Adding `safetensors` variant of this model

model-00001-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:deaeed8feb8abf1f945c6b619ab8848ceed2f7b9a5712e0279e6b581859c5736
+size 4194304136

model-00002-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0dec38f81760e66077269fc35308a5492006ef6b452b76da0feaef5fd4fe75fd
+size 839174176

model-00003-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:16f46b7a253d20b3221384af26ec30bc6fc392ef75a9bda6c9953d61cb40912a
+size 805586672

model-00004-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c773d58b438febe85615c28f8160b28dc5fae34f685495e637f1aa888de81cc8
+size 805586672

model-00005-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28c3016d9b152c7c48a552e3e1c223b53eb0eeb4b499a1255f53bc4555f92194
+size 805586672

model-00006-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73ed2c08fb86373b1e7b57d487ee86fcb4c4018b0bb78540ed88bcec2a6e9eef
+size 805586672

model-00007-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:44b2e8960a44eaf5875608930538c0a0fd65558d23e2ef71878a1ef3d95ca2a8
+size 805586672

model-00008-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b579c79305c2b130fb05c1c47251d35a76f2b7b2bf2eafc2931c5ece6e77df62
+size 805586672

model-00009-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52136ee22dba581a88262c46a28b72a7594be8f774881b77e7d405a6e94fe6c5
+size 805586672

model-00010-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d95eea2c56fb0c2a92ea077ae211f38a838952091e07d7dd5bc88204338bc3aa
+size 805586672

model-00011-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9b87376e8b34a13675b461cc7547b69162cadec44a7d8cef1fab912b58fd05a
+size 805586656

model-00012-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e15a43d0723a36cf9f686b9de4fff23b5e63fca1e95ebf81c13c210dd01e1a8
+size 805586688

model-00013-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90513230e71483120ff83b9d2d6442a8ebd074e69735ce7e808c229c2a7f15ae
+size 805586688

model-00014-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a36d102ae7cd0cb3a4b2327fe9cf9ee777bce0b31885b60b8320de88a7b8ab7
+size 805586688

model-00015-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f05592d827dbe314284e98854bb31fe7cddd0bd5d2e5707c63fdb34fcaac8a0
+size 805586688

model-00016-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:701f349e4a00f25706923b084f01b57b4dd7b4383ea6d487c24d40944c9df125
+size 805586688

model-00017-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5adccfc91909ea31ba907c9e39400e90c2b85261b27658654ae095b602be415f
+size 805586688

model-00018-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3742ea07fb89a706ea262396d34ce5cfa54340c4514d1b46bd640ed0d11302e1
+size 805586688

model-00019-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64c9a082d21e2b69f0f91229461461094a118376f61ebcdd8a42b6718e2e132f
+size 805586688

model-00020-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81e3c49ee523941def5b8a46fc300b70ec1ee50e990297ce3e12ea6f1d319616
+size 805586688

model-00021-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5166f5db16617f594396556a6d3deacbaff0f0a8ebb0bc98421f126f5cffe049
+size 805586688

model-00022-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab480dd76d6e1bb9352222bf3db81f6f861db43c0b8ae7be453ec9fa3b15dbeb
+size 805586688

model-00023-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e14785a3e9adf86c1188d5ce787f976f7330f386b5907554301fad14a7bb910e
+size 805586688

model-00024-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f250d57280ffa259a3490a5b841671e661e535d7cfe064769c858defd519feee
+size 805586688

model-00025-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5964c2d9cb582ce399fe9550ac59a2bb49b740034635120a62f581f5d41b007c
+size 805586688

model-00026-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9155da26219a03b0d9fbbdf9903cc6fff03560c12c1e313f565ee37162ee55b
+size 805586688

model-00027-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50aa47237d51362f6d02cd4912ee6f5ce74d134f13ff05fe92f6f2e25398b823
+size 805586688

model-00028-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fce2af0363e260e9f4e6d00d77fc22e39b13eab969f428324d71353fe51c8b23
+size 805586688

model-00029-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c87bce9caad5b8011d3be9788c4a7ccd028e8fb7ccfdd97580a055bbd45ec0b7
+size 805586688

model-00030-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd51372d868254f0c2dab7838316b83c7d403f8dc291bf5b4b2ddbc2532396e4
+size 805586688

model-00031-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:11a2179e9d3698c08138c9aae035da5174d27c62bbfbc0e22e924b93089b14e1
+size 805586688

model-00032-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99805412752421d93e0e2ce5fe9ef8d9bf4ad908049070b1d5dcf19fb84d367e
+size 805586688

model-00033-of-00033.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:914566b751363d81e1e441a4e23867d3bac9f6b74bd9a3eb246a9e72493d3dbd
+size 805586664

