Upload LlamaForCausalLM

config.json

@@ -0,0 +1,23 @@
+{
+  "_name_or_path": "./alpaca-combined-hub",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 2048,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "pad_token_id": 0,
+  "rms_norm_eps": 1e-06,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.29.2",
+  "use_cache": true,
+  "vocab_size": 49954
+}

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": "4.29.2"
+}

pytorch_model-00001-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a84f3830738588b03a50e6a43d9267feee9ec16978a0565920d8d027531a418d
+size 537

pytorch_model-00002-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d80a31c223494cc2e29ba5b884504bc30cca9f2ffacf48b7cef1dc8ca6dff868
+size 409224106

pytorch_model-00003-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ebf968755d743d29e46d0c3a2cb1d88974efd8bdce4964ea6ff1b821b6cdacb3
+size 314575824

pytorch_model-00004-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c053da160b7d33c4df1c001675142a4035c318d748fad0844dc085b472ab71d7
+size 314592818

pytorch_model-00005-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:548ee4d16bcbfbc03138d2ea5875e75c280585d6baddb2837c958cf07064930e
+size 314592818

pytorch_model-00006-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57e74ac537b2995f57304fbf0ebef9f940f0f445212de6e067e74e6a26799edb
+size 371215457

pytorch_model-00007-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e963fa50e10196a410a051a07993d4230b6bc97331291d16cafbbe349947cd7
+size 371215922

pytorch_model-00008-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6457278ba11c7a2092fabde96773f5854e73ddb0bbae51c112e4338be9f33b87
+size 371215922

pytorch_model-00009-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0825143fce89486635a078001a332ebb6c3c0360e8b37cbc69dfaea23471767f
+size 371215922

pytorch_model-00010-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d195e332ded61fb82838852a044dc193dec49970907f1779270749b05c7ac0fc
+size 314575824

pytorch_model-00011-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fcf50a49a848534aff6b4c81d97f3092c8b1e41e93c14f7641998dfc5cfa7361
+size 314592818

pytorch_model-00012-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50e35c94ff9ba6607d4d3575dd1881c7aaff9ce3aa352c8c5c1194494642aee3
+size 314592818

pytorch_model-00013-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d642593c30ccd6a1966a3354ab6714c7ef9cf2587a99162b338f2e0815480e1d
+size 371215457

pytorch_model-00014-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1210e241064e524683f5b5482f32c53ac74677f794ee0c7fdfdf1588bc5ea415
+size 371215922

pytorch_model-00015-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3992bfbe60c3d25499141820c9745ab074244054cbf673c53c089e4b5ab29b8
+size 371215922

pytorch_model-00016-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97b5d2f5afc989f8fbe34a63a2ee128fe4cd34a3928a81fcf051976e41641e84
+size 371215922

pytorch_model-00017-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2457c77fa53d5fec868e1da522708e872490a2dc76c60d9066ae516cfa99186
+size 314575824

pytorch_model-00018-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1fb634e0797a9c5faf6c6e80145b537a4cf407638824b70ef53839b2b9341bb
+size 314592818

pytorch_model-00019-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1014faee86735930f2d94368c6a30a7b19bb9f55079d2d74483bc9ea94684112
+size 314592818

pytorch_model-00020-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b6296ce3ec1d589af1d30f6ee32c3342032e722e5e452948e45e91a6e55d88a
+size 371215521

pytorch_model-00021-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9476926489c7ee8e15f51d5b53ceba25d0d1a11fe0cc793dd434c300cd9d464f
+size 371215922

pytorch_model-00022-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2927bb3db88f53a8c3146550cb0931252f6b0f4c584b06463e7460e584a371ef
+size 371215922

pytorch_model-00023-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3cf68182290aa8db684d9f0a55451bb2401937444ab1678b7e899c56c9cc2d12
+size 371215922

pytorch_model-00024-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:237a031f46333ef625529ab9f4739586bc3417a95ea6e9cd52da861684c46f2c
+size 314575824

pytorch_model-00025-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d05a28ef4cc6fdc3359a016f30ad37153481ed8ccea7d16c588c89db77d41e4f
+size 314592818

pytorch_model-00026-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fccf5c3d15a32c235de2be1cf131da02426e9fffb051bf1d89c4cee020766a14
+size 314592818

