Add files using upload-large-folder tool

LICENSE

@@ -0,0 +1,51 @@
+Qwen LICENSE AGREEMENT
+ 
+Qwen LICENSE AGREEMENT Release Date: September 19, 2024
+ 
+By clicking to agree or by using or distributing any portion or element of the Qwen Materials, you will be deemed to have recognized and accepted the content of this Agreement, which is effective immediately.
+ 
+1. Definitions
+    a. This Qwen LICENSE AGREEMENT (this "Agreement") shall mean the terms and conditions for use, reproduction, distribution and modification of the Materials as defined by this Agreement.
+    b. "We" (or "Us") shall mean Alibaba Cloud.
+    c. "You" (or "Your") shall mean a natural person or legal entity exercising the rights granted by this Agreement and/or using the Materials for any purpose and in any field of use.
+    d. "Third Parties" shall mean individuals or legal entities that are not under common control with us or you.
+    e. "Qwen" shall mean the large language models, and software and algorithms, consisting of trained model weights, parameters (including optimizer states), machine-learning model code, inference-enabling code, training-enabling code, fine-tuning enabling code and other elements of the foregoing distributed by us.
+    f. "Materials" shall mean, collectively, Alibaba Cloud's proprietary Qwen and Documentation (and any portion thereof) made available under this Agreement.
+    g. "Source" form shall mean the preferred form for making modifications, including but not limited to model source code, documentation source, and configuration files.
+    h. "Object" form shall mean any form resulting from mechanical transformation or translation of a Source form, including but not limited to compiled object code, generated documentation, and conversions to other media types.
+ 
+2. Grant of Rights
+You are granted a non-exclusive, worldwide, non-transferable and royalty-free limited license under Alibaba Cloud's intellectual property or other rights owned by us embodied in the Materials to use, reproduce, distribute, copy, create derivative works of, and make modifications to the Materials.
+ 
+3. Redistribution
+You may distribute copies or make the Materials, or derivative works thereof, available as part of a product or service that contains any of them, with or without modifications, and in Source or Object form, provided that you meet the following conditions:
+    a. You shall give any other recipients of the Materials or derivative works a copy of this Agreement;
+    b. You shall cause any modified files to carry prominent notices stating that you changed the files;
+    c. You shall retain in all copies of the Materials that you distribute the following attribution notices within a "Notice" text file distributed as a part of such copies: "Qwen is licensed under the Qwen LICENSE AGREEMENT, Copyright (c) Alibaba Cloud. All Rights Reserved."; and
+    d. You may add your own copyright statement to your modifications and may provide additional or different license terms and conditions for use, reproduction, or distribution of your modifications, or for any such derivative works as a whole, provided your use, reproduction, and distribution of the work otherwise complies with the terms and conditions of this Agreement.
+ 
+4. Restrictions
+If you are commercially using the Materials, and your product or service has more than 100 million monthly active users, you shall request a license from us. You cannot exercise your rights under this Agreement without our express authorization.
+ 
+5. Rules of use
+    a. The Materials may be subject to export controls or restrictions in China, the United States or other countries or regions. You shall comply with applicable laws and regulations in your use of the Materials.
+    b. If you use the Materials or any outputs or results therefrom to create, train, fine-tune, or improve an AI model that is distributed or made available, you shall prominently display “Built with Qwen” or “Improved using Qwen” in the related product documentation.
+ 
+6. Intellectual Property
+    a. We retain ownership of all intellectual property rights in and to the Materials and derivatives made by or for us. Conditioned upon compliance with the terms and conditions of this Agreement, with respect to any derivative works and modifications of the Materials that are made by you, you are and will be the owner of such derivative works and modifications.
+    b. No trademark license is granted to use the trade names, trademarks, service marks, or product names of us, except as required to fulfill notice requirements under this Agreement or as required for reasonable and customary use in describing and redistributing the Materials.
+    c. If you commence a lawsuit or other proceedings (including a cross-claim or counterclaim in a lawsuit) against us or any entity alleging that the Materials or any output therefrom, or any part of the foregoing, infringe any intellectual property or other right owned or licensable by you, then all licenses granted to you under this Agreement shall terminate as of the date such lawsuit or other proceeding is commenced or brought.
