Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

config.json

@@ -0,0 +1,28 @@
+{
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 8192,
+  "initializer_range": 0.02,
+  "intermediate_size": 29568,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 70,
+  "model_type": "qwen3",
+  "num_attention_heads": 64,
+  "num_hidden_layers": 80,
+  "num_key_value_heads": 8,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": 131072,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.51.3",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

config_diff.json

@@ -0,0 +1,41 @@
+{
+  "changed": {
+    "model_type": {
+      "from": "qwen2",
+      "to": "qwen3"
+    },
+    "vocab_size": {
+      "from": 152064,
+      "to": 151936
+    },
+    "_name_or_path": {
+      "from": "Qwen/Qwen2.5-72B-Instruct",
+      "to": ""
+    },
+    "bos_token_id": {
+      "from": 151643,
+      "to": null
+    },
+    "tie_word_embeddings": {
+      "from": false,
+      "to": true
+    },
+    "eos_token_id": {
+      "from": 151645,
+      "to": null
+    },
+    "architectures": {
+      "from": [
+        "Qwen2ForCausalLM"
+      ],
+      "to": [
+        "Qwen3ForCausalLM"
+      ]
+    }
+  },
+  "added": {
+    "attention_bias": false,
+    "head_dim": 128
+  },
+  "removed": {}
+}

generation_config.json

@@ -0,0 +1,4 @@
+{
+  "_from_model_config": true,
+  "transformers_version": "4.51.3"
+}

model-00001-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5c87bd7774dc08a98b2c164cfaca0996191acbc3c26540247d124e15497a73a
+size 4546661424

model-00002-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ffcd7ab15b09139a24d7654b21cd4b9c69278763e352aad92f969cd46642f5fa
+size 4964061232

model-00003-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:270771240d32d0ba80eef822b5e2dae77d93caea1f84490a5785a7f8525e2b2d
+size 4781577096

model-00004-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a1b72f211aaaa2ffa7c16ea1a0d37a256bdcb0ba415289f323fcd760270d7d9
+size 4781610096

model-00005-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9cfb54e3de8f1ae5a64151cfa7c25a6e06019a34063423b87ad5aed8d79beb9d
+size 4781610128

model-00006-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7ed65d7eca8c82a4791db1b75832aa9fe373b75f09b1b35f8cddafb89f4e036
+size 4964061264

model-00007-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e31f78b85a609f805ebe4290cf9d750456daa5ffc0a7d5a41dd97ddf2a964a9d
+size 4781577128

model-00008-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38c67c566ea3bbf10c08d2df63afb11acd5cf39ee00b7e9484dacf0f0e79710a
+size 4781610128

model-00009-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:31f8003f72913bd3bacbaaf8383be93b7f40a2f7e5f6190d325e44c3af1c2e02
+size 4781610128

model-00010-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10bd214214221ce8295fc1687d2d0fccbd6d11e9946f0d05ad4915c3ff2f4254
+size 4964061264

model-00011-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a46f1a789145e670afd86bf61334cb45d03e2fcd70d6a5c7d2a3a40deb60216
+size 4781577128

model-00012-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d7c6b95eaf5b23a33355425b3df730be9e95927975d5b3e4be2cdd79502586bf
+size 4781610128

model-00013-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:200c159cc8f4806a0f389f6cb8e82abba3dcf7914bf2681334e60a2b1c3cfe31
+size 4781610128

model-00014-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1c9a27aaa954d10fe2998fe3d9fcd2c0eab000eca497159d58c63d9fa10ddc3
+size 4964061264

model-00015-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:173c88345a156c20e200be235556edb4e29495c7075196f4443ca9de1287916c
+size 4781577128

model-00016-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3074880772c728570a99e0286417ea440d6bd9c5795ef60328531deeaf41870
+size 4781610128

model-00017-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c27b08d5d44471c784e33f2944d5b69e8b950bd1822e23165f75520410802a1
+size 4781610128

model-00018-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a231baedea1b49219c450bdc95207f4b555ff7599a14133f0fc27d7340817a4
+size 4964061264

model-00019-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63065018deb37f61e72410afce3484592d4af3951c2ebb8251628e94eb9c6e4f
+size 4781577128

model-00020-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb1225790f1ba1da0bbe2a24f896240f610d75faa23703ec9ef2cb4b1a362cbc
+size 4781610128

model-00021-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e850aaea0846d305c8a6abf544a4f63c0c35fc9028f187f10d93e0b5b636198
+size 4781610128

model-00022-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:222aa197a63dc450d912b6f34bc993ccfb3cf3e637427226aaf164ffa669606e
+size 4964061264

model-00023-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff80f4e57fbb429fe116c66b14043098e51b55480a8e5b9f095148322e000182
+size 4781577128

model-00024-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3265be553586f3be7dc41756370d06bb9059c2b88afbf4e3ce6b3efd0c9c8327
+size 4781610128

model-00025-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a92b55701aee163cfee55e3fb19c798e705390b205fb397c2356897aee4e4c1
+size 4781610128

model-00026-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b80f8c6ac7894fc07526cd0a6248b92265251366f2d8df2dfdabdad048a4ac4
+size 4964061264

model-00027-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af1de5423acb6f99947bbcced145619a673b3412ac16f2283e25b489e7f5ff92
+size 4781577128

model-00028-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69a3c8bfc19b7e94b095aa91c1be5088f07eebe4eb490d9411ba25e7aa081696
+size 4781610128

model-00029-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:574be2d6d1d69ba58c34769798c76ff5f0f0a717d9118b62cdd57e262013a5c4
+size 4781610128

model-00030-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd7a418d62bd41a7487963686b036f14051124a0d66f269c7efc8557b5afd62e
+size 3208726960

model-00031-of-00031.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ac450e7a68b764065b7dde03e3e7e54d64a5fb972567881ff144ce4a7c64220
+size 2489319552