model.safetensors.index.json

@@ -0,0 +1,523 @@
+{
+    "metadata": {
+        "total_size": 30006607872
+    },
+    "weight_map": {
+        "transformer.blocks.0.attn.Wqkv.bias": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.attn.Wqkv.weight": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.attn.k_ln.bias": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.attn.k_ln.weight": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.attn.out_proj.bias": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.attn.out_proj.weight": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.attn.q_ln.bias": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.attn.q_ln.weight": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.ffn.down_proj.bias": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.ffn.down_proj.weight": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.ffn.up_proj.bias": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.ffn.up_proj.weight": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.norm_1.bias": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.norm_1.weight": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.norm_2.bias": "model-00002-of-00033.safetensors",
+        "transformer.blocks.0.norm_2.weight": "model-00002-of-00033.safetensors",
+        "transformer.blocks.1.attn.Wqkv.bias": "model-00003-of-00033.safetensors",
+        "transformer.blocks.1.attn.Wqkv.weight": "model-00003-of-00033.safetensors",
+        "transformer.blocks.1.attn.k_ln.bias": "model-00003-of-00033.safetensors",
+        "transformer.blocks.1.attn.k_ln.weight": "model-00003-of-00033.safetensors",
+        "transformer.blocks.1.attn.out_proj.bias": "model-00003-of-00033.safetensors",
+        "transformer.blocks.1.attn.out_proj.weight": "model-00003-of-00033.safetensors",
+        "transformer.blocks.1.attn.q_ln.bias": "model-00003-of-00033.safetensors",
+        "transformer.blocks.1.attn.q_ln.weight": "model-00003-of-00033.safetensors",
+        "transformer.blocks.1.ffn.down_proj.bias": "model-00003-of-00033.safetensors",
+        "transformer.blocks.1.ffn.down_proj.weight": "model-00003-of-00033.safetensors",
+        "transformer.blocks.1.ffn.up_proj.bias": "model-00003-of-00033.safetensors",
+        "transformer.blocks.1.ffn.up_proj.weight": "model-00003-of-00033.safetensors",
+        "transformer.blocks.1.norm_1.bias": "model-00002-of-00033.safetensors",
+        "transformer.blocks.1.norm_1.weight": "model-00002-of-00033.safetensors",
+        "transformer.blocks.1.norm_2.bias": "model-00003-of-00033.safetensors",
+        "transformer.blocks.1.norm_2.weight": "model-00003-of-00033.safetensors",
+        "transformer.blocks.10.attn.Wqkv.bias": "model-00012-of-00033.safetensors",
+        "transformer.blocks.10.attn.Wqkv.weight": "model-00012-of-00033.safetensors",
+        "transformer.blocks.10.attn.k_ln.bias": "model-00012-of-00033.safetensors",
+        "transformer.blocks.10.attn.k_ln.weight": "model-00012-of-00033.safetensors",
+        "transformer.blocks.10.attn.out_proj.bias": "model-00012-of-00033.safetensors",
+        "transformer.blocks.10.attn.out_proj.weight": "model-00012-of-00033.safetensors",
+        "transformer.blocks.10.attn.q_ln.bias": "model-00012-of-00033.safetensors",
+        "transformer.blocks.10.attn.q_ln.weight": "model-00012-of-00033.safetensors",
+        "transformer.blocks.10.ffn.down_proj.bias": "model-00012-of-00033.safetensors",
+        "transformer.blocks.10.ffn.down_proj.weight": "model-00012-of-00033.safetensors",
+        "transformer.blocks.10.ffn.up_proj.bias": "model-00012-of-00033.safetensors",
+        "transformer.blocks.10.ffn.up_proj.weight": "model-00012-of-00033.safetensors",
+        "transformer.blocks.10.norm_1.bias": "model-00011-of-00033.safetensors",
+        "transformer.blocks.10.norm_1.weight": "model-00011-of-00033.safetensors",
+        "transformer.blocks.10.norm_2.bias": "model-00012-of-00033.safetensors",
+        "transformer.blocks.10.norm_2.weight": "model-00012-of-00033.safetensors",
+        "transformer.blocks.11.attn.Wqkv.bias": "model-00013-of-00033.safetensors",
+        "transformer.blocks.11.attn.Wqkv.weight": "model-00013-of-00033.safetensors",
+        "transformer.blocks.11.attn.k_ln.bias": "model-00013-of-00033.safetensors",
+        "transformer.blocks.11.attn.k_ln.weight": "model-00013-of-00033.safetensors",
+        "transformer.blocks.11.attn.out_proj.bias": "model-00013-of-00033.safetensors",
+        "transformer.blocks.11.attn.out_proj.weight": "model-00013-of-00033.safetensors",
+        "transformer.blocks.11.attn.q_ln.bias": "model-00013-of-00033.safetensors",
+        "transformer.blocks.11.attn.q_ln.weight": "model-00013-of-00033.safetensors",
+        "transformer.blocks.11.ffn.down_proj.bias": "model-00013-of-00033.safetensors",
+        "transformer.blocks.11.ffn.down_proj.weight": "model-00013-of-00033.safetensors",
+        "transformer.blocks.11.ffn.up_proj.bias": "model-00013-of-00033.safetensors",
+        "transformer.blocks.11.ffn.up_proj.weight": "model-00013-of-00033.safetensors",
+        "transformer.blocks.11.norm_1.bias": "model-00012-of-00033.safetensors",
+        "transformer.blocks.11.norm_1.weight": "model-00012-of-00033.safetensors",
+        "transformer.blocks.11.norm_2.bias": "model-00013-of-00033.safetensors",
+        "transformer.blocks.11.norm_2.weight": "model-00013-of-00033.safetensors",
+        "transformer.blocks.12.attn.Wqkv.bias": "model-00014-of-00033.safetensors",
+        "transformer.blocks.12.attn.Wqkv.weight": "model-00014-of-00033.safetensors",
+        "transformer.blocks.12.attn.k_ln.bias": "model-00014-of-00033.safetensors",
+        "transformer.blocks.12.attn.k_ln.weight": "model-00014-of-00033.safetensors",
+        "transformer.blocks.12.attn.out_proj.bias": "model-00014-of-00033.safetensors",
+        "transformer.blocks.12.attn.out_proj.weight": "model-00014-of-00033.safetensors",
+        "transformer.blocks.12.attn.q_ln.bias": "model-00014-of-00033.safetensors",
+        "transformer.blocks.12.attn.q_ln.weight": "model-00014-of-00033.safetensors",
+        "transformer.blocks.12.ffn.down_proj.bias": "model-00014-of-00033.safetensors",
+        "transformer.blocks.12.ffn.down_proj.weight": "model-00014-of-00033.safetensors",
