Upload LlamaForCausalLM

config.json

@@ -0,0 +1,23 @@
+{
+  "_name_or_path": "sharpbai/llama-13b-hf",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 5120,
+  "initializer_range": 0.02,
+  "intermediate_size": 13824,
+  "max_position_embeddings": 2048,
+  "model_type": "llama",
+  "num_attention_heads": 40,
+  "num_hidden_layers": 40,
+  "pad_token_id": 0,
+  "rms_norm_eps": 1e-06,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.30.1",
+  "use_cache": true,
+  "vocab_size": 32000
+}

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "pad_token_id": 0,
+  "transformers_version": "4.30.1"
+}

pytorch_model-00001-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49ed476de4201911439b088d5c6bfb57ae0789858a59f45dcbc814e946051559
+size 537397951

pytorch_model-00002-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:060b08cef95051a860d4df5a1bf8b8acd2383464d2e80b1829424f30fdd655cb
+size 634412995

pytorch_model-00003-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3389b32e4e1624d20e0902786ee6e7b70db36c227733710787065fe67e295bf
+size 634412995

pytorch_model-00004-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e84e4f13cdcc34b0cd7ce81b0b84e5fb33db2d7efaa60d95ac3aece822f24080
+size 634412995

pytorch_model-00005-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8c759dd236ba578da30ff0acfc1ee70cf73aa87cfa40eb0df3f60cfd9c47c3f
+size 634412995

pytorch_model-00006-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6c89208f8a688cee6b2472f1ec0f7b48042882930bb881c96791e3bb99b2dc6
+size 634412995

pytorch_model-00007-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9f2041804d0038d9c94bc623fb09d524abb36b53fc517836e1405797efffc0c
+size 634412995

pytorch_model-00008-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0dc9f9c4220521b7b754a8fbd092993291d366024b678d4031bbc4ad4545ee4
+size 634412995

pytorch_model-00009-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e960ad014360b7e631b37b43b1090dd9b6547e43c361539c44793d72ce5f899d
+size 634412995

pytorch_model-00010-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d032ac7467288a5d014cf40ffc35591b59183ac5850aac93a273a12ba63f370f
+size 634412995

pytorch_model-00011-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae7e946cd3a06b414494862cc53ecf7d45a78ca233d24c20bbd13c331f2831f6
+size 634412995

pytorch_model-00012-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef31dd926c9c415f8fce8985f6481e41ec9f93df0407c58716dfc32df99bf0e9
+size 634412995

pytorch_model-00013-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80ca97bef67e3649c02e53c9e986aefeda5ba9aad4fc8720e0244b93e0091175
+size 634412995

pytorch_model-00014-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:226f787eee2f71e99ace227465f3ed60222554d8ac70bfbba74ee9d140bc7e27
+size 634412995

pytorch_model-00015-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:392351699b422ff79b8fbc9fde0f32dadd06ded75d7af8de62ed58c045e14fdb
+size 634412995

pytorch_model-00016-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2bb3236a3ec99748f62ffe85378c8c8b182fda2a1bb206be941e162c75863982
+size 634412995

pytorch_model-00017-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:246657b659fef5126dac5e16c3aae4409c292468ccdaf8e2042823fa23c6ab35
+size 634412995

pytorch_model-00018-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f6fb2e42acc782eca34d1b49c888150462ea2720479cfb54cc0ea455abdceed
+size 634412995

pytorch_model-00019-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:463d67bb44c0b3ccb704906fba0c60e988791309e6abaacae25b67d7e95514fb
+size 634412995

pytorch_model-00020-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:494e86b0e75c4343c86b8e488fef7bdcff7fd9be82b6b6df4618abc77285a12e
+size 634412995

pytorch_model-00021-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:301e001fa44a57bd533379f969334bf08d031543bf5c509d28d9654dd7690989
+size 634412995

pytorch_model-00022-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef7a378c4511b4fa8ed971dcfd5574c358dae1d6542fca1fef03bf8f2d8b341f
+size 634412995

pytorch_model-00023-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a005fc69f32b665cbe504f0f02d233da8c7fcb254052831c7a08e33eb0ea9604
+size 634412995

pytorch_model-00024-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5170ca5858c2966e6c38d674e38faa2e99294d850fb3e2604f930a8de01ab091
+size 634412995

pytorch_model-00025-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6010839b1ceb88d302a0913437fd6e4e7d456b2b57d60dac981ed4623ccbeb4c
+size 634412995