model.safetensors.index.json

@@ -0,0 +1,890 @@
+{
+  "metadata": {
+    "total_size": 145406615552
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00031-of-00031.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00031.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00031.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00031.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00031.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00031.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00031.safetensors",
+    "model.layers.0.self_attn.k_norm.weight": "model-00001-of-00031.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00031.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00031.safetensors",
+    "model.layers.0.self_attn.q_norm.weight": "model-00001-of-00031.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00031.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00031.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00002-of-00031.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00002-of-00031.safetensors",
+    "model.layers.1.self_attn.k_norm.weight": "model-00001-of-00031.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00031.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00031.safetensors",
+    "model.layers.1.self_attn.q_norm.weight": "model-00001-of-00031.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00031.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00031.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00005-of-00031.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00005-of-00031.safetensors",
+    "model.layers.10.self_attn.k_norm.weight": "model-00005-of-00031.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.10.self_attn.q_norm.weight": "model-00005-of-00031.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00005-of-00031.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00005-of-00031.safetensors",
+    "model.layers.11.self_attn.k_norm.weight": "model-00005-of-00031.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.11.self_attn.q_norm.weight": "model-00005-of-00031.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00006-of-00031.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00006-of-00031.safetensors",
+    "model.layers.12.self_attn.k_norm.weight": "model-00005-of-00031.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.12.self_attn.q_norm.weight": "model-00005-of-00031.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00006-of-00031.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00006-of-00031.safetensors",
+    "model.layers.13.self_attn.k_norm.weight": "model-00006-of-00031.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.13.self_attn.q_norm.weight": "model-00006-of-00031.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00006-of-00031.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00006-of-00031.safetensors",
+    "model.layers.14.self_attn.k_norm.weight": "model-00006-of-00031.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.14.self_attn.q_norm.weight": "model-00006-of-00031.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00006-of-00031.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00007-of-00031.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00007-of-00031.safetensors",
+    "model.layers.15.self_attn.k_norm.weight": "model-00007-of-00031.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.15.self_attn.q_norm.weight": "model-00007-of-00031.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00007-of-00031.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00007-of-00031.safetensors",
+    "model.layers.16.self_attn.k_norm.weight": "model-00007-of-00031.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.16.self_attn.q_norm.weight": "model-00007-of-00031.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00008-of-00031.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00008-of-00031.safetensors",
+    "model.layers.17.self_attn.k_norm.weight": "model-00007-of-00031.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.17.self_attn.q_norm.weight": "model-00007-of-00031.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00007-of-00031.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00008-of-00031.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00008-of-00031.safetensors",
+    "model.layers.18.self_attn.k_norm.weight": "model-00008-of-00031.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.18.self_attn.q_norm.weight": "model-00008-of-00031.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00008-of-00031.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00008-of-00031.safetensors",
+    "model.layers.19.self_attn.k_norm.weight": "model-00008-of-00031.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.19.self_attn.q_norm.weight": "model-00008-of-00031.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00002-of-00031.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00002-of-00031.safetensors",
+    "model.layers.2.self_attn.k_norm.weight": "model-00002-of-00031.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.2.self_attn.q_norm.weight": "model-00002-of-00031.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00009-of-00031.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00009-of-00031.safetensors",
+    "model.layers.20.self_attn.k_norm.weight": "model-00008-of-00031.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.20.self_attn.q_norm.weight": "model-00008-of-00031.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00008-of-00031.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00009-of-00031.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00009-of-00031.safetensors",
+    "model.layers.21.self_attn.k_norm.weight": "model-00009-of-00031.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.21.self_attn.q_norm.weight": "model-00009-of-00031.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00009-of-00031.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00009-of-00031.safetensors",
+    "model.layers.22.self_attn.k_norm.weight": "model-00009-of-00031.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.22.self_attn.q_norm.weight": "model-00009-of-00031.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00010-of-00031.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00010-of-00031.safetensors",
+    "model.layers.23.self_attn.k_norm.weight": "model-00009-of-00031.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.23.self_attn.q_norm.weight": "model-00009-of-00031.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00009-of-00031.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00010-of-00031.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00010-of-00031.safetensors",
+    "model.layers.24.self_attn.k_norm.weight": "model-00010-of-00031.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.24.self_attn.q_norm.weight": "model-00010-of-00031.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00010-of-00031.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00010-of-00031.safetensors",
+    "model.layers.25.self_attn.k_norm.weight": "model-00010-of-00031.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.25.self_attn.q_norm.weight": "model-00010-of-00031.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00010-of-00031.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00011-of-00031.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00011-of-00031.safetensors",
+    "model.layers.26.self_attn.k_norm.weight": "model-00011-of-00031.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.26.self_attn.q_norm.weight": "model-00011-of-00031.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00011-of-00031.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00011-of-00031.safetensors",
+    "model.layers.27.self_attn.k_norm.weight": "model-00011-of-00031.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.27.self_attn.q_norm.weight": "model-00011-of-00031.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00012-of-00031.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00012-of-00031.safetensors",
+    "model.layers.28.self_attn.k_norm.weight": "model-00011-of-00031.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.28.self_attn.q_norm.weight": "model-00011-of-00031.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00011-of-00031.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00012-of-00031.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00012-of-00031.safetensors",
+    "model.layers.29.self_attn.k_norm.weight": "model-00012-of-00031.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.29.self_attn.q_norm.weight": "model-00012-of-00031.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00002-of-00031.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00002-of-00031.safetensors",
+    "model.layers.3.self_attn.k_norm.weight": "model-00002-of-00031.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.3.self_attn.q_norm.weight": "model-00002-of-00031.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00002-of-00031.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00012-of-00031.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00012-of-00031.safetensors",