+        "transformer.blocks.12.ffn.up_proj.bias": "model-00014-of-00033.safetensors",
+        "transformer.blocks.12.ffn.up_proj.weight": "model-00014-of-00033.safetensors",
+        "transformer.blocks.12.norm_1.bias": "model-00013-of-00033.safetensors",
+        "transformer.blocks.12.norm_1.weight": "model-00013-of-00033.safetensors",
+        "transformer.blocks.12.norm_2.bias": "model-00014-of-00033.safetensors",
+        "transformer.blocks.12.norm_2.weight": "model-00014-of-00033.safetensors",
+        "transformer.blocks.13.attn.Wqkv.bias": "model-00015-of-00033.safetensors",
+        "transformer.blocks.13.attn.Wqkv.weight": "model-00015-of-00033.safetensors",
+        "transformer.blocks.13.attn.k_ln.bias": "model-00015-of-00033.safetensors",
+        "transformer.blocks.13.attn.k_ln.weight": "model-00015-of-00033.safetensors",
+        "transformer.blocks.13.attn.out_proj.bias": "model-00015-of-00033.safetensors",
+        "transformer.blocks.13.attn.out_proj.weight": "model-00015-of-00033.safetensors",
+        "transformer.blocks.13.attn.q_ln.bias": "model-00015-of-00033.safetensors",
+        "transformer.blocks.13.attn.q_ln.weight": "model-00015-of-00033.safetensors",
+        "transformer.blocks.13.ffn.down_proj.bias": "model-00015-of-00033.safetensors",
+        "transformer.blocks.13.ffn.down_proj.weight": "model-00015-of-00033.safetensors",
+        "transformer.blocks.13.ffn.up_proj.bias": "model-00015-of-00033.safetensors",
+        "transformer.blocks.13.ffn.up_proj.weight": "model-00015-of-00033.safetensors",
+        "transformer.blocks.13.norm_1.bias": "model-00014-of-00033.safetensors",
+        "transformer.blocks.13.norm_1.weight": "model-00014-of-00033.safetensors",
+        "transformer.blocks.13.norm_2.bias": "model-00015-of-00033.safetensors",
+        "transformer.blocks.13.norm_2.weight": "model-00015-of-00033.safetensors",
+        "transformer.blocks.14.attn.Wqkv.bias": "model-00016-of-00033.safetensors",
+        "transformer.blocks.14.attn.Wqkv.weight": "model-00016-of-00033.safetensors",
+        "transformer.blocks.14.attn.k_ln.bias": "model-00016-of-00033.safetensors",
+        "transformer.blocks.14.attn.k_ln.weight": "model-00016-of-00033.safetensors",
+        "transformer.blocks.14.attn.out_proj.bias": "model-00016-of-00033.safetensors",
+        "transformer.blocks.14.attn.out_proj.weight": "model-00016-of-00033.safetensors",
+        "transformer.blocks.14.attn.q_ln.bias": "model-00016-of-00033.safetensors",
+        "transformer.blocks.14.attn.q_ln.weight": "model-00016-of-00033.safetensors",
+        "transformer.blocks.14.ffn.down_proj.bias": "model-00016-of-00033.safetensors",
+        "transformer.blocks.14.ffn.down_proj.weight": "model-00016-of-00033.safetensors",
+        "transformer.blocks.14.ffn.up_proj.bias": "model-00016-of-00033.safetensors",
+        "transformer.blocks.14.ffn.up_proj.weight": "model-00016-of-00033.safetensors",
+        "transformer.blocks.14.norm_1.bias": "model-00015-of-00033.safetensors",
+        "transformer.blocks.14.norm_1.weight": "model-00015-of-00033.safetensors",
+        "transformer.blocks.14.norm_2.bias": "model-00016-of-00033.safetensors",
+        "transformer.blocks.14.norm_2.weight": "model-00016-of-00033.safetensors",
+        "transformer.blocks.15.attn.Wqkv.bias": "model-00017-of-00033.safetensors",
+        "transformer.blocks.15.attn.Wqkv.weight": "model-00017-of-00033.safetensors",
+        "transformer.blocks.15.attn.k_ln.bias": "model-00017-of-00033.safetensors",
+        "transformer.blocks.15.attn.k_ln.weight": "model-00017-of-00033.safetensors",
+        "transformer.blocks.15.attn.out_proj.bias": "model-00017-of-00033.safetensors",
+        "transformer.blocks.15.attn.out_proj.weight": "model-00017-of-00033.safetensors",
+        "transformer.blocks.15.attn.q_ln.bias": "model-00017-of-00033.safetensors",
+        "transformer.blocks.15.attn.q_ln.weight": "model-00017-of-00033.safetensors",
+        "transformer.blocks.15.ffn.down_proj.bias": "model-00017-of-00033.safetensors",
+        "transformer.blocks.15.ffn.down_proj.weight": "model-00017-of-00033.safetensors",
+        "transformer.blocks.15.ffn.up_proj.bias": "model-00017-of-00033.safetensors",
+        "transformer.blocks.15.ffn.up_proj.weight": "model-00017-of-00033.safetensors",
+        "transformer.blocks.15.norm_1.bias": "model-00016-of-00033.safetensors",
+        "transformer.blocks.15.norm_1.weight": "model-00016-of-00033.safetensors",
+        "transformer.blocks.15.norm_2.bias": "model-00017-of-00033.safetensors",
+        "transformer.blocks.15.norm_2.weight": "model-00017-of-00033.safetensors",
+        "transformer.blocks.16.attn.Wqkv.bias": "model-00018-of-00033.safetensors",
+        "transformer.blocks.16.attn.Wqkv.weight": "model-00018-of-00033.safetensors",
+        "transformer.blocks.16.attn.k_ln.bias": "model-00018-of-00033.safetensors",
+        "transformer.blocks.16.attn.k_ln.weight": "model-00018-of-00033.safetensors",
+        "transformer.blocks.16.attn.out_proj.bias": "model-00018-of-00033.safetensors",
+        "transformer.blocks.16.attn.out_proj.weight": "model-00018-of-00033.safetensors",
+        "transformer.blocks.16.attn.q_ln.bias": "model-00018-of-00033.safetensors",
+        "transformer.blocks.16.attn.q_ln.weight": "model-00018-of-00033.safetensors",
+        "transformer.blocks.16.ffn.down_proj.bias": "model-00018-of-00033.safetensors",
+        "transformer.blocks.16.ffn.down_proj.weight": "model-00018-of-00033.safetensors",
+        "transformer.blocks.16.ffn.up_proj.bias": "model-00018-of-00033.safetensors",
+        "transformer.blocks.16.ffn.up_proj.weight": "model-00018-of-00033.safetensors",
+        "transformer.blocks.16.norm_1.bias": "model-00017-of-00033.safetensors",
+        "transformer.blocks.16.norm_1.weight": "model-00017-of-00033.safetensors",
+        "transformer.blocks.16.norm_2.bias": "model-00018-of-00033.safetensors",
+        "transformer.blocks.16.norm_2.weight": "model-00018-of-00033.safetensors",
+        "transformer.blocks.17.attn.Wqkv.bias": "model-00019-of-00033.safetensors",
+        "transformer.blocks.17.attn.Wqkv.weight": "model-00019-of-00033.safetensors",
+        "transformer.blocks.17.attn.k_ln.bias": "model-00019-of-00033.safetensors",
+        "transformer.blocks.17.attn.k_ln.weight": "model-00019-of-00033.safetensors",