pytorch_model-00026-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56fbb64cd51ab0ba19eebc2ea779c9689ea6140764411d4a96aef98bd0c4ebda
+size 634412995

pytorch_model-00027-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fbf7f9e2849767f380c8e6efd3988a7e5c898368bfed763ec9aa0b2f856d614f
+size 634412995

pytorch_model-00028-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f439e5a2fc3692a075e4b706d212ec62bcc02c18d46ff5c00c1a053066688297
+size 634412995

pytorch_model-00029-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d07292bc9b9e40664c113fd0e5871e16d6f2cd6b4766aced8e93ffc30c67f28
+size 634412995

pytorch_model-00030-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3371d79af348cb122f42e97dd91178b0c3a74a26f04e6a0bd161e1c1ef8ee3ce
+size 634412995

pytorch_model-00031-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e7edbc850a0e20c04e49e963fdb1e4279ca97dd1927c35c37d8ab9e9ffc7d4c8
+size 634412995

pytorch_model-00032-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd8ab169429293b54befdc7645529b377951df45db52382c2527304c074ffae9
+size 634412995

pytorch_model-00033-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8dfbec3e17b8627393140860a1c44b4d78bb9e29d4c985322c994d6bb8b4768f
+size 634412995

pytorch_model-00034-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4339e78b99179387901182e8fca889dcfe9f573e221a2d21708989a071adbe56
+size 634412995

pytorch_model-00035-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52105ef4561e559b054f2a1df4ecc118d3927da6f41f39df832962ba29a0e5b2
+size 634412995

pytorch_model-00036-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91f2d566d8400b2f16b89bf4b605827f9bc2d98a0b6f03bf0fd9bc27149703ec
+size 634412995

pytorch_model-00037-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:530f8a1a2bd61dd1c6e805f347c138d5aa75b5787a0e91decfeb700fef10d2e8
+size 634412995

pytorch_model-00038-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0eb31bd6937ee5c8e0a0172954b35f7ce2ace47703a6bc4ef61d8527a707da4c
+size 634412995

pytorch_model-00039-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff35382995ce7cb654912ce25dfbc36e1fd01f57dfd0a417940172a66a50e0cc
+size 634412995

pytorch_model-00040-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0f02357838b086491f2b2202ae764d6ee2d8e96403a95227aa92f33e8e39548
+size 634412995

pytorch_model-00041-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aeca5cce1f105b9ac32614bc5b4020e66c257d910df1ecfaa9ffd18e99e34f6a
+size 752386495

pytorch_model-00042-of-00042.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a98cefb947b9bdf9a87682636b2996e326140e0b26970d6fc2d49a6148c27
+size 327680938