+ 
+7. Disclaimer of Warranty and Limitation of Liability
+    a. We are not obligated to support, update, provide training for, or develop any further version of the Qwen Materials or to grant any license thereto.
+    b. THE MATERIALS ARE PROVIDED "AS IS" WITHOUT ANY EXPRESS OR IMPLIED WARRANTY OF ANY KIND INCLUDING WARRANTIES OF MERCHANTABILITY, NONINFRINGEMENT, OR FITNESS FOR A PARTICULAR PURPOSE. WE MAKE NO WARRANTY AND ASSUME NO RESPONSIBILITY FOR THE SAFETY OR STABILITY OF THE MATERIALS AND ANY OUTPUT THEREFROM.
+    c. IN NO EVENT SHALL WE BE LIABLE TO YOU FOR ANY DAMAGES, INCLUDING, BUT NOT LIMITED TO ANY DIRECT, OR INDIRECT, SPECIAL OR CONSEQUENTIAL DAMAGES ARISING FROM YOUR USE OR INABILITY TO USE THE MATERIALS OR ANY OUTPUT OF IT, NO MATTER HOW IT’S CAUSED.
+    d. You will defend, indemnify and hold harmless us from and against any claim by any third party arising out of or related to your use or distribution of the Materials.
+ 
+8. Survival and Termination.
+    a. The term of this Agreement shall commence upon your acceptance of this Agreement or access to the Materials and will continue in full force and effect until terminated in accordance with the terms and conditions herein.
+    b. We may terminate this Agreement if you breach any of the terms or conditions of this Agreement. Upon termination of this Agreement, you must delete and cease use of the Materials. Sections 7 and 9 shall survive the termination of this Agreement.
+ 
+9. Governing Law and Jurisdiction.
+    a. This Agreement and any dispute arising out of or relating to it will be governed by the laws of China, without regard to conflict of law principles, and the UN Convention on Contracts for the International Sale of Goods does not apply to this Agreement.
+    b. The People's Courts in Hangzhou City shall have exclusive jurisdiction over any dispute arising out of this Agreement.

config.json

@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "Qwen2ForRewardModel"
+  ],
+  "attention_dropout": 0.0,
+  "auto_map": {
+    "AutoConfig": "configuration_qwen2_rm.Qwen2RMConfig",
+    "AutoModel": "modeling_qwen2_rm.Qwen2ForRewardModel"
+  },
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 8192,
+  "initializer_range": 0.02,
+  "intermediate_size": 29568,
+  "max_position_embeddings": 2048,
+  "max_window_layers": 70,
+  "model_type": "qwen2",
+  "num_attention_heads": 64,
+  "num_hidden_layers": 80,
+  "num_key_value_heads": 8,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 1000000.0,
+  "sliding_window": 131072,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.37.0",
+  "use_cache": true,
+  "use_mrope": false,
+  "use_sliding_window": false,
+  "vocab_size": 152064
+}

configuration.json

@@ -0,0 +1 @@
+{"framework":"Pytorch","task":"text-generation"}

configuration_qwen2_rm.py

@@ -0,0 +1,140 @@
+# coding=utf-8
+# Copyright 2024 The Qwen team, Alibaba Group and the HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""Qwen2 model configuration"""
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+
+class Qwen2RMConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`Qwen2Model`]. It is used to instantiate a
+    Qwen2 model according to the specified arguments, defining the model architecture. Instantiating a configuration
+    with the defaults will yield a similar configuration to that of
+    Qwen2-7B-beta [Qwen/Qwen2-7B-beta](https://huggingface.co/Qwen/Qwen2-7B-beta).
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        vocab_size (`int`, *optional*, defaults to 151936):
+            Vocabulary size of the Qwen2 model. Defines the number of different tokens that can be represented by the
+            `inputs_ids` passed when calling [`Qwen2Model`]
+        hidden_size (`int`, *optional*, defaults to 4096):
+            Dimension of the hidden representations.
