39.attn.c_proj.weight": "model-00007-of-00007.safetensors",
+        "transformer.h.39.ln_1.bias": "model-00007-of-00007.safetensors",
+        "transformer.h.39.ln_1.weight": "model-00007-of-00007.safetensors",
+        "transformer.h.39.ln_2.bias": "model-00007-of-00007.safetensors",
+        "transformer.h.39.ln_2.weight": "model-00007-of-00007.safetensors",
+        "transformer.h.39.mlp.c_fc.bias": "model-00007-of-00007.safetensors",
+        "transformer.h.39.mlp.c_fc.weight": "model-00007-of-00007.safetensors",
+        "transformer.h.39.mlp.c_proj.bias": "model-00007-of-00007.safetensors",
+        "transformer.h.39.mlp.c_proj.weight": "model-00007-of-00007.safetensors",
+        "transformer.h.4.attn.c_attn.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.4.attn.c_attn.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.4.attn.c_proj.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.4.attn.c_proj.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.4.ln_1.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.4.ln_1.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.4.ln_2.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.4.ln_2.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.4.mlp.c_fc.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.4.mlp.c_fc.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.4.mlp.c_proj.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.4.mlp.c_proj.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.5.attn.c_attn.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.5.attn.c_attn.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.5.attn.c_proj.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.5.attn.c_proj.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.5.ln_1.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.5.ln_1.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.5.ln_2.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.5.ln_2.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.5.mlp.c_fc.bias": "model-00001-of-00007.safetensors",
+        "transformer.h.5.mlp.c_fc.weight": "model-00001-of-00007.safetensors",
+        "transformer.h.5.mlp.c_proj.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.5.mlp.c_proj.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.6.attn.c_attn.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.6.attn.c_attn.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.6.attn.c_proj.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.6.attn.c_proj.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.6.ln_1.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.6.ln_1.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.6.ln_2.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.6.ln_2.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.6.mlp.c_fc.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.6.mlp.c_fc.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.6.mlp.c_proj.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.6.mlp.c_proj.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.7.attn.c_attn.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.7.attn.c_attn.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.7.attn.c_proj.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.7.attn.c_proj.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.7.ln_1.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.7.ln_1.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.7.ln_2.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.7.ln_2.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.7.mlp.c_fc.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.7.mlp.c_fc.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.7.mlp.c_proj.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.7.mlp.c_proj.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.8.attn.c_attn.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.8.attn.c_attn.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.8.attn.c_proj.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.8.attn.c_proj.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.8.ln_1.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.8.ln_1.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.8.ln_2.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.8.ln_2.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.8.mlp.c_fc.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.8.mlp.c_fc.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.8.mlp.c_proj.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.8.mlp.c_proj.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.9.attn.c_attn.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.9.attn.c_attn.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.9.attn.c_proj.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.9.attn.c_proj.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.9.ln_1.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.9.ln_1.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.9.ln_2.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.9.ln_2.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.9.mlp.c_fc.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.9.mlp.c_fc.weight": "model-00002-of-00007.safetensors",
+        "transformer.h.9.mlp.c_proj.bias": "model-00002-of-00007.safetensors",
+        "transformer.h.9.mlp.c_proj.weight": "model-00002-of-00007.safetensors",
+        "transformer.ln_f.bias": "model-00007-of-00007.safetensors",
+        "transformer.ln_f.weight": "model-00007-of-00007.safetensors",
+        "transformer.wpe.weight": "model-00001-of-00007.safetensors",
+        "transformer.wte.weight": "model-00001-of-00007.safetensors"
+    }
+}