pytorch_model.bin.index.json

@@ -0,0 +1,410 @@
+{
+  "metadata": {
+    "total_size": 26031738880
+  },
+  "weight_map": {
+    "lm_head.weight": "pytorch_model-00042-of-00042.bin",
+    "model.embed_tokens.weight": "pytorch_model-00001-of-00042.bin",
+    "model.layers.0.input_layernorm.weight": "pytorch_model-00002-of-00042.bin",
+    "model.layers.0.mlp.down_proj.weight": "pytorch_model-00002-of-00042.bin",
+    "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00002-of-00042.bin",
+    "model.layers.0.mlp.up_proj.weight": "pytorch_model-00002-of-00042.bin",
+    "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00002-of-00042.bin",
+    "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00042.bin",
+    "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00042.bin",
+    "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00042.bin",
+    "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00042.bin",
+    "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00042.bin",
+    "model.layers.1.input_layernorm.weight": "pytorch_model-00003-of-00042.bin",
+    "model.layers.1.mlp.down_proj.weight": "pytorch_model-00003-of-00042.bin",
+    "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00003-of-00042.bin",
+    "model.layers.1.mlp.up_proj.weight": "pytorch_model-00003-of-00042.bin",
+    "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00003-of-00042.bin",
+    "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00002-of-00042.bin",
+    "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00002-of-00042.bin",
+    "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00002-of-00042.bin",
+    "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00042.bin",
+    "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00002-of-00042.bin",
+    "model.layers.10.input_layernorm.weight": "pytorch_model-00012-of-00042.bin",
+    "model.layers.10.mlp.down_proj.weight": "pytorch_model-00012-of-00042.bin",
+    "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00012-of-00042.bin",
+    "model.layers.10.mlp.up_proj.weight": "pytorch_model-00012-of-00042.bin",
+    "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00012-of-00042.bin",
+    "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00011-of-00042.bin",
+    "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00011-of-00042.bin",
+    "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00011-of-00042.bin",
+    "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00011-of-00042.bin",
+    "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00011-of-00042.bin",
+    "model.layers.11.input_layernorm.weight": "pytorch_model-00013-of-00042.bin",
+    "model.layers.11.mlp.down_proj.weight": "pytorch_model-00013-of-00042.bin",
+    "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00013-of-00042.bin",
+    "model.layers.11.mlp.up_proj.weight": "pytorch_model-00013-of-00042.bin",
+    "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00013-of-00042.bin",
+    "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00012-of-00042.bin",
+    "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00012-of-00042.bin",
+    "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00012-of-00042.bin",
+    "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00012-of-00042.bin",
+    "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00012-of-00042.bin",
+    "model.layers.12.input_layernorm.weight": "pytorch_model-00014-of-00042.bin",
+    "model.layers.12.mlp.down_proj.weight": "pytorch_model-00014-of-00042.bin",
+    "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00014-of-00042.bin",
+    "model.layers.12.mlp.up_proj.weight": "pytorch_model-00014-of-00042.bin",
+    "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00014-of-00042.bin",
+    "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00013-of-00042.bin",
+    "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00013-of-00042.bin",
+    "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00013-of-00042.bin",
+    "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00013-of-00042.bin",
+    "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00013-of-00042.bin",
+    "model.layers.13.input_layernorm.weight": "pytorch_model-00015-of-00042.bin",
+    "model.layers.13.mlp.down_proj.weight": "pytorch_model-00015-of-00042.bin",
+    "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00015-of-00042.bin",
+    "model.layers.13.mlp.up_proj.weight": "pytorch_model-00015-of-00042.bin",
+    "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00015-of-00042.bin",
+    "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00014-of-00042.bin",
+    "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00014-of-00042.bin",
+    "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00014-of-00042.bin",
+    "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00014-of-00042.bin",
+    "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00014-of-00042.bin",
+    "model.layers.14.input_layernorm.weight": "pytorch_model-00016-of-00042.bin",
+    "model.layers.14.mlp.down_proj.weight": "pytorch_model-00016-of-00042.bin",
+    "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00016-of-00042.bin",
+    "model.layers.14.mlp.up_proj.weight": "pytorch_model-00016-of-00042.bin",
+    "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00016-of-00042.bin",