+        "transformer.blocks.17.attn.out_proj.bias": "model-00019-of-00033.safetensors",
+        "transformer.blocks.17.attn.out_proj.weight": "model-00019-of-00033.safetensors",
+        "transformer.blocks.17.attn.q_ln.bias": "model-00019-of-00033.safetensors",
+        "transformer.blocks.17.attn.q_ln.weight": "model-00019-of-00033.safetensors",
+        "transformer.blocks.17.ffn.down_proj.bias": "model-00019-of-00033.safetensors",
+        "transformer.blocks.17.ffn.down_proj.weight": "model-00019-of-00033.safetensors",
+        "transformer.blocks.17.ffn.up_proj.bias": "model-00019-of-00033.safetensors",
+        "transformer.blocks.17.ffn.up_proj.weight": "model-00019-of-00033.safetensors",
+        "transformer.blocks.17.norm_1.bias": "model-00018-of-00033.safetensors",
+        "transformer.blocks.17.norm_1.weight": "model-00018-of-00033.safetensors",
+        "transformer.blocks.17.norm_2.bias": "model-00019-of-00033.safetensors",
+        "transformer.blocks.17.norm_2.weight": "model-00019-of-00033.safetensors",
+        "transformer.blocks.18.attn.Wqkv.bias": "model-00020-of-00033.safetensors",
+        "transformer.blocks.18.attn.Wqkv.weight": "model-00020-of-00033.safetensors",
+        "transformer.blocks.18.attn.k_ln.bias": "model-00020-of-00033.safetensors",
+        "transformer.blocks.18.attn.k_ln.weight": "model-00020-of-00033.safetensors",
+        "transformer.blocks.18.attn.out_proj.bias": "model-00020-of-00033.safetensors",
+        "transformer.blocks.18.attn.out_proj.weight": "model-00020-of-00033.safetensors",
+        "transformer.blocks.18.attn.q_ln.bias": "model-00020-of-00033.safetensors",
+        "transformer.blocks.18.attn.q_ln.weight": "model-00020-of-00033.safetensors",
+        "transformer.blocks.18.ffn.down_proj.bias": "model-00020-of-00033.safetensors",
+        "transformer.blocks.18.ffn.down_proj.weight": "model-00020-of-00033.safetensors",
+        "transformer.blocks.18.ffn.up_proj.bias": "model-00020-of-00033.safetensors",
+        "transformer.blocks.18.ffn.up_proj.weight": "model-00020-of-00033.safetensors",
+        "transformer.blocks.18.norm_1.bias": "model-00019-of-00033.safetensors",
+        "transformer.blocks.18.norm_1.weight": "model-00019-of-00033.safetensors",
+        "transformer.blocks.18.norm_2.bias": "model-00020-of-00033.safetensors",
+        "transformer.blocks.18.norm_2.weight": "model-00020-of-00033.safetensors",
+        "transformer.blocks.19.attn.Wqkv.bias": "model-00021-of-00033.safetensors",
+        "transformer.blocks.19.attn.Wqkv.weight": "model-00021-of-00033.safetensors",
+        "transformer.blocks.19.attn.k_ln.bias": "model-00021-of-00033.safetensors",
+        "transformer.blocks.19.attn.k_ln.weight": "model-00021-of-00033.safetensors",
+        "transformer.blocks.19.attn.out_proj.bias": "model-00021-of-00033.safetensors",
+        "transformer.blocks.19.attn.out_proj.weight": "model-00021-of-00033.safetensors",
+        "transformer.blocks.19.attn.q_ln.bias": "model-00021-of-00033.safetensors",
+        "transformer.blocks.19.attn.q_ln.weight": "model-00021-of-00033.safetensors",
+        "transformer.blocks.19.ffn.down_proj.bias": "model-00021-of-00033.safetensors",
+        "transformer.blocks.19.ffn.down_proj.weight": "model-00021-of-00033.safetensors",
+        "transformer.blocks.19.ffn.up_proj.bias": "model-00021-of-00033.safetensors",
+        "transformer.blocks.19.ffn.up_proj.weight": "model-00021-of-00033.safetensors",
+        "transformer.blocks.19.norm_1.bias": "model-00020-of-00033.safetensors",
+        "transformer.blocks.19.norm_1.weight": "model-00020-of-00033.safetensors",
+        "transformer.blocks.19.norm_2.bias": "model-00021-of-00033.safetensors",
+        "transformer.blocks.19.norm_2.weight": "model-00021-of-00033.safetensors",
+        "transformer.blocks.2.attn.Wqkv.bias": "model-00004-of-00033.safetensors",
+        "transformer.blocks.2.attn.Wqkv.weight": "model-00004-of-00033.safetensors",
+        "transformer.blocks.2.attn.k_ln.bias": "model-00004-of-00033.safetensors",
+        "transformer.blocks.2.attn.k_ln.weight": "model-00004-of-00033.safetensors",
+        "transformer.blocks.2.attn.out_proj.bias": "model-00004-of-00033.safetensors",
+        "transformer.blocks.2.attn.out_proj.weight": "model-00004-of-00033.safetensors",
+        "transformer.blocks.2.attn.q_ln.bias": "model-00004-of-00033.safetensors",
+        "transformer.blocks.2.attn.q_ln.weight": "model-00004-of-00033.safetensors",
+        "transformer.blocks.2.ffn.down_proj.bias": "model-00004-of-00033.safetensors",
+        "transformer.blocks.2.ffn.down_proj.weight": "model-00004-of-00033.safetensors",
+        "transformer.blocks.2.ffn.up_proj.bias": "model-00004-of-00033.safetensors",
+        "transformer.blocks.2.ffn.up_proj.weight": "model-00004-of-00033.safetensors",
+        "transformer.blocks.2.norm_1.bias": "model-00003-of-00033.safetensors",
+        "transformer.blocks.2.norm_1.weight": "model-00003-of-00033.safetensors",
+        "transformer.blocks.2.norm_2.bias": "model-00004-of-00033.safetensors",
+        "transformer.blocks.2.norm_2.weight": "model-00004-of-00033.safetensors",
+        "transformer.blocks.20.attn.Wqkv.bias": "model-00022-of-00033.safetensors",
+        "transformer.blocks.20.attn.Wqkv.weight": "model-00022-of-00033.safetensors",
+        "transformer.blocks.20.attn.k_ln.bias": "model-00022-of-00033.safetensors",
+        "transformer.blocks.20.attn.k_ln.weight": "model-00022-of-00033.safetensors",
+        "transformer.blocks.20.attn.out_proj.bias": "model-00022-of-00033.safetensors",
+        "transformer.blocks.20.attn.out_proj.weight": "model-00022-of-00033.safetensors",
+        "transformer.blocks.20.attn.q_ln.bias": "model-00022-of-00033.safetensors",
+        "transformer.blocks.20.attn.q_ln.weight": "model-00022-of-00033.safetensors",
+        "transformer.blocks.20.ffn.down_proj.bias": "model-00022-of-00033.safetensors",
+        "transformer.blocks.20.ffn.down_proj.weight": "model-00022-of-00033.safetensors",
+        "transformer.blocks.20.ffn.up_proj.bias": "model-00022-of-00033.safetensors",
+        "transformer.blocks.20.ffn.up_proj.weight": "model-00022-of-00033.safetensors",
+        "transformer.blocks.20.norm_1.bias": "model-00021-of-00033.safetensors",