+    "model.layers.30.self_attn.k_norm.weight": "model-00012-of-00031.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.30.self_attn.q_norm.weight": "model-00012-of-00031.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00013-of-00031.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00013-of-00031.safetensors",
+    "model.layers.31.self_attn.k_norm.weight": "model-00012-of-00031.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.31.self_attn.q_norm.weight": "model-00012-of-00031.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00012-of-00031.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00013-of-00031.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00013-of-00031.safetensors",
+    "model.layers.32.self_attn.k_norm.weight": "model-00013-of-00031.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.32.self_attn.q_norm.weight": "model-00013-of-00031.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00013-of-00031.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00013-of-00031.safetensors",
+    "model.layers.33.self_attn.k_norm.weight": "model-00013-of-00031.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.33.self_attn.q_norm.weight": "model-00013-of-00031.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00014-of-00031.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00014-of-00031.safetensors",
+    "model.layers.34.self_attn.k_norm.weight": "model-00013-of-00031.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.34.self_attn.q_norm.weight": "model-00013-of-00031.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00013-of-00031.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00014-of-00031.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00014-of-00031.safetensors",
+    "model.layers.35.self_attn.k_norm.weight": "model-00014-of-00031.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.35.self_attn.q_norm.weight": "model-00014-of-00031.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00014-of-00031.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00014-of-00031.safetensors",
+    "model.layers.36.self_attn.k_norm.weight": "model-00014-of-00031.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.36.self_attn.q_norm.weight": "model-00014-of-00031.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00014-of-00031.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00015-of-00031.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00015-of-00031.safetensors",
+    "model.layers.37.self_attn.k_norm.weight": "model-00015-of-00031.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.37.self_attn.q_norm.weight": "model-00015-of-00031.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00015-of-00031.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00015-of-00031.safetensors",
+    "model.layers.38.self_attn.k_norm.weight": "model-00015-of-00031.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.38.self_attn.q_norm.weight": "model-00015-of-00031.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00016-of-00031.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00016-of-00031.safetensors",
+    "model.layers.39.self_attn.k_norm.weight": "model-00015-of-00031.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.39.self_attn.q_norm.weight": "model-00015-of-00031.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00015-of-00031.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00003-of-00031.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00003-of-00031.safetensors",
+    "model.layers.4.self_attn.k_norm.weight": "model-00003-of-00031.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.4.self_attn.q_norm.weight": "model-00003-of-00031.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00016-of-00031.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00016-of-00031.safetensors",
+    "model.layers.40.self_attn.k_norm.weight": "model-00016-of-00031.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.40.self_attn.q_norm.weight": "model-00016-of-00031.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00016-of-00031.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00016-of-00031.safetensors",
+    "model.layers.41.self_attn.k_norm.weight": "model-00016-of-00031.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.41.self_attn.q_norm.weight": "model-00016-of-00031.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.42.input_layernorm.weight": "model-00017-of-00031.safetensors",
+    "model.layers.42.mlp.down_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.42.mlp.gate_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.42.mlp.up_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.42.post_attention_layernorm.weight": "model-00017-of-00031.safetensors",
+    "model.layers.42.self_attn.k_norm.weight": "model-00016-of-00031.safetensors",
+    "model.layers.42.self_attn.k_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.42.self_attn.o_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.42.self_attn.q_norm.weight": "model-00016-of-00031.safetensors",
+    "model.layers.42.self_attn.q_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.42.self_attn.v_proj.weight": "model-00016-of-00031.safetensors",
+    "model.layers.43.input_layernorm.weight": "model-00017-of-00031.safetensors",
+    "model.layers.43.mlp.down_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.43.mlp.gate_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.43.mlp.up_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.43.post_attention_layernorm.weight": "model-00017-of-00031.safetensors",
+    "model.layers.43.self_attn.k_norm.weight": "model-00017-of-00031.safetensors",
+    "model.layers.43.self_attn.k_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.43.self_attn.o_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.43.self_attn.q_norm.weight": "model-00017-of-00031.safetensors",
+    "model.layers.43.self_attn.q_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.43.self_attn.v_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.44.input_layernorm.weight": "model-00017-of-00031.safetensors",
+    "model.layers.44.mlp.down_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.44.mlp.gate_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.44.mlp.up_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.44.post_attention_layernorm.weight": "model-00017-of-00031.safetensors",
+    "model.layers.44.self_attn.k_norm.weight": "model-00017-of-00031.safetensors",
+    "model.layers.44.self_attn.k_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.44.self_attn.o_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.44.self_attn.q_norm.weight": "model-00017-of-00031.safetensors",
+    "model.layers.44.self_attn.q_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.44.self_attn.v_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.45.input_layernorm.weight": "model-00018-of-00031.safetensors",
+    "model.layers.45.mlp.down_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.45.mlp.gate_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.45.mlp.up_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.45.post_attention_layernorm.weight": "model-00018-of-00031.safetensors",
+    "model.layers.45.self_attn.k_norm.weight": "model-00017-of-00031.safetensors",
+    "model.layers.45.self_attn.k_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.45.self_attn.o_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.45.self_attn.q_norm.weight": "model-00017-of-00031.safetensors",
+    "model.layers.45.self_attn.q_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.45.self_attn.v_proj.weight": "model-00017-of-00031.safetensors",
+    "model.layers.46.input_layernorm.weight": "model-00018-of-00031.safetensors",
+    "model.layers.46.mlp.down_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.46.mlp.gate_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.46.mlp.up_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.46.post_attention_layernorm.weight": "model-00018-of-00031.safetensors",
+    "model.layers.46.self_attn.k_norm.weight": "model-00018-of-00031.safetensors",
+    "model.layers.46.self_attn.k_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.46.self_attn.o_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.46.self_attn.q_norm.weight": "model-00018-of-00031.safetensors",
+    "model.layers.46.self_attn.q_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.46.self_attn.v_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.47.input_layernorm.weight": "model-00018-of-00031.safetensors",
+    "model.layers.47.mlp.down_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.47.mlp.gate_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.47.mlp.up_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.47.post_attention_layernorm.weight": "model-00018-of-00031.safetensors",
+    "model.layers.47.self_attn.k_norm.weight": "model-00018-of-00031.safetensors",
+    "model.layers.47.self_attn.k_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.47.self_attn.o_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.47.self_attn.q_norm.weight": "model-00018-of-00031.safetensors",
+    "model.layers.47.self_attn.q_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.47.self_attn.v_proj.weight": "model-00018-of-00031.safetensors",
+    "model.layers.48.input_layernorm.weight": "model-00019-of-00031.safetensors",