+        intermediate_size (`int`, *optional*, defaults to 22016):
+            Dimension of the MLP representations.
+        num_hidden_layers (`int`, *optional*, defaults to 32):
+            Number of hidden layers in the Transformer encoder.
+        num_attention_heads (`int`, *optional*, defaults to 32):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        num_key_value_heads (`int`, *optional*, defaults to 32):
+            This is the number of key_value heads that should be used to implement Grouped Query Attention. If
+            `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
+            `num_key_value_heads=1` the model will use Multi Query Attention (MQA) otherwise GQA is used. When
+            converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
+            by meanpooling all the original heads within that group. For more details checkout [this
+            paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `32`.
+        hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
+            The non-linear activation function (function or string) in the decoder.
+        max_position_embeddings (`int`, *optional*, defaults to 32768):
+            The maximum sequence length that this model might ever be used with.
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        rms_norm_eps (`float`, *optional*, defaults to 1e-06):
+            The epsilon used by the rms normalization layers.
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should return the last key/values attentions (not used by all models). Only
+            relevant if `config.is_decoder=True`.
+        tie_word_embeddings (`bool`, *optional*, defaults to `False`):
+            Whether the model's input and output word embeddings should be tied.
+        rope_theta (`float`, *optional*, defaults to 10000.0):
+            The base period of the RoPE embeddings.
+        use_sliding_window (`bool`, *optional*, defaults to `False`):
+            Whether to use sliding window attention.
+        sliding_window (`int`, *optional*, defaults to 4096):
+            Sliding window attention (SWA) window size. If not specified, will default to `4096`.
+        max_window_layers (`int`, *optional*, defaults to 28):
+            The number of layers that use SWA (Sliding Window Attention). The bottom layers use SWA while the top use full attention.
+        attention_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio for the attention probabilities.
+
+    ```python
+    >>> from transformers import Qwen2Model, Qwen2Config
+
+    >>> # Initializing a Qwen2 style configuration
+    >>> configuration = Qwen2Config()
+
+    >>> # Initializing a model from the Qwen2-7B style configuration
+    >>> model = Qwen2Model(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "qwen2"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=151936,
+        hidden_size=4096,
+        intermediate_size=22016,
+        num_hidden_layers=32,
+        num_attention_heads=32,
+        num_key_value_heads=32,
+        hidden_act="silu",
+        max_position_embeddings=32768,
+        initializer_range=0.02,
+        rms_norm_eps=1e-6,
+        use_cache=True,
+        tie_word_embeddings=False,
+        rope_theta=10000.0,
+        use_sliding_window=False,
+        sliding_window=4096,
+        max_window_layers=28,
+        attention_dropout=0.0,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.max_position_embeddings = max_position_embeddings
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.use_sliding_window = use_sliding_window
+        self.sliding_window = sliding_window if use_sliding_window else None
+        self.max_window_layers = max_window_layers
+
+        # for backward compatibility
+        if num_key_value_heads is None:
+            num_key_value_heads = num_attention_heads
+
+        self.num_key_value_heads = num_key_value_heads
+        self.hidden_act = hidden_act
+        self.initializer_range = initializer_range
+        self.rms_norm_eps = rms_norm_eps
+        self.use_cache = use_cache
+        self.rope_theta = rope_theta
+        self.attention_dropout = attention_dropout
+
+        super().__init__(
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )

generation_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "pad_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "repetition_penalty": 1.05,
+  "temperature": 0.7,
+  "top_p": 0.8,
+  "top_k": 20,
+  "transformers_version": "4.37.0"
+}

model-00001-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a51e9240c91e877cad64d8988195d9e8c66abe88a98e095a51e303f08f175a1
+size 3896596186

model-00002-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78f643b5b89a35c939d95dd43b980c3d9f37a9ea93ca456cb60d92a93f58ef4a
+size 3995200440

model-00003-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4cb04ff53772c29b499893415d0e8cc67d61eb04115a7885d518bf72a8e6788
+size 3812769392

model-00004-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cae0a89fa74a51b4d76021b57eaea8ddb67c1a3f4e7938f73ccf59d12fd78fbf
+size 3995183944

model-00005-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4e05235061b8f23a4d74a4732ef1049792316dc8c9a64481f9abe5c9f253946
+size 3995183944

model-00006-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81f43c93a47ae5069bd2e3ba4bd670aa36bec23eaf2651230d8282c0428705f9