+        "transformer.blocks.20.norm_1.weight": "model-00021-of-00033.safetensors",
+        "transformer.blocks.20.norm_2.bias": "model-00022-of-00033.safetensors",
+        "transformer.blocks.20.norm_2.weight": "model-00022-of-00033.safetensors",
+        "transformer.blocks.21.attn.Wqkv.bias": "model-00023-of-00033.safetensors",
+        "transformer.blocks.21.attn.Wqkv.weight": "model-00023-of-00033.safetensors",
+        "transformer.blocks.21.attn.k_ln.bias": "model-00023-of-00033.safetensors",
+        "transformer.blocks.21.attn.k_ln.weight": "model-00023-of-00033.safetensors",
+        "transformer.blocks.21.attn.out_proj.bias": "model-00023-of-00033.safetensors",
+        "transformer.blocks.21.attn.out_proj.weight": "model-00023-of-00033.safetensors",
+        "transformer.blocks.21.attn.q_ln.bias": "model-00023-of-00033.safetensors",
+        "transformer.blocks.21.attn.q_ln.weight": "model-00023-of-00033.safetensors",
+        "transformer.blocks.21.ffn.down_proj.bias": "model-00023-of-00033.safetensors",
+        "transformer.blocks.21.ffn.down_proj.weight": "model-00023-of-00033.safetensors",
+        "transformer.blocks.21.ffn.up_proj.bias": "model-00023-of-00033.safetensors",
+        "transformer.blocks.21.ffn.up_proj.weight": "model-00023-of-00033.safetensors",
+        "transformer.blocks.21.norm_1.bias": "model-00022-of-00033.safetensors",
+        "transformer.blocks.21.norm_1.weight": "model-00022-of-00033.safetensors",
+        "transformer.blocks.21.norm_2.bias": "model-00023-of-00033.safetensors",
+        "transformer.blocks.21.norm_2.weight": "model-00023-of-00033.safetensors",
+        "transformer.blocks.22.attn.Wqkv.bias": "model-00024-of-00033.safetensors",
+        "transformer.blocks.22.attn.Wqkv.weight": "model-00024-of-00033.safetensors",
+        "transformer.blocks.22.attn.k_ln.bias": "model-00024-of-00033.safetensors",
+        "transformer.blocks.22.attn.k_ln.weight": "model-00024-of-00033.safetensors",
+        "transformer.blocks.22.attn.out_proj.bias": "model-00024-of-00033.safetensors",
+        "transformer.blocks.22.attn.out_proj.weight": "model-00024-of-00033.safetensors",
+        "transformer.blocks.22.attn.q_ln.bias": "model-00024-of-00033.safetensors",
+        "transformer.blocks.22.attn.q_ln.weight": "model-00024-of-00033.safetensors",
+        "transformer.blocks.22.ffn.down_proj.bias": "model-00024-of-00033.safetensors",
+        "transformer.blocks.22.ffn.down_proj.weight": "model-00024-of-00033.safetensors",
+        "transformer.blocks.22.ffn.up_proj.bias": "model-00024-of-00033.safetensors",
+        "transformer.blocks.22.ffn.up_proj.weight": "model-00024-of-00033.safetensors",
+        "transformer.blocks.22.norm_1.bias": "model-00023-of-00033.safetensors",
+        "transformer.blocks.22.norm_1.weight": "model-00023-of-00033.safetensors",
+        "transformer.blocks.22.norm_2.bias": "model-00024-of-00033.safetensors",
+        "transformer.blocks.22.norm_2.weight": "model-00024-of-00033.safetensors",
+        "transformer.blocks.23.attn.Wqkv.bias": "model-00025-of-00033.safetensors",
+        "transformer.blocks.23.attn.Wqkv.weight": "model-00025-of-00033.safetensors",
+        "transformer.blocks.23.attn.k_ln.bias": "model-00025-of-00033.safetensors",
+        "transformer.blocks.23.attn.k_ln.weight": "model-00025-of-00033.safetensors",
+        "transformer.blocks.23.attn.out_proj.bias": "model-00025-of-00033.safetensors",
+        "transformer.blocks.23.attn.out_proj.weight": "model-00025-of-00033.safetensors",
+        "transformer.blocks.23.attn.q_ln.bias": "model-00025-of-00033.safetensors",
+        "transformer.blocks.23.attn.q_ln.weight": "model-00025-of-00033.safetensors",
+        "transformer.blocks.23.ffn.down_proj.bias": "model-00025-of-00033.safetensors",
+        "transformer.blocks.23.ffn.down_proj.weight": "model-00025-of-00033.safetensors",
+        "transformer.blocks.23.ffn.up_proj.bias": "model-00025-of-00033.safetensors",
+        "transformer.blocks.23.ffn.up_proj.weight": "model-00025-of-00033.safetensors",
+        "transformer.blocks.23.norm_1.bias": "model-00024-of-00033.safetensors",
+        "transformer.blocks.23.norm_1.weight": "model-00024-of-00033.safetensors",
+        "transformer.blocks.23.norm_2.bias": "model-00025-of-00033.safetensors",
+        "transformer.blocks.23.norm_2.weight": "model-00025-of-00033.safetensors",
+        "transformer.blocks.24.attn.Wqkv.bias": "model-00026-of-00033.safetensors",
+        "transformer.blocks.24.attn.Wqkv.weight": "model-00026-of-00033.safetensors",
+        "transformer.blocks.24.attn.k_ln.bias": "model-00026-of-00033.safetensors",
+        "transformer.blocks.24.attn.k_ln.weight": "model-00026-of-00033.safetensors",
+        "transformer.blocks.24.attn.out_proj.bias": "model-00026-of-00033.safetensors",
+        "transformer.blocks.24.attn.out_proj.weight": "model-00026-of-00033.safetensors",
+        "transformer.blocks.24.attn.q_ln.bias": "model-00026-of-00033.safetensors",
+        "transformer.blocks.24.attn.q_ln.weight": "model-00026-of-00033.safetensors",
+        "transformer.blocks.24.ffn.down_proj.bias": "model-00026-of-00033.safetensors",
+        "transformer.blocks.24.ffn.down_proj.weight": "model-00026-of-00033.safetensors",
+        "transformer.blocks.24.ffn.up_proj.bias": "model-00026-of-00033.safetensors",
+        "transformer.blocks.24.ffn.up_proj.weight": "model-00026-of-00033.safetensors",
+        "transformer.blocks.24.norm_1.bias": "model-00025-of-00033.safetensors",
+        "transformer.blocks.24.norm_1.weight": "model-00025-of-00033.safetensors",
+        "transformer.blocks.24.norm_2.bias": "model-00026-of-00033.safetensors",
+        "transformer.blocks.24.norm_2.weight": "model-00026-of-00033.safetensors",
+        "transformer.blocks.25.attn.Wqkv.bias": "model-00027-of-00033.safetensors",
+        "transformer.blocks.25.attn.Wqkv.weight": "model-00027-of-00033.safetensors",
+        "transformer.blocks.25.attn.k_ln.bias": "model-00027-of-00033.safetensors",
+        "transformer.blocks.25.attn.k_ln.weight": "model-00027-of-00033.safetensors",
+        "transformer.blocks.25.attn.out_proj.bias": "model-00027-of-00033.safetensors",
+        "transformer.blocks.25.attn.out_proj.weight": "model-00027-of-00033.safetensors",