+    "model.layers.48.mlp.down_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.48.mlp.gate_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.48.mlp.up_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.48.post_attention_layernorm.weight": "model-00019-of-00031.safetensors",
+    "model.layers.48.self_attn.k_norm.weight": "model-00019-of-00031.safetensors",
+    "model.layers.48.self_attn.k_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.48.self_attn.o_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.48.self_attn.q_norm.weight": "model-00019-of-00031.safetensors",
+    "model.layers.48.self_attn.q_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.48.self_attn.v_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.49.input_layernorm.weight": "model-00019-of-00031.safetensors",
+    "model.layers.49.mlp.down_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.49.mlp.gate_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.49.mlp.up_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.49.post_attention_layernorm.weight": "model-00019-of-00031.safetensors",
+    "model.layers.49.self_attn.k_norm.weight": "model-00019-of-00031.safetensors",
+    "model.layers.49.self_attn.k_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.49.self_attn.o_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.49.self_attn.q_norm.weight": "model-00019-of-00031.safetensors",
+    "model.layers.49.self_attn.q_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.49.self_attn.v_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00003-of-00031.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00003-of-00031.safetensors",
+    "model.layers.5.self_attn.k_norm.weight": "model-00003-of-00031.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.5.self_attn.q_norm.weight": "model-00003-of-00031.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.50.input_layernorm.weight": "model-00020-of-00031.safetensors",
+    "model.layers.50.mlp.down_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.50.mlp.gate_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.50.mlp.up_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.50.post_attention_layernorm.weight": "model-00020-of-00031.safetensors",
+    "model.layers.50.self_attn.k_norm.weight": "model-00019-of-00031.safetensors",
+    "model.layers.50.self_attn.k_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.50.self_attn.o_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.50.self_attn.q_norm.weight": "model-00019-of-00031.safetensors",
+    "model.layers.50.self_attn.q_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.50.self_attn.v_proj.weight": "model-00019-of-00031.safetensors",
+    "model.layers.51.input_layernorm.weight": "model-00020-of-00031.safetensors",
+    "model.layers.51.mlp.down_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.51.mlp.gate_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.51.mlp.up_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.51.post_attention_layernorm.weight": "model-00020-of-00031.safetensors",
+    "model.layers.51.self_attn.k_norm.weight": "model-00020-of-00031.safetensors",
+    "model.layers.51.self_attn.k_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.51.self_attn.o_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.51.self_attn.q_norm.weight": "model-00020-of-00031.safetensors",
+    "model.layers.51.self_attn.q_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.51.self_attn.v_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.52.input_layernorm.weight": "model-00020-of-00031.safetensors",
+    "model.layers.52.mlp.down_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.52.mlp.gate_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.52.mlp.up_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.52.post_attention_layernorm.weight": "model-00020-of-00031.safetensors",
+    "model.layers.52.self_attn.k_norm.weight": "model-00020-of-00031.safetensors",
+    "model.layers.52.self_attn.k_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.52.self_attn.o_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.52.self_attn.q_norm.weight": "model-00020-of-00031.safetensors",
+    "model.layers.52.self_attn.q_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.52.self_attn.v_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.53.input_layernorm.weight": "model-00021-of-00031.safetensors",
+    "model.layers.53.mlp.down_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.53.mlp.gate_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.53.mlp.up_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.53.post_attention_layernorm.weight": "model-00021-of-00031.safetensors",
+    "model.layers.53.self_attn.k_norm.weight": "model-00020-of-00031.safetensors",
+    "model.layers.53.self_attn.k_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.53.self_attn.o_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.53.self_attn.q_norm.weight": "model-00020-of-00031.safetensors",
+    "model.layers.53.self_attn.q_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.53.self_attn.v_proj.weight": "model-00020-of-00031.safetensors",
+    "model.layers.54.input_layernorm.weight": "model-00021-of-00031.safetensors",
+    "model.layers.54.mlp.down_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.54.mlp.gate_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.54.mlp.up_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.54.post_attention_layernorm.weight": "model-00021-of-00031.safetensors",
+    "model.layers.54.self_attn.k_norm.weight": "model-00021-of-00031.safetensors",
+    "model.layers.54.self_attn.k_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.54.self_attn.o_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.54.self_attn.q_norm.weight": "model-00021-of-00031.safetensors",
+    "model.layers.54.self_attn.q_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.54.self_attn.v_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.55.input_layernorm.weight": "model-00021-of-00031.safetensors",
+    "model.layers.55.mlp.down_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.55.mlp.gate_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.55.mlp.up_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.55.post_attention_layernorm.weight": "model-00021-of-00031.safetensors",
+    "model.layers.55.self_attn.k_norm.weight": "model-00021-of-00031.safetensors",
+    "model.layers.55.self_attn.k_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.55.self_attn.o_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.55.self_attn.q_norm.weight": "model-00021-of-00031.safetensors",
+    "model.layers.55.self_attn.q_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.55.self_attn.v_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.56.input_layernorm.weight": "model-00022-of-00031.safetensors",
+    "model.layers.56.mlp.down_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.56.mlp.gate_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.56.mlp.up_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.56.post_attention_layernorm.weight": "model-00022-of-00031.safetensors",
+    "model.layers.56.self_attn.k_norm.weight": "model-00021-of-00031.safetensors",
+    "model.layers.56.self_attn.k_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.56.self_attn.o_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.56.self_attn.q_norm.weight": "model-00021-of-00031.safetensors",
+    "model.layers.56.self_attn.q_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.56.self_attn.v_proj.weight": "model-00021-of-00031.safetensors",
+    "model.layers.57.input_layernorm.weight": "model-00022-of-00031.safetensors",
+    "model.layers.57.mlp.down_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.57.mlp.gate_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.57.mlp.up_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.57.post_attention_layernorm.weight": "model-00022-of-00031.safetensors",
+    "model.layers.57.self_attn.k_norm.weight": "model-00022-of-00031.safetensors",
+    "model.layers.57.self_attn.k_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.57.self_attn.o_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.57.self_attn.q_norm.weight": "model-00022-of-00031.safetensors",
+    "model.layers.57.self_attn.q_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.57.self_attn.v_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.58.input_layernorm.weight": "model-00022-of-00031.safetensors",
+    "model.layers.58.mlp.down_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.58.mlp.gate_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.58.mlp.up_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.58.post_attention_layernorm.weight": "model-00022-of-00031.safetensors",
+    "model.layers.58.self_attn.k_norm.weight": "model-00022-of-00031.safetensors",
+    "model.layers.58.self_attn.k_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.58.self_attn.o_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.58.self_attn.q_norm.weight": "model-00022-of-00031.safetensors",
+    "model.layers.58.self_attn.q_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.58.self_attn.v_proj.weight": "model-00022-of-00031.safetensors",
+    "model.layers.59.input_layernorm.weight": "model-00023-of-00031.safetensors",
+    "model.layers.59.mlp.down_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.59.mlp.gate_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.59.mlp.up_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.59.post_attention_layernorm.weight": "model-00023-of-00031.safetensors",