+    "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00015-of-00042.bin",
+    "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00015-of-00042.bin",
+    "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00015-of-00042.bin",
+    "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00015-of-00042.bin",
+    "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00015-of-00042.bin",
+    "model.layers.15.input_layernorm.weight": "pytorch_model-00017-of-00042.bin",
+    "model.layers.15.mlp.down_proj.weight": "pytorch_model-00017-of-00042.bin",
+    "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00017-of-00042.bin",
+    "model.layers.15.mlp.up_proj.weight": "pytorch_model-00017-of-00042.bin",
+    "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00017-of-00042.bin",
+    "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00016-of-00042.bin",
+    "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00016-of-00042.bin",
+    "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00016-of-00042.bin",
+    "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00016-of-00042.bin",
+    "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00016-of-00042.bin",
+    "model.layers.16.input_layernorm.weight": "pytorch_model-00018-of-00042.bin",
+    "model.layers.16.mlp.down_proj.weight": "pytorch_model-00018-of-00042.bin",
+    "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00018-of-00042.bin",
+    "model.layers.16.mlp.up_proj.weight": "pytorch_model-00018-of-00042.bin",
+    "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00018-of-00042.bin",
+    "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00017-of-00042.bin",
+    "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00017-of-00042.bin",
+    "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00017-of-00042.bin",
+    "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00017-of-00042.bin",
+    "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00017-of-00042.bin",
+    "model.layers.17.input_layernorm.weight": "pytorch_model-00019-of-00042.bin",
+    "model.layers.17.mlp.down_proj.weight": "pytorch_model-00019-of-00042.bin",
+    "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00019-of-00042.bin",
+    "model.layers.17.mlp.up_proj.weight": "pytorch_model-00019-of-00042.bin",
+    "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00019-of-00042.bin",
+    "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00018-of-00042.bin",
+    "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00018-of-00042.bin",
+    "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00018-of-00042.bin",
+    "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00018-of-00042.bin",
+    "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00018-of-00042.bin",
+    "model.layers.18.input_layernorm.weight": "pytorch_model-00020-of-00042.bin",
+    "model.layers.18.mlp.down_proj.weight": "pytorch_model-00020-of-00042.bin",
+    "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00020-of-00042.bin",
+    "model.layers.18.mlp.up_proj.weight": "pytorch_model-00020-of-00042.bin",
+    "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00020-of-00042.bin",
+    "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00019-of-00042.bin",
+    "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00019-of-00042.bin",
+    "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00019-of-00042.bin",
+    "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00019-of-00042.bin",
+    "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00019-of-00042.bin",
+    "model.layers.19.input_layernorm.weight": "pytorch_model-00021-of-00042.bin",
+    "model.layers.19.mlp.down_proj.weight": "pytorch_model-00021-of-00042.bin",
+    "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00021-of-00042.bin",
+    "model.layers.19.mlp.up_proj.weight": "pytorch_model-00021-of-00042.bin",
+    "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00021-of-00042.bin",
+    "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00020-of-00042.bin",
+    "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00020-of-00042.bin",
+    "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00020-of-00042.bin",
+    "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00020-of-00042.bin",
+    "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00020-of-00042.bin",
+    "model.layers.2.input_layernorm.weight": "pytorch_model-00004-of-00042.bin",
+    "model.layers.2.mlp.down_proj.weight": "pytorch_model-00004-of-00042.bin",
+    "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00004-of-00042.bin",
+    "model.layers.2.mlp.up_proj.weight": "pytorch_model-00004-of-00042.bin",
+    "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00004-of-00042.bin",
+    "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00003-of-00042.bin",
+    "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00003-of-00042.bin",
+    "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00003-of-00042.bin",
+    "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00042.bin",
+    "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00003-of-00042.bin",
+    "model.layers.20.input_layernorm.weight": "pytorch_model-00022-of-00042.bin",
+    "model.layers.20.mlp.down_proj.weight": "pytorch_model-00022-of-00042.bin",
+    "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00022-of-00042.bin",