pytorch_model-00027-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c8ca0e59684cded44b667fb1776df77a85cb5bf6b3b44f34497b2ea5d9340a5
+size 371215521

pytorch_model-00028-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef39c0f49d9d23f7a13550951a3386816d27bdf705125c80bc0f51eccf3cac99
+size 371215922

pytorch_model-00029-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5b931563d9e3a0adbab4d450dab22bec906f0dfd418aceed422d78fbd5ab756
+size 371215922

pytorch_model-00030-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:291d0568cf1cb09f479625aa56edfd2abba08c9fbae6fe4845ac605bd85d9049
+size 371215922

pytorch_model-00031-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09bd9f227c79edf635dde0aee9fc154e6b8b1d1e9bcd0939b5420d777b96ad29
+size 314575824

pytorch_model-00032-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e66d1c2ad6a28fc8f4f9bddae1dd913a5fcf40b3cfff36db99a759be8e3f5093
+size 314592818

pytorch_model-00033-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b0dc5bfddaf3d997f46f9a0914cce5cec589eb34d5dedf57156a6967efbd086
+size 314592818

pytorch_model-00034-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c6562e1eaedf4f0a94642a45e2a3bc4f0249f07df5f40862c44e7bd3021391d
+size 371215521

pytorch_model-00035-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5b6be5d5b666c7e1ce6f662f7c64f15a089b426df11894c5dc15a7be174d982
+size 371215922

pytorch_model-00036-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:384c45f090245d37103911909c595780d42ba6233fdf9b9452f3de1bdce4bc51
+size 371215922

pytorch_model-00037-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f43f05556abc63f63ce866cc24dbaeb06434e0a41d5a5ae20ee451c348d61081
+size 371215922

pytorch_model-00038-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:adc477cf150e7e6ce6b28177a5b7c4b72730f98febca50bbbc4ec997fb4e5e8f
+size 314575824

pytorch_model-00039-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:827066494120a479e01709fba05099acf8b094ed3250939960067520d8152b60
+size 314592818

pytorch_model-00040-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:287882961f76926aa762ccac822f740ae93f8de252731a96b1ef9a4f8aa0067a
+size 180381870

pytorch_model-00041-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:622d58ba6013453551434b25b02827363c30ed4a2dc91aaaa1af0722c804fb17
+size 409224106