+    "model.layers.59.self_attn.k_norm.weight": "model-00023-of-00031.safetensors",
+    "model.layers.59.self_attn.k_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.59.self_attn.o_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.59.self_attn.q_norm.weight": "model-00023-of-00031.safetensors",
+    "model.layers.59.self_attn.q_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.59.self_attn.v_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00004-of-00031.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00004-of-00031.safetensors",
+    "model.layers.6.self_attn.k_norm.weight": "model-00003-of-00031.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.6.self_attn.q_norm.weight": "model-00003-of-00031.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00003-of-00031.safetensors",
+    "model.layers.60.input_layernorm.weight": "model-00023-of-00031.safetensors",
+    "model.layers.60.mlp.down_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.60.mlp.gate_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.60.mlp.up_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.60.post_attention_layernorm.weight": "model-00023-of-00031.safetensors",
+    "model.layers.60.self_attn.k_norm.weight": "model-00023-of-00031.safetensors",
+    "model.layers.60.self_attn.k_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.60.self_attn.o_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.60.self_attn.q_norm.weight": "model-00023-of-00031.safetensors",
+    "model.layers.60.self_attn.q_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.60.self_attn.v_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.61.input_layernorm.weight": "model-00024-of-00031.safetensors",
+    "model.layers.61.mlp.down_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.61.mlp.gate_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.61.mlp.up_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.61.post_attention_layernorm.weight": "model-00024-of-00031.safetensors",
+    "model.layers.61.self_attn.k_norm.weight": "model-00023-of-00031.safetensors",
+    "model.layers.61.self_attn.k_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.61.self_attn.o_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.61.self_attn.q_norm.weight": "model-00023-of-00031.safetensors",
+    "model.layers.61.self_attn.q_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.61.self_attn.v_proj.weight": "model-00023-of-00031.safetensors",
+    "model.layers.62.input_layernorm.weight": "model-00024-of-00031.safetensors",
+    "model.layers.62.mlp.down_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.62.mlp.gate_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.62.mlp.up_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.62.post_attention_layernorm.weight": "model-00024-of-00031.safetensors",
+    "model.layers.62.self_attn.k_norm.weight": "model-00024-of-00031.safetensors",
+    "model.layers.62.self_attn.k_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.62.self_attn.o_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.62.self_attn.q_norm.weight": "model-00024-of-00031.safetensors",
+    "model.layers.62.self_attn.q_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.62.self_attn.v_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.63.input_layernorm.weight": "model-00024-of-00031.safetensors",
+    "model.layers.63.mlp.down_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.63.mlp.gate_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.63.mlp.up_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.63.post_attention_layernorm.weight": "model-00024-of-00031.safetensors",
+    "model.layers.63.self_attn.k_norm.weight": "model-00024-of-00031.safetensors",
+    "model.layers.63.self_attn.k_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.63.self_attn.o_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.63.self_attn.q_norm.weight": "model-00024-of-00031.safetensors",
+    "model.layers.63.self_attn.q_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.63.self_attn.v_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.64.input_layernorm.weight": "model-00025-of-00031.safetensors",
+    "model.layers.64.mlp.down_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.64.mlp.gate_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.64.mlp.up_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.64.post_attention_layernorm.weight": "model-00025-of-00031.safetensors",
+    "model.layers.64.self_attn.k_norm.weight": "model-00024-of-00031.safetensors",
+    "model.layers.64.self_attn.k_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.64.self_attn.o_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.64.self_attn.q_norm.weight": "model-00024-of-00031.safetensors",
+    "model.layers.64.self_attn.q_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.64.self_attn.v_proj.weight": "model-00024-of-00031.safetensors",
+    "model.layers.65.input_layernorm.weight": "model-00025-of-00031.safetensors",
+    "model.layers.65.mlp.down_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.65.mlp.gate_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.65.mlp.up_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.65.post_attention_layernorm.weight": "model-00025-of-00031.safetensors",
+    "model.layers.65.self_attn.k_norm.weight": "model-00025-of-00031.safetensors",
+    "model.layers.65.self_attn.k_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.65.self_attn.o_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.65.self_attn.q_norm.weight": "model-00025-of-00031.safetensors",
+    "model.layers.65.self_attn.q_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.65.self_attn.v_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.66.input_layernorm.weight": "model-00025-of-00031.safetensors",
+    "model.layers.66.mlp.down_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.66.mlp.gate_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.66.mlp.up_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.66.post_attention_layernorm.weight": "model-00025-of-00031.safetensors",
+    "model.layers.66.self_attn.k_norm.weight": "model-00025-of-00031.safetensors",
+    "model.layers.66.self_attn.k_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.66.self_attn.o_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.66.self_attn.q_norm.weight": "model-00025-of-00031.safetensors",
+    "model.layers.66.self_attn.q_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.66.self_attn.v_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.67.input_layernorm.weight": "model-00026-of-00031.safetensors",
+    "model.layers.67.mlp.down_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.67.mlp.gate_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.67.mlp.up_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.67.post_attention_layernorm.weight": "model-00026-of-00031.safetensors",
+    "model.layers.67.self_attn.k_norm.weight": "model-00025-of-00031.safetensors",
+    "model.layers.67.self_attn.k_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.67.self_attn.o_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.67.self_attn.q_norm.weight": "model-00025-of-00031.safetensors",
+    "model.layers.67.self_attn.q_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.67.self_attn.v_proj.weight": "model-00025-of-00031.safetensors",
+    "model.layers.68.input_layernorm.weight": "model-00026-of-00031.safetensors",
+    "model.layers.68.mlp.down_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.68.mlp.gate_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.68.mlp.up_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.68.post_attention_layernorm.weight": "model-00026-of-00031.safetensors",
+    "model.layers.68.self_attn.k_norm.weight": "model-00026-of-00031.safetensors",
+    "model.layers.68.self_attn.k_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.68.self_attn.o_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.68.self_attn.q_norm.weight": "model-00026-of-00031.safetensors",
+    "model.layers.68.self_attn.q_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.68.self_attn.v_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.69.input_layernorm.weight": "model-00026-of-00031.safetensors",
+    "model.layers.69.mlp.down_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.69.mlp.gate_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.69.mlp.up_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.69.post_attention_layernorm.weight": "model-00026-of-00031.safetensors",
+    "model.layers.69.self_attn.k_norm.weight": "model-00026-of-00031.safetensors",
+    "model.layers.69.self_attn.k_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.69.self_attn.o_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.69.self_attn.q_norm.weight": "model-00026-of-00031.safetensors",
+    "model.layers.69.self_attn.q_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.69.self_attn.v_proj.weight": "model-00026-of-00031.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00004-of-00031.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00004-of-00031.safetensors",