+    "model.layers.20.mlp.up_proj.weight": "pytorch_model-00022-of-00042.bin",
+    "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00022-of-00042.bin",
+    "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00021-of-00042.bin",
+    "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00021-of-00042.bin",
+    "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00021-of-00042.bin",
+    "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00021-of-00042.bin",
+    "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00021-of-00042.bin",
+    "model.layers.21.input_layernorm.weight": "pytorch_model-00023-of-00042.bin",
+    "model.layers.21.mlp.down_proj.weight": "pytorch_model-00023-of-00042.bin",
+    "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00023-of-00042.bin",
+    "model.layers.21.mlp.up_proj.weight": "pytorch_model-00023-of-00042.bin",
+    "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00023-of-00042.bin",
+    "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00022-of-00042.bin",
+    "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00022-of-00042.bin",
+    "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00022-of-00042.bin",
+    "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00022-of-00042.bin",
+    "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00022-of-00042.bin",
+    "model.layers.22.input_layernorm.weight": "pytorch_model-00024-of-00042.bin",
+    "model.layers.22.mlp.down_proj.weight": "pytorch_model-00024-of-00042.bin",
+    "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00024-of-00042.bin",
+    "model.layers.22.mlp.up_proj.weight": "pytorch_model-00024-of-00042.bin",
+    "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00024-of-00042.bin",
+    "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00023-of-00042.bin",
+    "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00023-of-00042.bin",
+    "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00023-of-00042.bin",
+    "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00023-of-00042.bin",
+    "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00023-of-00042.bin",
+    "model.layers.23.input_layernorm.weight": "pytorch_model-00025-of-00042.bin",
+    "model.layers.23.mlp.down_proj.weight": "pytorch_model-00025-of-00042.bin",
+    "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00025-of-00042.bin",
+    "model.layers.23.mlp.up_proj.weight": "pytorch_model-00025-of-00042.bin",
+    "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00025-of-00042.bin",
+    "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00024-of-00042.bin",
+    "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00024-of-00042.bin",
+    "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00024-of-00042.bin",
+    "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00024-of-00042.bin",
+    "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00024-of-00042.bin",
+    "model.layers.24.input_layernorm.weight": "pytorch_model-00026-of-00042.bin",
+    "model.layers.24.mlp.down_proj.weight": "pytorch_model-00026-of-00042.bin",
+    "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00026-of-00042.bin",
+    "model.layers.24.mlp.up_proj.weight": "pytorch_model-00026-of-00042.bin",
+    "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00026-of-00042.bin",
+    "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00025-of-00042.bin",
+    "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00025-of-00042.bin",
+    "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00025-of-00042.bin",
+    "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00025-of-00042.bin",
+    "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00025-of-00042.bin",
+    "model.layers.25.input_layernorm.weight": "pytorch_model-00027-of-00042.bin",
+    "model.layers.25.mlp.down_proj.weight": "pytorch_model-00027-of-00042.bin",
+    "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00027-of-00042.bin",
+    "model.layers.25.mlp.up_proj.weight": "pytorch_model-00027-of-00042.bin",
+    "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00027-of-00042.bin",
+    "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00026-of-00042.bin",
+    "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00026-of-00042.bin",
+    "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00026-of-00042.bin",
+    "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00026-of-00042.bin",
+    "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00026-of-00042.bin",
+    "model.layers.26.input_layernorm.weight": "pytorch_model-00028-of-00042.bin",
+    "model.layers.26.mlp.down_proj.weight": "pytorch_model-00028-of-00042.bin",
+    "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00028-of-00042.bin",
+    "model.layers.26.mlp.up_proj.weight": "pytorch_model-00028-of-00042.bin",
+    "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00028-of-00042.bin",
+    "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00027-of-00042.bin",
+    "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00027-of-00042.bin",
+    "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00027-of-00042.bin",
+    "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00027-of-00042.bin",
+    "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00027-of-00042.bin",
+    "model.layers.27.input_layernorm.weight": "pytorch_model-00029-of-00042.bin",