+size 3995200456

model-00007-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8400c01a2a1056edb32a377e66ab0b1bc3b0a06a03cf09cd22b6c43cc42b4f8f
+size 3812769424

model-00008-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2d9535681399fec92f53123518500ffd010f7bafe6eb2e4994c5a9b3e72e8ef
+size 3995183968

model-00009-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3defdc2a2184ff007ba2083e9e4f65714478db8ac6a1242817257b72b5ec513
+size 3995183968

model-00010-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf79f0328d48a76cd0bd98f163706285c7091f9c3cb3e6596217425deb803af9
+size 3995200464

model-00011-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:469b0f9c514935e0aeed006eabd479b454b88f97f45b486630047d8b7b336b8a
+size 3812769424

model-00012-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4c362e7d0bb8b122a6d084be68e0d00410fe7ffeb636ea7ed98ce66fbbb5b4d
+size 3995183968

model-00013-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e2bebd29bb1559360892c6493be9a2b0a9f2769fccd2194ad0a94358c488896
+size 3995183968

model-00014-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a2360c019a1d9d91f9cb299bb413f7f1f5b923e35ca5568402f76e9eaa89c11
+size 3995200464

model-00015-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d136b7a1eb7455949667877625f35a6f72576472d61f995271ee600377482e7
+size 3812769424

model-00016-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2170c194b75963f2ffbe440ba832774c5c12e2d5bcb68518b83341bc68c8bef9
+size 3995183968

model-00017-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c68a761c5cac68f822931f4b70a35fab0d6d5e1eb59760a3d61e7a642a62d272
+size 3995183968

model-00018-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2434d3d50078275b2d88b6eb8456816e4e34b0fd8ddc402db2ddbe5d6bba4dce
+size 3995200464

model-00019-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d6a6f2df2e36938cb6d752935b038c543596ef464875a8d27cefadfce23b310
+size 3812769424

model-00020-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ce7deef30c47cc1e965c7790c2f923f0b7ee2606436155e2edcc89e69c9afe3
+size 3995183968

model-00021-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e2e0684a0b826621c72d60b232d0d6eece2767058635a06b7b9374374e7c3c6
+size 3995183968

model-00022-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:510af817b70d11de63df9e1c90aeaaf1d6656aa5621101aaf3962d0df2aa43eb
+size 3995200464

model-00023-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41c44a4729399d7f8e161616ad2dd5ce39627176c3faa480709d8521b9b8e71b
+size 3812769424

model-00024-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6ac7860bea68b6181ec1d0350ab728268593f087d50ed3155150db7dac33858f
+size 3995183968

model-00025-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:219576a24a15338fa7fff19b673cc75f3964512be19200935068f66e2605e9a9
+size 3995183968

model-00026-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b58f0c4484e7ab8c8a85c860f85043adafd87495b030952e75d3169831f18b6
+size 3995200464

model-00027-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bdc606f27e9d12b6f81c21568e40f304e56bd5a3ff94aa81abeefc46313b2014
+size 3812769424

model-00028-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d1a3c15641be3e98a417213a95d8a63df434aa970ff7f2e48ec4b31ae73fa77
+size 3995183968

model-00029-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f444259ff180103090fcc1fbee00f38332c133352c458be6b0d0f648af2de39a
+size 3995183968

model-00030-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05f115361acef294ad1986f2d156681306ad39a4589c42003203528976be1c82
+size 3995200464

model-00031-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fbd380e7eb1b1acd2033b080c6a18c4f30cccd1fcb579e1bf9fc9537cc7648f2
+size 3812769424

model-00032-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e00cfb5c7c1f9ed366d194c7a33f92b5bc2b923c4525374f5d0699ef8c791abb
+size 3995183968

model-00033-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42ba0d5b29224cfdaeb1e0a1c887a114063f28ccbfc048d8cdf59c5aeb69a66f
+size 3995183968

model-00034-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddcec0c3dfadadd316329ecba7e21c08b1b9d0b10ef061dd883da65ec4b2fce1
+size 3995200464

model-00035-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:85281b5599dade6709abcabe9ee2391cf851dc489e48586ad742656c28c55dec
+size 3812769424

model-00036-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:266984ffcc4eec34c468203d9a53a6c1064bd2e0e2407951785f68860a743684
+size 3995183968

model-00037-of-00037.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa571248670fe1b3d498f3dda740f96422677de020914792ddc68aa6165c77b7
+size 3460317640

model.safetensors.index.json

@@ -0,0 +1,974 @@
+{
+  "metadata": {
+    "total_size": 145546657794
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00037-of-00037.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00037.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00037.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00037.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00037.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00037.safetensors",
+    "model.layers.0.self_attn.k_proj.bias": "model-00001-of-00037.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00037.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00037.safetensors",
+    "model.layers.0.self_attn.q_proj.bias": "model-00001-of-00037.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00037.safetensors",