pytorch_model.bin.index.json

@@ -0,0 +1,330 @@
+{
+  "metadata": {
+    "total_size": 13770993664
+  },
+  "weight_map": {
+    "lm_head.weight": "pytorch_model-00041-of-00041.bin",
+    "model.embed_tokens.weight": "pytorch_model-00002-of-00041.bin",
+    "model.layers.0.input_layernorm.weight": "pytorch_model-00004-of-00041.bin",
+    "model.layers.0.mlp.down_proj.weight": "pytorch_model-00003-of-00041.bin",
+    "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00003-of-00041.bin",
+    "model.layers.0.mlp.up_proj.weight": "pytorch_model-00004-of-00041.bin",
+    "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00004-of-00041.bin",
+    "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00003-of-00041.bin",
+    "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00003-of-00041.bin",
+    "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00003-of-00041.bin",
+    "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00041.bin",
+    "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00003-of-00041.bin",
+    "model.layers.1.input_layernorm.weight": "pytorch_model-00005-of-00041.bin",
+    "model.layers.1.mlp.down_proj.weight": "pytorch_model-00005-of-00041.bin",
+    "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00004-of-00041.bin",
+    "model.layers.1.mlp.up_proj.weight": "pytorch_model-00005-of-00041.bin",
+    "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00005-of-00041.bin",
+    "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00004-of-00041.bin",
+    "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00004-of-00041.bin",
+    "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00004-of-00041.bin",
+    "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00041.bin",
+    "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00004-of-00041.bin",
+    "model.layers.10.input_layernorm.weight": "pytorch_model-00015-of-00041.bin",
+    "model.layers.10.mlp.down_proj.weight": "pytorch_model-00015-of-00041.bin",
+    "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00015-of-00041.bin",
+    "model.layers.10.mlp.up_proj.weight": "pytorch_model-00015-of-00041.bin",
+    "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00015-of-00041.bin",
+    "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00014-of-00041.bin",
+    "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00015-of-00041.bin",
+    "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00014-of-00041.bin",
+    "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00015-of-00041.bin",
+    "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00015-of-00041.bin",
+    "model.layers.11.input_layernorm.weight": "pytorch_model-00016-of-00041.bin",
+    "model.layers.11.mlp.down_proj.weight": "pytorch_model-00016-of-00041.bin",
+    "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00016-of-00041.bin",
+    "model.layers.11.mlp.up_proj.weight": "pytorch_model-00016-of-00041.bin",
+    "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00016-of-00041.bin",
+    "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00016-of-00041.bin",
+    "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00016-of-00041.bin",
+    "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00015-of-00041.bin",
+    "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00016-of-00041.bin",
+    "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00016-of-00041.bin",
+    "model.layers.12.input_layernorm.weight": "pytorch_model-00018-of-00041.bin",
+    "model.layers.12.mlp.down_proj.weight": "pytorch_model-00017-of-00041.bin",
+    "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00017-of-00041.bin",
+    "model.layers.12.mlp.up_proj.weight": "pytorch_model-00018-of-00041.bin",
+    "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00018-of-00041.bin",
+    "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00017-of-00041.bin",
+    "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00017-of-00041.bin",
+    "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00017-of-00041.bin",
+    "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00017-of-00041.bin",
+    "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00017-of-00041.bin",
+    "model.layers.13.input_layernorm.weight": "pytorch_model-00019-of-00041.bin",
+    "model.layers.13.mlp.down_proj.weight": "pytorch_model-00019-of-00041.bin",
+    "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00018-of-00041.bin",
+    "model.layers.13.mlp.up_proj.weight": "pytorch_model-00019-of-00041.bin",
+    "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00019-of-00041.bin",
+    "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00018-of-00041.bin",
+    "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00018-of-00041.bin",
+    "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00018-of-00041.bin",
+    "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00018-of-00041.bin",
+    "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00018-of-00041.bin",
+    "model.layers.14.input_layernorm.weight": "pytorch_model-00020-of-00041.bin",
+    "model.layers.14.mlp.down_proj.weight": "pytorch_model-00020-of-00041.bin",
+    "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00020-of-00041.bin",
+    "model.layers.14.mlp.up_proj.weight": "pytorch_model-00020-of-00041.bin",
+    "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00020-of-00041.bin",
+    "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00019-of-00041.bin",
+    "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00019-of-00041.bin",
+    "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00019-of-00041.bin",
+    "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00019-of-00041.bin",
+    "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00019-of-00041.bin",
+    "model.layers.15.input_layernorm.weight": "pytorch_model-00021-of-00041.bin",
+    "model.layers.15.mlp.down_proj.weight": "pytorch_model-00021-of-00041.bin",
+    "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00021-of-00041.bin",
+    "model.layers.15.mlp.up_proj.weight": "pytorch_model-00021-of-00041.bin",
+    "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00021-of-00041.bin",
+    "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00020-of-00041.bin",
+    "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00021-of-00041.bin",
+    "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00020-of-00041.bin",