+    "model.layers.27.mlp.down_proj.weight": "pytorch_model-00029-of-00042.bin",
+    "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00029-of-00042.bin",
+    "model.layers.27.mlp.up_proj.weight": "pytorch_model-00029-of-00042.bin",
+    "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00029-of-00042.bin",
+    "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00028-of-00042.bin",
+    "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00028-of-00042.bin",
+    "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00028-of-00042.bin",
+    "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00028-of-00042.bin",
+    "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00028-of-00042.bin",
+    "model.layers.28.input_layernorm.weight": "pytorch_model-00030-of-00042.bin",
+    "model.layers.28.mlp.down_proj.weight": "pytorch_model-00030-of-00042.bin",
+    "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00030-of-00042.bin",
+    "model.layers.28.mlp.up_proj.weight": "pytorch_model-00030-of-00042.bin",
+    "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00030-of-00042.bin",
+    "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00029-of-00042.bin",
+    "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00029-of-00042.bin",
+    "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00029-of-00042.bin",
+    "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00029-of-00042.bin",
+    "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00029-of-00042.bin",
+    "model.layers.29.input_layernorm.weight": "pytorch_model-00031-of-00042.bin",
+    "model.layers.29.mlp.down_proj.weight": "pytorch_model-00031-of-00042.bin",
+    "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00031-of-00042.bin",
+    "model.layers.29.mlp.up_proj.weight": "pytorch_model-00031-of-00042.bin",
+    "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00031-of-00042.bin",
+    "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00030-of-00042.bin",
+    "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00030-of-00042.bin",
+    "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00030-of-00042.bin",
+    "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00030-of-00042.bin",
+    "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00030-of-00042.bin",
+    "model.layers.3.input_layernorm.weight": "pytorch_model-00005-of-00042.bin",
+    "model.layers.3.mlp.down_proj.weight": "pytorch_model-00005-of-00042.bin",
+    "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00005-of-00042.bin",
+    "model.layers.3.mlp.up_proj.weight": "pytorch_model-00005-of-00042.bin",
+    "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00005-of-00042.bin",
+    "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00004-of-00042.bin",
+    "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00004-of-00042.bin",
+    "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00004-of-00042.bin",
+    "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00042.bin",
+    "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00004-of-00042.bin",
+    "model.layers.30.input_layernorm.weight": "pytorch_model-00032-of-00042.bin",
+    "model.layers.30.mlp.down_proj.weight": "pytorch_model-00032-of-00042.bin",
+    "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00032-of-00042.bin",
+    "model.layers.30.mlp.up_proj.weight": "pytorch_model-00032-of-00042.bin",
+    "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00032-of-00042.bin",
+    "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00031-of-00042.bin",
+    "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00031-of-00042.bin",
+    "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00031-of-00042.bin",
+    "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00031-of-00042.bin",
+    "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00031-of-00042.bin",
+    "model.layers.31.input_layernorm.weight": "pytorch_model-00033-of-00042.bin",
+    "model.layers.31.mlp.down_proj.weight": "pytorch_model-00033-of-00042.bin",
+    "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00033-of-00042.bin",
+    "model.layers.31.mlp.up_proj.weight": "pytorch_model-00033-of-00042.bin",
+    "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00033-of-00042.bin",
+    "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00032-of-00042.bin",
+    "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00032-of-00042.bin",
+    "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00032-of-00042.bin",
+    "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00032-of-00042.bin",
+    "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00032-of-00042.bin",
+    "model.layers.32.input_layernorm.weight": "pytorch_model-00034-of-00042.bin",
+    "model.layers.32.mlp.down_proj.weight": "pytorch_model-00034-of-00042.bin",
+    "model.layers.32.mlp.gate_proj.weight": "pytorch_model-00034-of-00042.bin",
+    "model.layers.32.mlp.up_proj.weight": "pytorch_model-00034-of-00042.bin",
+    "model.layers.32.post_attention_layernorm.weight": "pytorch_model-00034-of-00042.bin",
+    "model.layers.32.self_attn.k_proj.weight": "pytorch_model-00033-of-00042.bin",
+    "model.layers.32.self_attn.o_proj.weight": "pytorch_model-00033-of-00042.bin",
+    "model.layers.32.self_attn.q_proj.weight": "pytorch_model-00033-of-00042.bin",