+        "transformer.blocks.25.attn.q_ln.bias": "model-00027-of-00033.safetensors",
+        "transformer.blocks.25.attn.q_ln.weight": "model-00027-of-00033.safetensors",
+        "transformer.blocks.25.ffn.down_proj.bias": "model-00027-of-00033.safetensors",
+        "transformer.blocks.25.ffn.down_proj.weight": "model-00027-of-00033.safetensors",
+        "transformer.blocks.25.ffn.up_proj.bias": "model-00027-of-00033.safetensors",
+        "transformer.blocks.25.ffn.up_proj.weight": "model-00027-of-00033.safetensors",
+        "transformer.blocks.25.norm_1.bias": "model-00026-of-00033.safetensors",
+        "transformer.blocks.25.norm_1.weight": "model-00026-of-00033.safetensors",
+        "transformer.blocks.25.norm_2.bias": "model-00027-of-00033.safetensors",
+        "transformer.blocks.25.norm_2.weight": "model-00027-of-00033.safetensors",
+        "transformer.blocks.26.attn.Wqkv.bias": "model-00028-of-00033.safetensors",
+        "transformer.blocks.26.attn.Wqkv.weight": "model-00028-of-00033.safetensors",
+        "transformer.blocks.26.attn.k_ln.bias": "model-00028-of-00033.safetensors",
+        "transformer.blocks.26.attn.k_ln.weight": "model-00028-of-00033.safetensors",
+        "transformer.blocks.26.attn.out_proj.bias": "model-00028-of-00033.safetensors",
+        "transformer.blocks.26.attn.out_proj.weight": "model-00028-of-00033.safetensors",
+        "transformer.blocks.26.attn.q_ln.bias": "model-00028-of-00033.safetensors",
+        "transformer.blocks.26.attn.q_ln.weight": "model-00028-of-00033.safetensors",
+        "transformer.blocks.26.ffn.down_proj.bias": "model-00028-of-00033.safetensors",
+        "transformer.blocks.26.ffn.down_proj.weight": "model-00028-of-00033.safetensors",
+        "transformer.blocks.26.ffn.up_proj.bias": "model-00028-of-00033.safetensors",
+        "transformer.blocks.26.ffn.up_proj.weight": "model-00028-of-00033.safetensors",
+        "transformer.blocks.26.norm_1.bias": "model-00027-of-00033.safetensors",
+        "transformer.blocks.26.norm_1.weight": "model-00027-of-00033.safetensors",
+        "transformer.blocks.26.norm_2.bias": "model-00028-of-00033.safetensors",
+        "transformer.blocks.26.norm_2.weight": "model-00028-of-00033.safetensors",
+        "transformer.blocks.27.attn.Wqkv.bias": "model-00029-of-00033.safetensors",
+        "transformer.blocks.27.attn.Wqkv.weight": "model-00029-of-00033.safetensors",
+        "transformer.blocks.27.attn.k_ln.bias": "model-00029-of-00033.safetensors",
+        "transformer.blocks.27.attn.k_ln.weight": "model-00029-of-00033.safetensors",
+        "transformer.blocks.27.attn.out_proj.bias": "model-00029-of-00033.safetensors",
+        "transformer.blocks.27.attn.out_proj.weight": "model-00029-of-00033.safetensors",
+        "transformer.blocks.27.attn.q_ln.bias": "model-00029-of-00033.safetensors",
+        "transformer.blocks.27.attn.q_ln.weight": "model-00029-of-00033.safetensors",
+        "transformer.blocks.27.ffn.down_proj.bias": "model-00029-of-00033.safetensors",
+        "transformer.blocks.27.ffn.down_proj.weight": "model-00029-of-00033.safetensors",
+        "transformer.blocks.27.ffn.up_proj.bias": "model-00029-of-00033.safetensors",
+        "transformer.blocks.27.ffn.up_proj.weight": "model-00029-of-00033.safetensors",
+        "transformer.blocks.27.norm_1.bias": "model-00028-of-00033.safetensors",
+        "transformer.blocks.27.norm_1.weight": "model-00028-of-00033.safetensors",
+        "transformer.blocks.27.norm_2.bias": "model-00029-of-00033.safetensors",
+        "transformer.blocks.27.norm_2.weight": "model-00029-of-00033.safetensors",
+        "transformer.blocks.28.attn.Wqkv.bias": "model-00030-of-00033.safetensors",
+        "transformer.blocks.28.attn.Wqkv.weight": "model-00030-of-00033.safetensors",
+        "transformer.blocks.28.attn.k_ln.bias": "model-00030-of-00033.safetensors",
+        "transformer.blocks.28.attn.k_ln.weight": "model-00030-of-00033.safetensors",
+        "transformer.blocks.28.attn.out_proj.bias": "model-00030-of-00033.safetensors",
+        "transformer.blocks.28.attn.out_proj.weight": "model-00030-of-00033.safetensors",
+        "transformer.blocks.28.attn.q_ln.bias": "model-00030-of-00033.safetensors",
+        "transformer.blocks.28.attn.q_ln.weight": "model-00030-of-00033.safetensors",
+        "transformer.blocks.28.ffn.down_proj.bias": "model-00030-of-00033.safetensors",
+        "transformer.blocks.28.ffn.down_proj.weight": "model-00030-of-00033.safetensors",
+        "transformer.blocks.28.ffn.up_proj.bias": "model-00030-of-00033.safetensors",
+        "transformer.blocks.28.ffn.up_proj.weight": "model-00030-of-00033.safetensors",
+        "transformer.blocks.28.norm_1.bias": "model-00029-of-00033.safetensors",
+        "transformer.blocks.28.norm_1.weight": "model-00029-of-00033.safetensors",
+        "transformer.blocks.28.norm_2.bias": "model-00030-of-00033.safetensors",
+        "transformer.blocks.28.norm_2.weight": "model-00030-of-00033.safetensors",
+        "transformer.blocks.29.attn.Wqkv.bias": "model-00031-of-00033.safetensors",
+        "transformer.blocks.29.attn.Wqkv.weight": "model-00031-of-00033.safetensors",
+        "transformer.blocks.29.attn.k_ln.bias": "model-00031-of-00033.safetensors",
+        "transformer.blocks.29.attn.k_ln.weight": "model-00031-of-00033.safetensors",
+        "transformer.blocks.29.attn.out_proj.bias": "model-00031-of-00033.safetensors",
+        "transformer.blocks.29.attn.out_proj.weight": "model-00031-of-00033.safetensors",
+        "transformer.blocks.29.attn.q_ln.bias": "model-00031-of-00033.safetensors",
+        "transformer.blocks.29.attn.q_ln.weight": "model-00031-of-00033.safetensors",
+        "transformer.blocks.29.ffn.down_proj.bias": "model-00031-of-00033.safetensors",
+        "transformer.blocks.29.ffn.down_proj.weight": "model-00031-of-00033.safetensors",
+        "transformer.blocks.29.ffn.up_proj.bias": "model-00031-of-00033.safetensors",
+        "transformer.blocks.29.ffn.up_proj.weight": "model-00031-of-00033.safetensors",
+        "transformer.blocks.29.norm_1.bias": "model-00030-of-00033.safetensors",
+        "transformer.blocks.29.norm_1.weight": "model-00030-of-00033.safetensors",
+        "transformer.blocks.29.norm_2.bias": "model-00031-of-00033.safetensors",
+        "transformer.blocks.29.norm_2.weight": "model-00031-of-00033.safetensors",