+    "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00021-of-00041.bin",
+    "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00020-of-00041.bin",
+    "model.layers.16.input_layernorm.weight": "pytorch_model-00022-of-00041.bin",
+    "model.layers.16.mlp.down_proj.weight": "pytorch_model-00022-of-00041.bin",
+    "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00022-of-00041.bin",
+    "model.layers.16.mlp.up_proj.weight": "pytorch_model-00022-of-00041.bin",
+    "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00022-of-00041.bin",
+    "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00021-of-00041.bin",
+    "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00022-of-00041.bin",
+    "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00021-of-00041.bin",
+    "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00022-of-00041.bin",
+    "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00022-of-00041.bin",
+    "model.layers.17.input_layernorm.weight": "pytorch_model-00023-of-00041.bin",
+    "model.layers.17.mlp.down_proj.weight": "pytorch_model-00023-of-00041.bin",
+    "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00023-of-00041.bin",
+    "model.layers.17.mlp.up_proj.weight": "pytorch_model-00023-of-00041.bin",
+    "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00023-of-00041.bin",
+    "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00023-of-00041.bin",
+    "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00023-of-00041.bin",
+    "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00022-of-00041.bin",
+    "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00023-of-00041.bin",
+    "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00023-of-00041.bin",
+    "model.layers.18.input_layernorm.weight": "pytorch_model-00025-of-00041.bin",
+    "model.layers.18.mlp.down_proj.weight": "pytorch_model-00024-of-00041.bin",
+    "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00024-of-00041.bin",
+    "model.layers.18.mlp.up_proj.weight": "pytorch_model-00025-of-00041.bin",
+    "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00025-of-00041.bin",
+    "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00024-of-00041.bin",
+    "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00024-of-00041.bin",
+    "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00024-of-00041.bin",
+    "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00024-of-00041.bin",
+    "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00024-of-00041.bin",
+    "model.layers.19.input_layernorm.weight": "pytorch_model-00026-of-00041.bin",
+    "model.layers.19.mlp.down_proj.weight": "pytorch_model-00026-of-00041.bin",
+    "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00025-of-00041.bin",
+    "model.layers.19.mlp.up_proj.weight": "pytorch_model-00026-of-00041.bin",
+    "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00026-of-00041.bin",
+    "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00025-of-00041.bin",
+    "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00025-of-00041.bin",
+    "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00025-of-00041.bin",
+    "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00025-of-00041.bin",
+    "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00025-of-00041.bin",
+    "model.layers.2.input_layernorm.weight": "pytorch_model-00006-of-00041.bin",
+    "model.layers.2.mlp.down_proj.weight": "pytorch_model-00006-of-00041.bin",
+    "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00006-of-00041.bin",
+    "model.layers.2.mlp.up_proj.weight": "pytorch_model-00006-of-00041.bin",
+    "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00006-of-00041.bin",
+    "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00005-of-00041.bin",
+    "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00005-of-00041.bin",
+    "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00005-of-00041.bin",
+    "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00041.bin",
+    "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00005-of-00041.bin",
+    "model.layers.20.input_layernorm.weight": "pytorch_model-00027-of-00041.bin",
+    "model.layers.20.mlp.down_proj.weight": "pytorch_model-00027-of-00041.bin",
+    "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00027-of-00041.bin",
+    "model.layers.20.mlp.up_proj.weight": "pytorch_model-00027-of-00041.bin",
+    "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00027-of-00041.bin",
+    "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00026-of-00041.bin",
+    "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00026-of-00041.bin",
+    "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00026-of-00041.bin",
+    "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00026-of-00041.bin",
+    "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00026-of-00041.bin",
+    "model.layers.21.input_layernorm.weight": "pytorch_model-00028-of-00041.bin",
+    "model.layers.21.mlp.down_proj.weight": "pytorch_model-00028-of-00041.bin",
+    "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00028-of-00041.bin",
+    "model.layers.21.mlp.up_proj.weight": "pytorch_model-00028-of-00041.bin",
+    "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00028-of-00041.bin",
+    "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00027-of-00041.bin",
+    "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00028-of-00041.bin",
+    "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00027-of-00041.bin",
+    "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00028-of-00041.bin",
+    "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00027-of-00041.bin",
+    "model.layers.22.input_layernorm.weight": "pytorch_model-00029-of-00041.bin",
+    "model.layers.22.mlp.down_proj.weight": "pytorch_model-00029-of-00041.bin",
+    "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00029-of-00041.bin",
+    "model.layers.22.mlp.up_proj.weight": "pytorch_model-00029-of-00041.bin",
+    "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00029-of-00041.bin",
+    "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00028-of-00041.bin",
+    "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00029-of-00041.bin",
+    "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00028-of-00041.bin",
+    "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00029-of-00041.bin",