+    "model.layers.32.self_attn.rotary_emb.inv_freq": "pytorch_model-00033-of-00042.bin",
+    "model.layers.32.self_attn.v_proj.weight": "pytorch_model-00033-of-00042.bin",
+    "model.layers.33.input_layernorm.weight": "pytorch_model-00035-of-00042.bin",
+    "model.layers.33.mlp.down_proj.weight": "pytorch_model-00035-of-00042.bin",
+    "model.layers.33.mlp.gate_proj.weight": "pytorch_model-00035-of-00042.bin",
+    "model.layers.33.mlp.up_proj.weight": "pytorch_model-00035-of-00042.bin",
+    "model.layers.33.post_attention_layernorm.weight": "pytorch_model-00035-of-00042.bin",
+    "model.layers.33.self_attn.k_proj.weight": "pytorch_model-00034-of-00042.bin",
+    "model.layers.33.self_attn.o_proj.weight": "pytorch_model-00034-of-00042.bin",
+    "model.layers.33.self_attn.q_proj.weight": "pytorch_model-00034-of-00042.bin",
+    "model.layers.33.self_attn.rotary_emb.inv_freq": "pytorch_model-00034-of-00042.bin",
+    "model.layers.33.self_attn.v_proj.weight": "pytorch_model-00034-of-00042.bin",
+    "model.layers.34.input_layernorm.weight": "pytorch_model-00036-of-00042.bin",
+    "model.layers.34.mlp.down_proj.weight": "pytorch_model-00036-of-00042.bin",
+    "model.layers.34.mlp.gate_proj.weight": "pytorch_model-00036-of-00042.bin",
+    "model.layers.34.mlp.up_proj.weight": "pytorch_model-00036-of-00042.bin",
+    "model.layers.34.post_attention_layernorm.weight": "pytorch_model-00036-of-00042.bin",
+    "model.layers.34.self_attn.k_proj.weight": "pytorch_model-00035-of-00042.bin",
+    "model.layers.34.self_attn.o_proj.weight": "pytorch_model-00035-of-00042.bin",
+    "model.layers.34.self_attn.q_proj.weight": "pytorch_model-00035-of-00042.bin",
+    "model.layers.34.self_attn.rotary_emb.inv_freq": "pytorch_model-00035-of-00042.bin",
+    "model.layers.34.self_attn.v_proj.weight": "pytorch_model-00035-of-00042.bin",
+    "model.layers.35.input_layernorm.weight": "pytorch_model-00037-of-00042.bin",
+    "model.layers.35.mlp.down_proj.weight": "pytorch_model-00037-of-00042.bin",
+    "model.layers.35.mlp.gate_proj.weight": "pytorch_model-00037-of-00042.bin",
+    "model.layers.35.mlp.up_proj.weight": "pytorch_model-00037-of-00042.bin",
+    "model.layers.35.post_attention_layernorm.weight": "pytorch_model-00037-of-00042.bin",
+    "model.layers.35.self_attn.k_proj.weight": "pytorch_model-00036-of-00042.bin",
+    "model.layers.35.self_attn.o_proj.weight": "pytorch_model-00036-of-00042.bin",
+    "model.layers.35.self_attn.q_proj.weight": "pytorch_model-00036-of-00042.bin",
+    "model.layers.35.self_attn.rotary_emb.inv_freq": "pytorch_model-00036-of-00042.bin",
+    "model.layers.35.self_attn.v_proj.weight": "pytorch_model-00036-of-00042.bin",
+    "model.layers.36.input_layernorm.weight": "pytorch_model-00038-of-00042.bin",
+    "model.layers.36.mlp.down_proj.weight": "pytorch_model-00038-of-00042.bin",
+    "model.layers.36.mlp.gate_proj.weight": "pytorch_model-00038-of-00042.bin",
+    "model.layers.36.mlp.up_proj.weight": "pytorch_model-00038-of-00042.bin",
+    "model.layers.36.post_attention_layernorm.weight": "pytorch_model-00038-of-00042.bin",
+    "model.layers.36.self_attn.k_proj.weight": "pytorch_model-00037-of-00042.bin",
+    "model.layers.36.self_attn.o_proj.weight": "pytorch_model-00037-of-00042.bin",
+    "model.layers.36.self_attn.q_proj.weight": "pytorch_model-00037-of-00042.bin",
+    "model.layers.36.self_attn.rotary_emb.inv_freq": "pytorch_model-00037-of-00042.bin",
+    "model.layers.36.self_attn.v_proj.weight": "pytorch_model-00037-of-00042.bin",
+    "model.layers.37.input_layernorm.weight": "pytorch_model-00039-of-00042.bin",
+    "model.layers.37.mlp.down_proj.weight": "pytorch_model-00039-of-00042.bin",
+    "model.layers.37.mlp.gate_proj.weight": "pytorch_model-00039-of-00042.bin",
+    "model.layers.37.mlp.up_proj.weight": "pytorch_model-00039-of-00042.bin",
+    "model.layers.37.post_attention_layernorm.weight": "pytorch_model-00039-of-00042.bin",
+    "model.layers.37.self_attn.k_proj.weight": "pytorch_model-00038-of-00042.bin",
+    "model.layers.37.self_attn.o_proj.weight": "pytorch_model-00038-of-00042.bin",
+    "model.layers.37.self_attn.q_proj.weight": "pytorch_model-00038-of-00042.bin",
+    "model.layers.37.self_attn.rotary_emb.inv_freq": "pytorch_model-00038-of-00042.bin",
+    "model.layers.37.self_attn.v_proj.weight": "pytorch_model-00038-of-00042.bin",
+    "model.layers.38.input_layernorm.weight": "pytorch_model-00040-of-00042.bin",
+    "model.layers.38.mlp.down_proj.weight": "pytorch_model-00040-of-00042.bin",
+    "model.layers.38.mlp.gate_proj.weight": "pytorch_model-00040-of-00042.bin",
+    "model.layers.38.mlp.up_proj.weight": "pytorch_model-00040-of-00042.bin",
+    "model.layers.38.post_attention_layernorm.weight": "pytorch_model-00040-of-00042.bin",
+    "model.layers.38.self_attn.k_proj.weight": "pytorch_model-00039-of-00042.bin",
+    "model.layers.38.self_attn.o_proj.weight": "pytorch_model-00039-of-00042.bin",
+    "model.layers.38.self_attn.q_proj.weight": "pytorch_model-00039-of-00042.bin",
+    "model.layers.38.self_attn.rotary_emb.inv_freq": "pytorch_model-00039-of-00042.bin",
+    "model.layers.38.self_attn.v_proj.weight": "pytorch_model-00039-of-00042.bin",
+    "model.layers.39.input_layernorm.weight": "pytorch_model-00041-of-00042.bin",
+    "model.layers.39.mlp.down_proj.weight": "pytorch_model-00041-of-00042.bin",
+    "model.layers.39.mlp.gate_proj.weight": "pytorch_model-00041-of-00042.bin",