+    "model.layers.0.self_attn.v_proj.bias": "model-00001-of-00037.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00037.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00002-of-00037.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00002-of-00037.safetensors",
+    "model.layers.1.self_attn.k_proj.bias": "model-00002-of-00037.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.1.self_attn.q_proj.bias": "model-00002-of-00037.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.1.self_attn.v_proj.bias": "model-00002-of-00037.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00006-of-00037.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00006-of-00037.safetensors",
+    "model.layers.10.self_attn.k_proj.bias": "model-00006-of-00037.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.10.self_attn.q_proj.bias": "model-00006-of-00037.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.10.self_attn.v_proj.bias": "model-00006-of-00037.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00006-of-00037.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00006-of-00037.safetensors",
+    "model.layers.11.self_attn.k_proj.bias": "model-00006-of-00037.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.11.self_attn.q_proj.bias": "model-00006-of-00037.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.11.self_attn.v_proj.bias": "model-00006-of-00037.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00006-of-00037.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00007-of-00037.safetensors",
+    "model.layers.12.self_attn.k_proj.bias": "model-00007-of-00037.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.12.self_attn.q_proj.bias": "model-00007-of-00037.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.12.self_attn.v_proj.bias": "model-00007-of-00037.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00007-of-00037.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00007-of-00037.safetensors",
+    "model.layers.13.self_attn.k_proj.bias": "model-00007-of-00037.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.13.self_attn.q_proj.bias": "model-00007-of-00037.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.13.self_attn.v_proj.bias": "model-00007-of-00037.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00007-of-00037.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00007-of-00037.safetensors",
+    "model.layers.14.self_attn.k_proj.bias": "model-00007-of-00037.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.14.self_attn.q_proj.bias": "model-00007-of-00037.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.14.self_attn.v_proj.bias": "model-00007-of-00037.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00007-of-00037.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00008-of-00037.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00008-of-00037.safetensors",
+    "model.layers.15.self_attn.k_proj.bias": "model-00008-of-00037.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.15.self_attn.q_proj.bias": "model-00008-of-00037.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.15.self_attn.v_proj.bias": "model-00008-of-00037.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00008-of-00037.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00008-of-00037.safetensors",
+    "model.layers.16.self_attn.k_proj.bias": "model-00008-of-00037.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.16.self_attn.q_proj.bias": "model-00008-of-00037.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.16.self_attn.v_proj.bias": "model-00008-of-00037.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00008-of-00037.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00009-of-00037.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00009-of-00037.safetensors",
+    "model.layers.17.self_attn.k_proj.bias": "model-00009-of-00037.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.17.self_attn.q_proj.bias": "model-00009-of-00037.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.17.self_attn.v_proj.bias": "model-00009-of-00037.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00009-of-00037.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00009-of-00037.safetensors",
+    "model.layers.18.self_attn.k_proj.bias": "model-00009-of-00037.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.18.self_attn.q_proj.bias": "model-00009-of-00037.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.18.self_attn.v_proj.bias": "model-00009-of-00037.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00009-of-00037.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00010-of-00037.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00010-of-00037.safetensors",
+    "model.layers.19.self_attn.k_proj.bias": "model-00010-of-00037.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.19.self_attn.q_proj.bias": "model-00010-of-00037.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.19.self_attn.v_proj.bias": "model-00010-of-00037.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00002-of-00037.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00002-of-00037.safetensors",