+    "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00029-of-00041.bin",
+    "model.layers.23.input_layernorm.weight": "pytorch_model-00030-of-00041.bin",
+    "model.layers.23.mlp.down_proj.weight": "pytorch_model-00030-of-00041.bin",
+    "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00030-of-00041.bin",
+    "model.layers.23.mlp.up_proj.weight": "pytorch_model-00030-of-00041.bin",
+    "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00030-of-00041.bin",
+    "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00030-of-00041.bin",
+    "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00030-of-00041.bin",
+    "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00029-of-00041.bin",
+    "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00030-of-00041.bin",
+    "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00030-of-00041.bin",
+    "model.layers.24.input_layernorm.weight": "pytorch_model-00032-of-00041.bin",
+    "model.layers.24.mlp.down_proj.weight": "pytorch_model-00031-of-00041.bin",
+    "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00031-of-00041.bin",
+    "model.layers.24.mlp.up_proj.weight": "pytorch_model-00032-of-00041.bin",
+    "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00032-of-00041.bin",
+    "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00031-of-00041.bin",
+    "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00031-of-00041.bin",
+    "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00031-of-00041.bin",
+    "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00031-of-00041.bin",
+    "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00031-of-00041.bin",
+    "model.layers.25.input_layernorm.weight": "pytorch_model-00033-of-00041.bin",
+    "model.layers.25.mlp.down_proj.weight": "pytorch_model-00033-of-00041.bin",
+    "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00032-of-00041.bin",
+    "model.layers.25.mlp.up_proj.weight": "pytorch_model-00033-of-00041.bin",
+    "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00033-of-00041.bin",
+    "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00032-of-00041.bin",
+    "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00032-of-00041.bin",
+    "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00032-of-00041.bin",
+    "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00032-of-00041.bin",
+    "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00032-of-00041.bin",
+    "model.layers.26.input_layernorm.weight": "pytorch_model-00034-of-00041.bin",
+    "model.layers.26.mlp.down_proj.weight": "pytorch_model-00034-of-00041.bin",
+    "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00034-of-00041.bin",
+    "model.layers.26.mlp.up_proj.weight": "pytorch_model-00034-of-00041.bin",
+    "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00034-of-00041.bin",
+    "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00033-of-00041.bin",
+    "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00033-of-00041.bin",
+    "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00033-of-00041.bin",
+    "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00033-of-00041.bin",
+    "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00033-of-00041.bin",
+    "model.layers.27.input_layernorm.weight": "pytorch_model-00035-of-00041.bin",
+    "model.layers.27.mlp.down_proj.weight": "pytorch_model-00035-of-00041.bin",
+    "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00035-of-00041.bin",
+    "model.layers.27.mlp.up_proj.weight": "pytorch_model-00035-of-00041.bin",
+    "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00035-of-00041.bin",
+    "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00034-of-00041.bin",
+    "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00035-of-00041.bin",
+    "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00034-of-00041.bin",
+    "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00035-of-00041.bin",
+    "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00034-of-00041.bin",
+    "model.layers.28.input_layernorm.weight": "pytorch_model-00036-of-00041.bin",
+    "model.layers.28.mlp.down_proj.weight": "pytorch_model-00036-of-00041.bin",
+    "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00036-of-00041.bin",
+    "model.layers.28.mlp.up_proj.weight": "pytorch_model-00036-of-00041.bin",
+    "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00036-of-00041.bin",
+    "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00035-of-00041.bin",
+    "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00036-of-00041.bin",
+    "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00035-of-00041.bin",
+    "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00036-of-00041.bin",
+    "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00036-of-00041.bin",
+    "model.layers.29.input_layernorm.weight": "pytorch_model-00037-of-00041.bin",
+    "model.layers.29.mlp.down_proj.weight": "pytorch_model-00037-of-00041.bin",
+    "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00037-of-00041.bin",
+    "model.layers.29.mlp.up_proj.weight": "pytorch_model-00037-of-00041.bin",
+    "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00037-of-00041.bin",
+    "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00037-of-00041.bin",
+    "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00037-of-00041.bin",
+    "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00036-of-00041.bin",
+    "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00037-of-00041.bin",
+    "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00037-of-00041.bin",
+    "model.layers.3.input_layernorm.weight": "pytorch_model-00007-of-00041.bin",
+    "model.layers.3.mlp.down_proj.weight": "pytorch_model-00007-of-00041.bin",
+    "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00007-of-00041.bin",
+    "model.layers.3.mlp.up_proj.weight": "pytorch_model-00007-of-00041.bin",
+    "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00007-of-00041.bin",
+    "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00006-of-00041.bin",
+    "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00007-of-00041.bin",
+    "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00006-of-00041.bin",
+    "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00041.bin",
+    "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00006-of-00041.bin",