+    "model.layers.39.mlp.up_proj.weight": "pytorch_model-00041-of-00042.bin",
+    "model.layers.39.post_attention_layernorm.weight": "pytorch_model-00041-of-00042.bin",
+    "model.layers.39.self_attn.k_proj.weight": "pytorch_model-00040-of-00042.bin",
+    "model.layers.39.self_attn.o_proj.weight": "pytorch_model-00040-of-00042.bin",
+    "model.layers.39.self_attn.q_proj.weight": "pytorch_model-00040-of-00042.bin",
+    "model.layers.39.self_attn.rotary_emb.inv_freq": "pytorch_model-00040-of-00042.bin",
+    "model.layers.39.self_attn.v_proj.weight": "pytorch_model-00040-of-00042.bin",
+    "model.layers.4.input_layernorm.weight": "pytorch_model-00006-of-00042.bin",
+    "model.layers.4.mlp.down_proj.weight": "pytorch_model-00006-of-00042.bin",
+    "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00006-of-00042.bin",
+    "model.layers.4.mlp.up_proj.weight": "pytorch_model-00006-of-00042.bin",
+    "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00006-of-00042.bin",
+    "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00005-of-00042.bin",
+    "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00005-of-00042.bin",
+    "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00005-of-00042.bin",
+    "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00042.bin",
+    "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00005-of-00042.bin",
+    "model.layers.5.input_layernorm.weight": "pytorch_model-00007-of-00042.bin",
+    "model.layers.5.mlp.down_proj.weight": "pytorch_model-00007-of-00042.bin",
+    "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00007-of-00042.bin",
+    "model.layers.5.mlp.up_proj.weight": "pytorch_model-00007-of-00042.bin",
+    "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00007-of-00042.bin",
+    "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00006-of-00042.bin",
+    "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00006-of-00042.bin",
+    "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00006-of-00042.bin",
+    "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00042.bin",
+    "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00006-of-00042.bin",
+    "model.layers.6.input_layernorm.weight": "pytorch_model-00008-of-00042.bin",
+    "model.layers.6.mlp.down_proj.weight": "pytorch_model-00008-of-00042.bin",
+    "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00008-of-00042.bin",
+    "model.layers.6.mlp.up_proj.weight": "pytorch_model-00008-of-00042.bin",
+    "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00008-of-00042.bin",
+    "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00007-of-00042.bin",
+    "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00007-of-00042.bin",
+    "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00007-of-00042.bin",
+    "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00042.bin",
+    "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00007-of-00042.bin",
+    "model.layers.7.input_layernorm.weight": "pytorch_model-00009-of-00042.bin",
+    "model.layers.7.mlp.down_proj.weight": "pytorch_model-00009-of-00042.bin",
+    "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00009-of-00042.bin",
+    "model.layers.7.mlp.up_proj.weight": "pytorch_model-00009-of-00042.bin",
+    "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00009-of-00042.bin",
+    "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00008-of-00042.bin",
+    "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00008-of-00042.bin",
+    "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00008-of-00042.bin",
+    "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00008-of-00042.bin",
+    "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00008-of-00042.bin",
+    "model.layers.8.input_layernorm.weight": "pytorch_model-00010-of-00042.bin",
+    "model.layers.8.mlp.down_proj.weight": "pytorch_model-00010-of-00042.bin",
+    "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00010-of-00042.bin",
+    "model.layers.8.mlp.up_proj.weight": "pytorch_model-00010-of-00042.bin",
+    "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00010-of-00042.bin",
+    "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00009-of-00042.bin",
+    "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00009-of-00042.bin",
+    "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00009-of-00042.bin",
+    "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00009-of-00042.bin",
+    "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00009-of-00042.bin",
+    "model.layers.9.input_layernorm.weight": "pytorch_model-00011-of-00042.bin",
+    "model.layers.9.mlp.down_proj.weight": "pytorch_model-00011-of-00042.bin",
+    "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00011-of-00042.bin",
+    "model.layers.9.mlp.up_proj.weight": "pytorch_model-00011-of-00042.bin",
+    "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00011-of-00042.bin",
+    "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00010-of-00042.bin",
+    "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00010-of-00042.bin",
+    "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00010-of-00042.bin",
+    "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00010-of-00042.bin",
+    "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00010-of-00042.bin",
+    "model.norm.weight": "pytorch_model-00041-of-00042.bin"
+  }
+}