+        "transformer.blocks.3.attn.Wqkv.bias": "model-00005-of-00033.safetensors",
+        "transformer.blocks.3.attn.Wqkv.weight": "model-00005-of-00033.safetensors",
+        "transformer.blocks.3.attn.k_ln.bias": "model-00005-of-00033.safetensors",
+        "transformer.blocks.3.attn.k_ln.weight": "model-00005-of-00033.safetensors",
+        "transformer.blocks.3.attn.out_proj.bias": "model-00005-of-00033.safetensors",
+        "transformer.blocks.3.attn.out_proj.weight": "model-00005-of-00033.safetensors",
+        "transformer.blocks.3.attn.q_ln.bias": "model-00005-of-00033.safetensors",
+        "transformer.blocks.3.attn.q_ln.weight": "model-00005-of-00033.safetensors",
+        "transformer.blocks.3.ffn.down_proj.bias": "model-00005-of-00033.safetensors",
+        "transformer.blocks.3.ffn.down_proj.weight": "model-00005-of-00033.safetensors",
+        "transformer.blocks.3.ffn.up_proj.bias": "model-00005-of-00033.safetensors",
+        "transformer.blocks.3.ffn.up_proj.weight": "model-00005-of-00033.safetensors",
+        "transformer.blocks.3.norm_1.bias": "model-00004-of-00033.safetensors",
+        "transformer.blocks.3.norm_1.weight": "model-00004-of-00033.safetensors",
+        "transformer.blocks.3.norm_2.bias": "model-00005-of-00033.safetensors",
+        "transformer.blocks.3.norm_2.weight": "model-00005-of-00033.safetensors",
+        "transformer.blocks.30.attn.Wqkv.bias": "model-00032-of-00033.safetensors",
+        "transformer.blocks.30.attn.Wqkv.weight": "model-00032-of-00033.safetensors",
+        "transformer.blocks.30.attn.k_ln.bias": "model-00032-of-00033.safetensors",
+        "transformer.blocks.30.attn.k_ln.weight": "model-00032-of-00033.safetensors",
+        "transformer.blocks.30.attn.out_proj.bias": "model-00032-of-00033.safetensors",
+        "transformer.blocks.30.attn.out_proj.weight": "model-00032-of-00033.safetensors",
+        "transformer.blocks.30.attn.q_ln.bias": "model-00032-of-00033.safetensors",
+        "transformer.blocks.30.attn.q_ln.weight": "model-00032-of-00033.safetensors",
+        "transformer.blocks.30.ffn.down_proj.bias": "model-00032-of-00033.safetensors",
+        "transformer.blocks.30.ffn.down_proj.weight": "model-00032-of-00033.safetensors",
+        "transformer.blocks.30.ffn.up_proj.bias": "model-00032-of-00033.safetensors",
+        "transformer.blocks.30.ffn.up_proj.weight": "model-00032-of-00033.safetensors",
+        "transformer.blocks.30.norm_1.bias": "model-00031-of-00033.safetensors",
+        "transformer.blocks.30.norm_1.weight": "model-00031-of-00033.safetensors",
+        "transformer.blocks.30.norm_2.bias": "model-00032-of-00033.safetensors",
+        "transformer.blocks.30.norm_2.weight": "model-00032-of-00033.safetensors",
+        "transformer.blocks.31.attn.Wqkv.bias": "model-00033-of-00033.safetensors",
+        "transformer.blocks.31.attn.Wqkv.weight": "model-00033-of-00033.safetensors",
+        "transformer.blocks.31.attn.k_ln.bias": "model-00033-of-00033.safetensors",
+        "transformer.blocks.31.attn.k_ln.weight": "model-00033-of-00033.safetensors",
+        "transformer.blocks.31.attn.out_proj.bias": "model-00033-of-00033.safetensors",
+        "transformer.blocks.31.attn.out_proj.weight": "model-00033-of-00033.safetensors",
+        "transformer.blocks.31.attn.q_ln.bias": "model-00033-of-00033.safetensors",
+        "transformer.blocks.31.attn.q_ln.weight": "model-00033-of-00033.safetensors",
+        "transformer.blocks.31.ffn.down_proj.bias": "model-00033-of-00033.safetensors",
+        "transformer.blocks.31.ffn.down_proj.weight": "model-00033-of-00033.safetensors",
+        "transformer.blocks.31.ffn.up_proj.bias": "model-00033-of-00033.safetensors",
+        "transformer.blocks.31.ffn.up_proj.weight": "model-00033-of-00033.safetensors",
+        "transformer.blocks.31.norm_1.bias": "model-00032-of-00033.safetensors",
+        "transformer.blocks.31.norm_1.weight": "model-00032-of-00033.safetensors",
+        "transformer.blocks.31.norm_2.bias": "model-00033-of-00033.safetensors",
+        "transformer.blocks.31.norm_2.weight": "model-00033-of-00033.safetensors",
+        "transformer.blocks.4.attn.Wqkv.bias": "model-00006-of-00033.safetensors",
+        "transformer.blocks.4.attn.Wqkv.weight": "model-00006-of-00033.safetensors",
+        "transformer.blocks.4.attn.k_ln.bias": "model-00006-of-00033.safetensors",
+        "transformer.blocks.4.attn.k_ln.weight": "model-00006-of-00033.safetensors",
+        "transformer.blocks.4.attn.out_proj.bias": "model-00006-of-00033.safetensors",
+        "transformer.blocks.4.attn.out_proj.weight": "model-00006-of-00033.safetensors",
+        "transformer.blocks.4.attn.q_ln.bias": "model-00006-of-00033.safetensors",
+        "transformer.blocks.4.attn.q_ln.weight": "model-00006-of-00033.safetensors",
+        "transformer.blocks.4.ffn.down_proj.bias": "model-00006-of-00033.safetensors",
+        "transformer.blocks.4.ffn.down_proj.weight": "model-00006-of-00033.safetensors",
+        "transformer.blocks.4.ffn.up_proj.bias": "model-00006-of-00033.safetensors",
+        "transformer.blocks.4.ffn.up_proj.weight": "model-00006-of-00033.safetensors",
+        "transformer.blocks.4.norm_1.bias": "model-00005-of-00033.safetensors",
+        "transformer.blocks.4.norm_1.weight": "model-00005-of-00033.safetensors",
+        "transformer.blocks.4.norm_2.bias": "model-00006-of-00033.safetensors",
+        "transformer.blocks.4.norm_2.weight": "model-00006-of-00033.safetensors",
+        "transformer.blocks.5.attn.Wqkv.bias": "model-00007-of-00033.safetensors",
+        "transformer.blocks.5.attn.Wqkv.weight": "model-00007-of-00033.safetensors",
+        "transformer.blocks.5.attn.k_ln.bias": "model-00007-of-00033.safetensors",
+        "transformer.blocks.5.attn.k_ln.weight": "model-00007-of-00033.safetensors",
+        "transformer.blocks.5.attn.out_proj.bias": "model-00007-of-00033.safetensors",
+        "transformer.blocks.5.attn.out_proj.weight": "model-00007-of-00033.safetensors",
+        "transformer.blocks.5.attn.q_ln.bias": "model-00007-of-00033.safetensors",
+        "transformer.blocks.5.attn.q_ln.weight": "model-00007-of-00033.safetensors",
+        "transformer.blocks.5.ffn.down_proj.bias": "model-00007-of-00033.safetensors",