+    "model.layers.30.input_layernorm.weight": "pytorch_model-00039-of-00041.bin",
+    "model.layers.30.mlp.down_proj.weight": "pytorch_model-00038-of-00041.bin",
+    "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00038-of-00041.bin",
+    "model.layers.30.mlp.up_proj.weight": "pytorch_model-00039-of-00041.bin",
+    "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00039-of-00041.bin",
+    "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00038-of-00041.bin",
+    "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00038-of-00041.bin",
+    "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00038-of-00041.bin",
+    "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00038-of-00041.bin",
+    "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00038-of-00041.bin",
+    "model.layers.31.input_layernorm.weight": "pytorch_model-00040-of-00041.bin",
+    "model.layers.31.mlp.down_proj.weight": "pytorch_model-00040-of-00041.bin",
+    "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00039-of-00041.bin",
+    "model.layers.31.mlp.up_proj.weight": "pytorch_model-00040-of-00041.bin",
+    "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00040-of-00041.bin",
+    "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00039-of-00041.bin",
+    "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00039-of-00041.bin",
+    "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00039-of-00041.bin",
+    "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00039-of-00041.bin",
+    "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00039-of-00041.bin",
+    "model.layers.4.input_layernorm.weight": "pytorch_model-00008-of-00041.bin",
+    "model.layers.4.mlp.down_proj.weight": "pytorch_model-00008-of-00041.bin",
+    "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00008-of-00041.bin",
+    "model.layers.4.mlp.up_proj.weight": "pytorch_model-00008-of-00041.bin",
+    "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00008-of-00041.bin",
+    "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00007-of-00041.bin",
+    "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00008-of-00041.bin",
+    "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00007-of-00041.bin",
+    "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00008-of-00041.bin",
+    "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00008-of-00041.bin",
+    "model.layers.5.input_layernorm.weight": "pytorch_model-00009-of-00041.bin",
+    "model.layers.5.mlp.down_proj.weight": "pytorch_model-00009-of-00041.bin",
+    "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00009-of-00041.bin",
+    "model.layers.5.mlp.up_proj.weight": "pytorch_model-00009-of-00041.bin",
+    "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00009-of-00041.bin",
+    "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00009-of-00041.bin",
+    "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00009-of-00041.bin",
+    "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00008-of-00041.bin",
+    "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00009-of-00041.bin",
+    "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00009-of-00041.bin",
+    "model.layers.6.input_layernorm.weight": "pytorch_model-00011-of-00041.bin",
+    "model.layers.6.mlp.down_proj.weight": "pytorch_model-00010-of-00041.bin",
+    "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00010-of-00041.bin",
+    "model.layers.6.mlp.up_proj.weight": "pytorch_model-00011-of-00041.bin",
+    "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00011-of-00041.bin",
+    "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00010-of-00041.bin",
+    "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00010-of-00041.bin",
+    "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00010-of-00041.bin",
+    "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00010-of-00041.bin",
+    "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00010-of-00041.bin",
+    "model.layers.7.input_layernorm.weight": "pytorch_model-00012-of-00041.bin",
+    "model.layers.7.mlp.down_proj.weight": "pytorch_model-00012-of-00041.bin",
+    "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00011-of-00041.bin",
+    "model.layers.7.mlp.up_proj.weight": "pytorch_model-00012-of-00041.bin",
+    "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00012-of-00041.bin",
+    "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00011-of-00041.bin",
+    "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00011-of-00041.bin",
+    "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00011-of-00041.bin",
+    "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00011-of-00041.bin",
+    "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00011-of-00041.bin",
+    "model.layers.8.input_layernorm.weight": "pytorch_model-00013-of-00041.bin",
+    "model.layers.8.mlp.down_proj.weight": "pytorch_model-00013-of-00041.bin",
+    "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00013-of-00041.bin",
+    "model.layers.8.mlp.up_proj.weight": "pytorch_model-00013-of-00041.bin",
+    "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00013-of-00041.bin",
+    "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00012-of-00041.bin",
+    "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00012-of-00041.bin",
+    "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00012-of-00041.bin",
+    "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00012-of-00041.bin",
+    "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00012-of-00041.bin",
+    "model.layers.9.input_layernorm.weight": "pytorch_model-00014-of-00041.bin",
+    "model.layers.9.mlp.down_proj.weight": "pytorch_model-00014-of-00041.bin",
+    "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00014-of-00041.bin",
+    "model.layers.9.mlp.up_proj.weight": "pytorch_model-00014-of-00041.bin",
+    "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00014-of-00041.bin",
+    "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00013-of-00041.bin",
+    "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00014-of-00041.bin",
+    "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00013-of-00041.bin",
+    "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00014-of-00041.bin",
+    "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00013-of-00041.bin",
+    "model.norm.weight": "pytorch_model-00040-of-00041.bin"
+  }
+}