Initial Version

config.json

@@ -0,0 +1,26 @@
+{
+  "_name_or_path": "Adun/openthaigpt-1.0.0-beta-7b-ckpt-hf",
+  "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": 4096,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": null,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.28.0",
+  "use_cache": true,
+  "vocab_size": 56554
+}

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": "4.28.0"
+}

pytorch_model-00001-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01907711dd7ff2d3b41140a8db70515479b64ee6bfff82bade4c876d7f099ee5
+size 1012

pytorch_model-00002-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9d78d02e264931d8bca9572894d2b9b6dcc5cd8bf164c337b67e557477e7d1e
+size 463291781

pytorch_model-00003-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d0c24a15d2f9208cae06434cc08c8c1ffb7ba8828e513c87e438503aebb198b
+size 314576427

pytorch_model-00004-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:433e7402f3eeb9e409504cc68a71069e5bbd565f7feb07a2b3a9f32a0a225f45
+size 314593485

pytorch_model-00005-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:369c5094693a3f18e35cb7c26b8c8ba03b18f2b93beedf4ca92e50038d79427c
+size 314593485

pytorch_model-00006-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af59eee094561ce820da5754453ce40c91a4bfb593def6839776e89fd660e7e5
+size 371215932

pytorch_model-00007-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ce0ce0b1db5ea6747e6bcd2e8b3a784584fff0a1b03bcbe5d40694784224523
+size 371216525

pytorch_model-00008-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0dda0cb74a5f6ec5a06029331442f139f0f998da6ebce43cf5c5dfecca6822b9
+size 371216525

pytorch_model-00009-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62663e55200a644b809558428528c40cbf4dbeaff5291ca39b7fe0e3cb53b3f5
+size 371216525

pytorch_model-00010-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d88a3c9fbb42fc8ef245d5ff099f7ca895a0fcc97df9ac7b4202ab10237400dd
+size 314576427

pytorch_model-00011-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4f5f86b1ca674675dbe4a8744e47ae4558e004b0cc8f0ab917f24def4f3a2b9
+size 314593485

pytorch_model-00012-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2913e1e8d520a84f3ea1f30409df37664eb6a36eb6edb15f6d42949e63898de5
+size 314593485

pytorch_model-00013-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0492428be15eb1327c3b9f6836b621832c5508381052c3f5158e8d12d1e3d1da
+size 371215932

pytorch_model-00014-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2619dfd6e28378d300ebb22c112904099a8dbf3b1073cfb42b15de67221a33fe
+size 371216589

pytorch_model-00015-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50dbe5c4baaeb1df2eb8e2355ca5dc777f85a2084f220f713468dc122d67f010
+size 371216589

pytorch_model-00016-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4099b837212e5ca0d94aafc2add9040c3cbe3579a48e4f4d97597cf38bef146
+size 371216589

pytorch_model-00017-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64f98906b509827b2739e64873d1986736de52081c7b8a80d9c7b0fbcb36cbc6
+size 314576427

pytorch_model-00018-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8418025ecdb38867098b4665993ea04d16b4d67a2141ceffe6410fa06fdcdda
+size 314593485

pytorch_model-00019-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55355a29d3417bdae55f2cfd52a2a5fde59d16baa1908ca159944cf241105e85
+size 314593485

pytorch_model-00020-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bcf3651c25e46439d18e21162960465164c86c3c6c00a7568495734db1fce748
+size 371215996

pytorch_model-00021-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54fd2e39055222595acb764e8ea56cdac595def495269051fddbd7152903d284
+size 371216589

pytorch_model-00022-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ae5045ad2b0707684cf7dcc5d3329c3454c0a97a0f2e0ae3992129ecf4fe76f
+size 371216589

pytorch_model-00023-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8073cd6c0f377c2b8004e797dbcbbf75a57e6b60c4cb331f3fa7f4bf4bff401e
+size 371216589

pytorch_model-00024-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e1e594522c6cbe95ef4cbdd2f3d09a137ba01d26e050715ff09409fdc05d794
+size 314576427

pytorch_model-00025-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:309f2664dc276de99038fa69a9293acf75c5b7d214a52f7600dd0597b0a5f4f6
+size 314593485

pytorch_model-00026-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eca2c786a63e531bcb4329945caab3470192e43ba303a75dbf28f8e95e9b5b33
+size 314593485

pytorch_model-00027-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67b1281bc60c7174becd0603b59efa378fcad7f2d2f94375ca32d49c29f25fde
+size 371215996

pytorch_model-00028-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2fd17e979894b5f05145e465f961731b01db907be06c51e4017fa9ea0a0abb68
+size 371216589

pytorch_model-00029-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0573f8fc059beab40b2e9fe8c8229136c8899f67aa8dd2f722453d84c62a0423
+size 371216589

pytorch_model-00030-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91eb40f70f3cc5d9ac2299129bcf968715576b0d11d83c5b947b615a486af668
+size 371216589

pytorch_model-00031-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bc97428ec766f01007ce85afded6f1bfc767f4a7b7d13cabb245fd40c1c2afe
+size 314576427

pytorch_model-00032-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e5f2706d5f92e3da5f23a4850bebd6d3765cae430d19d0796d7df5adbee6e07
+size 314593485

pytorch_model-00033-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce2e86fe25797eb1b4626c510beffce9215ceb112c5026e22a8d05bc8a5b84c3
+size 314593485

pytorch_model-00034-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b20e154e7c90ccf3d1e4b19c44462373cf5f9f0b922363db59644d799fc72411
+size 371215996

pytorch_model-00035-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b59bef4cebee15d2c2c7ec8a8094e6861caa5d2dc76785be0c8226286cb7ccc
+size 371216589

pytorch_model-00036-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1e82881dfb2400f539e57c1413fd9caafe6eaafec1f60d2c32cefae8b07010b
+size 371216589

pytorch_model-00037-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e258a552642459f80ebb8c59a8c0c855e2cd51ca6db08c1994b075b505e37e90
+size 371216589

pytorch_model-00038-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c654f8c68fd8988d5cafa7a26b468848838d185e4d659700941a2c887fb7bf4e
+size 314576427

pytorch_model-00039-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:84ac26ded4b9a2fe245276f070315806718e800612996236362565826efefa5c
+size 314593485

pytorch_model-00040-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8d88dbd2118e5b909baf85428e8a758add045cbbff3dc528eeffeed5cd8d5ff
+size 180382345

pytorch_model-00041-of-00041.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1663a024637ae7aba6a0bf077b714f13f0c852dc8c7e31cb7385b5a49310f84
+size 463291781

pytorch_model.bin.index.json

@@ -0,0 +1,330 @@
+{
+  "metadata": {
+    "total_size": 13879132160
+  },
+  "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"
+  }
+}