+    "model.layers.7.self_attn.k_norm.weight": "model-00004-of-00031.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.7.self_attn.q_norm.weight": "model-00004-of-00031.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.70.input_layernorm.weight": "model-00027-of-00031.safetensors",
+    "model.layers.70.mlp.down_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.70.mlp.gate_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.70.mlp.up_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.70.post_attention_layernorm.weight": "model-00027-of-00031.safetensors",
+    "model.layers.70.self_attn.k_norm.weight": "model-00027-of-00031.safetensors",
+    "model.layers.70.self_attn.k_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.70.self_attn.o_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.70.self_attn.q_norm.weight": "model-00027-of-00031.safetensors",
+    "model.layers.70.self_attn.q_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.70.self_attn.v_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.71.input_layernorm.weight": "model-00027-of-00031.safetensors",
+    "model.layers.71.mlp.down_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.71.mlp.gate_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.71.mlp.up_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.71.post_attention_layernorm.weight": "model-00027-of-00031.safetensors",
+    "model.layers.71.self_attn.k_norm.weight": "model-00027-of-00031.safetensors",
+    "model.layers.71.self_attn.k_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.71.self_attn.o_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.71.self_attn.q_norm.weight": "model-00027-of-00031.safetensors",
+    "model.layers.71.self_attn.q_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.71.self_attn.v_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.72.input_layernorm.weight": "model-00028-of-00031.safetensors",
+    "model.layers.72.mlp.down_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.72.mlp.gate_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.72.mlp.up_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.72.post_attention_layernorm.weight": "model-00028-of-00031.safetensors",
+    "model.layers.72.self_attn.k_norm.weight": "model-00027-of-00031.safetensors",
+    "model.layers.72.self_attn.k_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.72.self_attn.o_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.72.self_attn.q_norm.weight": "model-00027-of-00031.safetensors",
+    "model.layers.72.self_attn.q_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.72.self_attn.v_proj.weight": "model-00027-of-00031.safetensors",
+    "model.layers.73.input_layernorm.weight": "model-00028-of-00031.safetensors",
+    "model.layers.73.mlp.down_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.73.mlp.gate_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.73.mlp.up_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.73.post_attention_layernorm.weight": "model-00028-of-00031.safetensors",
+    "model.layers.73.self_attn.k_norm.weight": "model-00028-of-00031.safetensors",
+    "model.layers.73.self_attn.k_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.73.self_attn.o_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.73.self_attn.q_norm.weight": "model-00028-of-00031.safetensors",
+    "model.layers.73.self_attn.q_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.73.self_attn.v_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.74.input_layernorm.weight": "model-00028-of-00031.safetensors",
+    "model.layers.74.mlp.down_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.74.mlp.gate_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.74.mlp.up_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.74.post_attention_layernorm.weight": "model-00028-of-00031.safetensors",
+    "model.layers.74.self_attn.k_norm.weight": "model-00028-of-00031.safetensors",
+    "model.layers.74.self_attn.k_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.74.self_attn.o_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.74.self_attn.q_norm.weight": "model-00028-of-00031.safetensors",
+    "model.layers.74.self_attn.q_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.74.self_attn.v_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.75.input_layernorm.weight": "model-00029-of-00031.safetensors",
+    "model.layers.75.mlp.down_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.75.mlp.gate_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.75.mlp.up_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.75.post_attention_layernorm.weight": "model-00029-of-00031.safetensors",
+    "model.layers.75.self_attn.k_norm.weight": "model-00028-of-00031.safetensors",
+    "model.layers.75.self_attn.k_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.75.self_attn.o_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.75.self_attn.q_norm.weight": "model-00028-of-00031.safetensors",
+    "model.layers.75.self_attn.q_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.75.self_attn.v_proj.weight": "model-00028-of-00031.safetensors",
+    "model.layers.76.input_layernorm.weight": "model-00029-of-00031.safetensors",
+    "model.layers.76.mlp.down_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.76.mlp.gate_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.76.mlp.up_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.76.post_attention_layernorm.weight": "model-00029-of-00031.safetensors",
+    "model.layers.76.self_attn.k_norm.weight": "model-00029-of-00031.safetensors",
+    "model.layers.76.self_attn.k_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.76.self_attn.o_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.76.self_attn.q_norm.weight": "model-00029-of-00031.safetensors",
+    "model.layers.76.self_attn.q_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.76.self_attn.v_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.77.input_layernorm.weight": "model-00029-of-00031.safetensors",
+    "model.layers.77.mlp.down_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.77.mlp.gate_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.77.mlp.up_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.77.post_attention_layernorm.weight": "model-00029-of-00031.safetensors",
+    "model.layers.77.self_attn.k_norm.weight": "model-00029-of-00031.safetensors",
+    "model.layers.77.self_attn.k_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.77.self_attn.o_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.77.self_attn.q_norm.weight": "model-00029-of-00031.safetensors",
+    "model.layers.77.self_attn.q_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.77.self_attn.v_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.78.input_layernorm.weight": "model-00030-of-00031.safetensors",
+    "model.layers.78.mlp.down_proj.weight": "model-00030-of-00031.safetensors",
+    "model.layers.78.mlp.gate_proj.weight": "model-00030-of-00031.safetensors",
+    "model.layers.78.mlp.up_proj.weight": "model-00030-of-00031.safetensors",
+    "model.layers.78.post_attention_layernorm.weight": "model-00030-of-00031.safetensors",
+    "model.layers.78.self_attn.k_norm.weight": "model-00029-of-00031.safetensors",
+    "model.layers.78.self_attn.k_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.78.self_attn.o_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.78.self_attn.q_norm.weight": "model-00029-of-00031.safetensors",
+    "model.layers.78.self_attn.q_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.78.self_attn.v_proj.weight": "model-00029-of-00031.safetensors",
+    "model.layers.79.input_layernorm.weight": "model-00030-of-00031.safetensors",
+    "model.layers.79.mlp.down_proj.weight": "model-00030-of-00031.safetensors",
+    "model.layers.79.mlp.gate_proj.weight": "model-00030-of-00031.safetensors",
+    "model.layers.79.mlp.up_proj.weight": "model-00030-of-00031.safetensors",
+    "model.layers.79.post_attention_layernorm.weight": "model-00030-of-00031.safetensors",
+    "model.layers.79.self_attn.k_norm.weight": "model-00030-of-00031.safetensors",
+    "model.layers.79.self_attn.k_proj.weight": "model-00030-of-00031.safetensors",
+    "model.layers.79.self_attn.o_proj.weight": "model-00030-of-00031.safetensors",
+    "model.layers.79.self_attn.q_norm.weight": "model-00030-of-00031.safetensors",
+    "model.layers.79.self_attn.q_proj.weight": "model-00030-of-00031.safetensors",
+    "model.layers.79.self_attn.v_proj.weight": "model-00030-of-00031.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00004-of-00031.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00004-of-00031.safetensors",
+    "model.layers.8.self_attn.k_norm.weight": "model-00004-of-00031.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.8.self_attn.q_norm.weight": "model-00004-of-00031.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00005-of-00031.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00005-of-00031.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00005-of-00031.safetensors",
+    "model.layers.9.self_attn.k_norm.weight": "model-00004-of-00031.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.9.self_attn.q_norm.weight": "model-00004-of-00031.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00004-of-00031.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00004-of-00031.safetensors",
+    "model.norm.weight": "model-00030-of-00031.safetensors"
+  }
+}