+        "transformer.blocks.5.ffn.down_proj.weight": "model-00007-of-00033.safetensors",
+        "transformer.blocks.5.ffn.up_proj.bias": "model-00007-of-00033.safetensors",
+        "transformer.blocks.5.ffn.up_proj.weight": "model-00007-of-00033.safetensors",
+        "transformer.blocks.5.norm_1.bias": "model-00006-of-00033.safetensors",
+        "transformer.blocks.5.norm_1.weight": "model-00006-of-00033.safetensors",
+        "transformer.blocks.5.norm_2.bias": "model-00007-of-00033.safetensors",
+        "transformer.blocks.5.norm_2.weight": "model-00007-of-00033.safetensors",
+        "transformer.blocks.6.attn.Wqkv.bias": "model-00008-of-00033.safetensors",
+        "transformer.blocks.6.attn.Wqkv.weight": "model-00008-of-00033.safetensors",
+        "transformer.blocks.6.attn.k_ln.bias": "model-00008-of-00033.safetensors",
+        "transformer.blocks.6.attn.k_ln.weight": "model-00008-of-00033.safetensors",
+        "transformer.blocks.6.attn.out_proj.bias": "model-00008-of-00033.safetensors",
+        "transformer.blocks.6.attn.out_proj.weight": "model-00008-of-00033.safetensors",
+        "transformer.blocks.6.attn.q_ln.bias": "model-00008-of-00033.safetensors",
+        "transformer.blocks.6.attn.q_ln.weight": "model-00008-of-00033.safetensors",
+        "transformer.blocks.6.ffn.down_proj.bias": "model-00008-of-00033.safetensors",
+        "transformer.blocks.6.ffn.down_proj.weight": "model-00008-of-00033.safetensors",
+        "transformer.blocks.6.ffn.up_proj.bias": "model-00008-of-00033.safetensors",
+        "transformer.blocks.6.ffn.up_proj.weight": "model-00008-of-00033.safetensors",
+        "transformer.blocks.6.norm_1.bias": "model-00007-of-00033.safetensors",
+        "transformer.blocks.6.norm_1.weight": "model-00007-of-00033.safetensors",
+        "transformer.blocks.6.norm_2.bias": "model-00008-of-00033.safetensors",
+        "transformer.blocks.6.norm_2.weight": "model-00008-of-00033.safetensors",
+        "transformer.blocks.7.attn.Wqkv.bias": "model-00009-of-00033.safetensors",
+        "transformer.blocks.7.attn.Wqkv.weight": "model-00009-of-00033.safetensors",
+        "transformer.blocks.7.attn.k_ln.bias": "model-00009-of-00033.safetensors",
+        "transformer.blocks.7.attn.k_ln.weight": "model-00009-of-00033.safetensors",
+        "transformer.blocks.7.attn.out_proj.bias": "model-00009-of-00033.safetensors",
+        "transformer.blocks.7.attn.out_proj.weight": "model-00009-of-00033.safetensors",
+        "transformer.blocks.7.attn.q_ln.bias": "model-00009-of-00033.safetensors",
+        "transformer.blocks.7.attn.q_ln.weight": "model-00009-of-00033.safetensors",
+        "transformer.blocks.7.ffn.down_proj.bias": "model-00009-of-00033.safetensors",
+        "transformer.blocks.7.ffn.down_proj.weight": "model-00009-of-00033.safetensors",
+        "transformer.blocks.7.ffn.up_proj.bias": "model-00009-of-00033.safetensors",
+        "transformer.blocks.7.ffn.up_proj.weight": "model-00009-of-00033.safetensors",
+        "transformer.blocks.7.norm_1.bias": "model-00008-of-00033.safetensors",
+        "transformer.blocks.7.norm_1.weight": "model-00008-of-00033.safetensors",
+        "transformer.blocks.7.norm_2.bias": "model-00009-of-00033.safetensors",
+        "transformer.blocks.7.norm_2.weight": "model-00009-of-00033.safetensors",
+        "transformer.blocks.8.attn.Wqkv.bias": "model-00010-of-00033.safetensors",
+        "transformer.blocks.8.attn.Wqkv.weight": "model-00010-of-00033.safetensors",
+        "transformer.blocks.8.attn.k_ln.bias": "model-00010-of-00033.safetensors",
+        "transformer.blocks.8.attn.k_ln.weight": "model-00010-of-00033.safetensors",
+        "transformer.blocks.8.attn.out_proj.bias": "model-00010-of-00033.safetensors",
+        "transformer.blocks.8.attn.out_proj.weight": "model-00010-of-00033.safetensors",
+        "transformer.blocks.8.attn.q_ln.bias": "model-00010-of-00033.safetensors",
+        "transformer.blocks.8.attn.q_ln.weight": "model-00010-of-00033.safetensors",
+        "transformer.blocks.8.ffn.down_proj.bias": "model-00010-of-00033.safetensors",
+        "transformer.blocks.8.ffn.down_proj.weight": "model-00010-of-00033.safetensors",
+        "transformer.blocks.8.ffn.up_proj.bias": "model-00010-of-00033.safetensors",
+        "transformer.blocks.8.ffn.up_proj.weight": "model-00010-of-00033.safetensors",
+        "transformer.blocks.8.norm_1.bias": "model-00009-of-00033.safetensors",
+        "transformer.blocks.8.norm_1.weight": "model-00009-of-00033.safetensors",
+        "transformer.blocks.8.norm_2.bias": "model-00010-of-00033.safetensors",
+        "transformer.blocks.8.norm_2.weight": "model-00010-of-00033.safetensors",
+        "transformer.blocks.9.attn.Wqkv.bias": "model-00011-of-00033.safetensors",
+        "transformer.blocks.9.attn.Wqkv.weight": "model-00011-of-00033.safetensors",
+        "transformer.blocks.9.attn.k_ln.bias": "model-00011-of-00033.safetensors",
+        "transformer.blocks.9.attn.k_ln.weight": "model-00011-of-00033.safetensors",
+        "transformer.blocks.9.attn.out_proj.bias": "model-00011-of-00033.safetensors",
+        "transformer.blocks.9.attn.out_proj.weight": "model-00011-of-00033.safetensors",
+        "transformer.blocks.9.attn.q_ln.bias": "model-00011-of-00033.safetensors",
+        "transformer.blocks.9.attn.q_ln.weight": "model-00011-of-00033.safetensors",
+        "transformer.blocks.9.ffn.down_proj.bias": "model-00011-of-00033.safetensors",
+        "transformer.blocks.9.ffn.down_proj.weight": "model-00011-of-00033.safetensors",
+        "transformer.blocks.9.ffn.up_proj.bias": "model-00011-of-00033.safetensors",
+        "transformer.blocks.9.ffn.up_proj.weight": "model-00011-of-00033.safetensors",
+        "transformer.blocks.9.norm_1.bias": "model-00010-of-00033.safetensors",
+        "transformer.blocks.9.norm_1.weight": "model-00010-of-00033.safetensors",
+        "transformer.blocks.9.norm_2.bias": "model-00011-of-00033.safetensors",
+        "transformer.blocks.9.norm_2.weight": "model-00011-of-00033.safetensors",
+        "transformer.norm_f.bias": "model-00033-of-00033.safetensors",
+        "transformer.norm_f.weight": "model-00033-of-00033.safetensors",
+        "transformer.wpe.weight": "model-00002-of-00033.safetensors",
+        "transformer.wte.weight": "model-00001-of-00033.safetensors"
+    }
+}