qwen2to3.py

@@ -0,0 +1,142 @@
+# file: convert_qwen2.5_to_qwen3_final_decoupled.py
+
+import torch
+import os
+import json
+import re
+from datetime import datetime
+from tqdm import tqdm
+from transformers import AutoTokenizer, AutoConfig, AutoModelForCausalLM
+from transformers import Qwen3Config, Qwen3ForCausalLM
+from collections import Counter
+
+# --- Helper Functions (Unchanged) ---
+
+def create_vocab_mapping(s_tok, t_tok):
+    s_vocab, t_vocab = s_tok.get_vocab(), t_tok.get_vocab()
+    s_tok_to_id = {t: i for t, i in s_vocab.items()}
+    mapping = {t_id: s_tok_to_id.get(t, -1) for t, t_id in t_vocab.items()}
+    matches = sum(1 for v in mapping.values() if v != -1)
+    print(f"Vocabulary overlap: {matches}/{len(t_vocab)} tokens ({matches/len(t_vocab)*100:.1f}%) will be transferred.")
+    return mapping
+
+def verify_special_tokens(s_tok, t_tok, mapping):
+    print("\nVerifying special token mappings...")
+    for name, token_value in t_tok.special_tokens_map.items():
+        def _process_token(token_str):
+            if token_str and token_str in t_tok.get_vocab():
+                t_id = t_tok.convert_tokens_to_ids(token_str)
+                s_id = mapping.get(t_id, -1)
+                status = f"Mapped (T: {t_id} -> S: {s_id})" if s_id != -1 else "NOT FOUND in source (initialized with mean)"
+                print(f"  ✓ ('{token_str}'): {status}")
+        if isinstance(token_value, str): _process_token(token_value)
+        elif isinstance(token_value, list):
+            for token_str_in_list in token_value: _process_token(token_str_in_list)
+
+def create_hybrid_matrix(s_matrix, mapping, shape):
+    print("  -> Calculating mean embedding from source model for new token initialization...")
+    mean_embedding = s_matrix.mean(dim=0, keepdim=True)
+    hybrid = torch.zeros(shape, dtype=s_matrix.dtype, device='cpu')
+    for t_id, s_id in mapping.items():
+        hybrid[t_id] = s_matrix[s_id] if s_id != -1 else mean_embedding
+    return hybrid.to(s_matrix.device)
+
+def save_config_diff(s_conf, t_conf, path):
+    s_dict, t_dict = s_conf.to_dict(), t_conf.to_dict()
+    diff = {'changed': {}, 'added': {}, 'removed': {}}
+    for k in set(s_dict.keys()) | set(t_dict.keys()):
+        if s_dict.get(k) != t_dict.get(k):
+            if k in s_dict and k in t_dict: diff['changed'][k] = {'from': s_dict[k], 'to': t_dict[k]}
+            elif k in t_dict: diff['added'][k] = t_dict[k]
+            else: diff['removed'][k] = s_dict[k]
+    with open(os.path.join(path, "config_diff.json"), "w") as f: json.dump(diff, f, indent=2)
+
+def validate_model(path):
+    print("\n[Step 6/6] Validating final model (smoke test)...")
+    try:
+        tokenizer = AutoTokenizer.from_pretrained(path)
+        model = AutoModelForCausalLM.from_pretrained(path, device_map="auto", torch_dtype=torch.bfloat16)
+        model.eval()
+        prompt = "The theory of relativity states that"
+        print(f"\nValidation Prompt: '{prompt}'")
+        inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+        with torch.no_grad():
+            outputs = model.generate(**inputs, max_new_tokens=25, do_sample=False, pad_token_id=tokenizer.eos_token_id)
+        response = tokenizer.decode(outputs[0], skip_special_tokens=True)
+        print(f"Generated Response: '{response}'")
+        assert len(response) > len(prompt) and "�" not in response, "Model generated no or garbled output."
+        print("\n  ✓ Validation successful: Model loads and generates coherent text.")
+    except Exception as e:
+        print(f"\n  ✗ Validation FAILED: {e}")
+
+# --- Main Conversion Logic ---
+def convert_qwen2_to_qwen3_decoupled():
+    source_model_id, donor_model_id = "Qwen/Qwen2.5-72B-Instruct", "Qwen/Qwen3-32B"
+    target_model_path = "./Qwen3-72B-Instruct"
+    print("Starting DECOUPLED conversion process (v5.4)...")
+
+    # --- 1. Pre-flight Checks ---
+    print("\n[Step 1/6] Running pre-flight architectural checks...")
+    s_config = AutoConfig.from_pretrained(source_model_id)
+    d_config = AutoConfig.from_pretrained(donor_model_id)
+    assert s_config.hidden_act == d_config.hidden_act, f"FATAL: Hidden activation mismatch! Source: {s_config.hidden_act}, Donor: {d_config.hidden_act}."
+    print("  ✓ Hidden activation functions match.")
+    if s_config.rope_theta != d_config.rope_theta:
+        print(f"  ✓ RoPE Theta: Using donor value {d_config.rope_theta} (source was {s_config.rope_theta})")
+    
+    # --- 2. Load Models & Tokenizers using AutoModel ---
+    print("\n[Step 2/6] Loading models & tokenizers using standard AutoClasses...")
+    dtype = torch.bfloat16
+    s_model = AutoModelForCausalLM.from_pretrained(source_model_id, torch_dtype=dtype, device_map="auto")
+    # Donor model is only needed for its config now, so we can load it with low_cpu_mem_usage
+    d_model = AutoModelForCausalLM.from_pretrained(donor_model_id, torch_dtype=dtype, low_cpu_mem_usage=True)
+    s_tokenizer = AutoTokenizer.from_pretrained(source_model_id)
+    t_tokenizer = AutoTokenizer.from_pretrained(donor_model_id)
+
+    # --- 3. Create Target Config & Initialize ---
+    print("\n[Step 3/6] Creating target Qwen3 72B config & initializing model shell...")
+    t_config = Qwen3Config(hidden_size=s_config.hidden_size, intermediate_size=s_config.intermediate_size, num_hidden_layers=s_config.num_hidden_layers, num_attention_heads=s_config.num_attention_heads, num_key_value_heads=s_config.num_key_value_heads, max_position_embeddings=s_config.max_position_embeddings, max_window_layers=s_config.max_window_layers, sliding_window=s_config.sliding_window, attention_bias=d_config.attention_bias, hidden_act=d_config.hidden_act, initializer_range=d_config.initializer_range, rms_norm_eps=d_config.rms_norm_eps, rope_theta=d_config.rope_theta, vocab_size=d_config.vocab_size, tie_word_embeddings=True)
+    with torch.device("meta"): t_model = Qwen3ForCausalLM(t_config)
+
+    # --- 4. Convert and Transfer Weights ---
+    print("\n[Step 4/6] Converting weights (memory-safe)...")
+    s_state_dict = {k: v.to('cpu', dtype=dtype) for k, v in tqdm(s_model.state_dict().items(), desc="Source state dict to CPU")}
+    
+    vocab_mapping = create_vocab_mapping(s_tokenizer, t_tokenizer)
+    verify_special_tokens(s_tokenizer, t_tokenizer, vocab_mapping)
+    
+    new_state_dict = {}
+    head_dim = s_config.hidden_size // s_config.num_attention_heads
+    
+    for key in tqdm(t_model.state_dict().keys(), desc="Transferring weights"):
+        if "q_norm.weight" in key or "k_norm.weight" in key:
+            # --- FINAL FIX: Initialize norm layers to a neutral state (all ones) ---
+            # This is the safe and correct way to handle grafting normalization layers
+            # between models of different scales.
+            new_state_dict[key] = torch.ones(head_dim, dtype=dtype)
+        elif "model.embed_tokens.weight" in key: new_state_dict[key] = create_hybrid_matrix(s_state_dict[key], vocab_mapping, (t_config.vocab_size, t_config.hidden_size))
+        elif "lm_head.weight" in key: new_state_dict[key] = create_hybrid_matrix(s_state_dict[key], vocab_mapping, (t_config.vocab_size, t_config.hidden_size))
+        elif key in s_state_dict: new_state_dict[key] = s_state_dict[key].clone()
+        else: print(f"  ⚠️ Unhandled key: {key} (not in source, skipping)")
+
+    t_model.load_state_dict(new_state_dict, strict=True, assign=True)
+    t_model = t_model.to(dtype)
+
+    # --- 5. Save Final Model & Metadata ---
+    print("\n[Step 5/6] Saving final model and supporting files...")
+    if not os.path.exists(target_model_path): os.makedirs(target_model_path)
+    t_model.save_pretrained(target_model_path, safe_serialization=True)
+    t_tokenizer.save_pretrained(target_model_path)
+    save_config_diff(s_config, t_config, target_model_path)
+    metadata = {"conversion_date_utc": datetime.now(datetime.UTC).isoformat(), "source_model": source_model_id, "donor_model": donor_model_id,
+                "warnings": ["This is a community-created model merge. Its behavior may be unpredictable.", "q_norm/k_norm layers were initialized to ones, not grafted. This is safer but may require fine-tuning to optimize.", "Sliding window config inherited from Qwen2.5 with Qwen3 RoPE theta - long context behavior MUST be validated.", "Post-conversion evaluation is highly recommended for numerical stability, quantization, and safety alignment."]}
+    with open(os.path.join(target_model_path, "conversion_metadata.json"), "w") as f: json.dump(metadata, f, indent=2)
+    print(f"✅ Model saved to: {target_model_path}")
+    
+    # --- 6. Final Validation ---
+    del s_model, d_model, s_state_dict, t_model, new_state_dict
+    torch.cuda.empty_cache()
+    validate_model(target_model_path)
+
+if __name__ == "__main__":
+    convert_qwen2_to_qwen3_decoupled()

special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4
+size 11422654

tokenizer_config.json

@@ -0,0 +1,240 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0].role == 'system' %}\n        {{- messages[0].content + '\\n\\n' }}\n    {%- endif %}\n    {{- \"# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0].role == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0].content + '<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}\n{%- for message in messages[::-1] %}\n    {%- set index = (messages|length - 1) - loop.index0 %}\n    {%- if ns.multi_step_tool and message.role == \"user\" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}\n        {%- set ns.multi_step_tool = false %}\n        {%- set ns.last_query_index = index %}\n    {%- endif %}\n{%- endfor %}\n{%- for message in messages %}\n    {%- if message.content is string %}\n        {%- set content = message.content %}\n    {%- else %}\n        {%- set content = '' %}\n    {%- endif %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) %}\n        {{- '<|im_start|>' + message.role + '\\n' + content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {%- set reasoning_content = '' %}\n        {%- if message.reasoning_content is string %}\n            {%- set reasoning_content = message.reasoning_content %}\n        {%- else %}\n            {%- if '</think>' in content %}\n                {%- set reasoning_content = content.split('</think>')[0].rstrip('\\n').split('<think>')[-1].lstrip('\\n') %}\n                {%- set content = content.split('</think>')[-1].lstrip('\\n') %}\n            {%- endif %}\n        {%- endif %}\n        {%- if loop.index0 > ns.last_query_index %}\n            {%- if loop.last or (not loop.last and reasoning_content) %}\n                {{- '<|im_start|>' + message.role + '\\n<think>\\n' + reasoning_content.strip('\\n') + '\\n</think>\\n\\n' + content.lstrip('\\n') }}\n            {%- else %}\n                {{- '<|im_start|>' + message.role + '\\n' + content }}\n            {%- endif %}\n        {%- else %}\n            {{- '<|im_start|>' + message.role + '\\n' + content }}\n        {%- endif %}\n        {%- if message.tool_calls %}\n            {%- for tool_call in message.tool_calls %}\n                {%- if (loop.first and content) or (not loop.first) %}\n                    {{- '\\n' }}\n                {%- endif %}\n                {%- if tool_call.function %}\n                    {%- set tool_call = tool_call.function %}\n                {%- endif %}\n                {{- '<tool_call>\\n{\"name\": \"' }}\n                {{- tool_call.name }}\n                {{- '\", \"arguments\": ' }}\n                {%- if tool_call.arguments is string %}\n                    {{- tool_call.arguments }}\n                {%- else %}\n                    {{- tool_call.arguments | tojson }}\n                {%- endif %}\n                {{- '}\\n</tool_call>' }}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if loop.first or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n    {%- if enable_thinking is defined and enable_thinking is false %}\n        {{- '<think>\\n\\n</think>\\n\\n' }}\n    {%- endif %}\n{%- endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}