+    "model.layers.2.self_attn.k_proj.bias": "model-00002-of-00037.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.2.self_attn.q_proj.bias": "model-00002-of-00037.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.2.self_attn.v_proj.bias": "model-00002-of-00037.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00002-of-00037.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00010-of-00037.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00010-of-00037.safetensors",
+    "model.layers.20.self_attn.k_proj.bias": "model-00010-of-00037.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.20.self_attn.q_proj.bias": "model-00010-of-00037.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.20.self_attn.v_proj.bias": "model-00010-of-00037.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00010-of-00037.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00010-of-00037.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00011-of-00037.safetensors",
+    "model.layers.21.self_attn.k_proj.bias": "model-00011-of-00037.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.21.self_attn.q_proj.bias": "model-00011-of-00037.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.21.self_attn.v_proj.bias": "model-00011-of-00037.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00011-of-00037.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00011-of-00037.safetensors",
+    "model.layers.22.self_attn.k_proj.bias": "model-00011-of-00037.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.22.self_attn.q_proj.bias": "model-00011-of-00037.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.22.self_attn.v_proj.bias": "model-00011-of-00037.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00011-of-00037.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00011-of-00037.safetensors",
+    "model.layers.23.self_attn.k_proj.bias": "model-00011-of-00037.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.23.self_attn.q_proj.bias": "model-00011-of-00037.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.23.self_attn.v_proj.bias": "model-00011-of-00037.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00011-of-00037.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00012-of-00037.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00012-of-00037.safetensors",
+    "model.layers.24.self_attn.k_proj.bias": "model-00012-of-00037.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.24.self_attn.q_proj.bias": "model-00012-of-00037.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.24.self_attn.v_proj.bias": "model-00012-of-00037.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00012-of-00037.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00012-of-00037.safetensors",
+    "model.layers.25.self_attn.k_proj.bias": "model-00012-of-00037.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.25.self_attn.q_proj.bias": "model-00012-of-00037.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.25.self_attn.v_proj.bias": "model-00012-of-00037.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00012-of-00037.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00013-of-00037.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00013-of-00037.safetensors",
+    "model.layers.26.self_attn.k_proj.bias": "model-00013-of-00037.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.26.self_attn.q_proj.bias": "model-00013-of-00037.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.26.self_attn.v_proj.bias": "model-00013-of-00037.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00013-of-00037.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00013-of-00037.safetensors",
+    "model.layers.27.self_attn.k_proj.bias": "model-00013-of-00037.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.27.self_attn.q_proj.bias": "model-00013-of-00037.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.27.self_attn.v_proj.bias": "model-00013-of-00037.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00013-of-00037.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00014-of-00037.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00014-of-00037.safetensors",
+    "model.layers.28.self_attn.k_proj.bias": "model-00014-of-00037.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.28.self_attn.q_proj.bias": "model-00014-of-00037.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.28.self_attn.v_proj.bias": "model-00014-of-00037.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00014-of-00037.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00014-of-00037.safetensors",
+    "model.layers.29.self_attn.k_proj.bias": "model-00014-of-00037.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.29.self_attn.q_proj.bias": "model-00014-of-00037.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.29.self_attn.v_proj.bias": "model-00014-of-00037.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00014-of-00037.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00002-of-00037.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00003-of-00037.safetensors",
+    "model.layers.3.self_attn.k_proj.bias": "model-00003-of-00037.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.3.self_attn.q_proj.bias": "model-00003-of-00037.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.3.self_attn.v_proj.bias": "model-00003-of-00037.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00014-of-00037.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00015-of-00037.safetensors",
+    "model.layers.30.self_attn.k_proj.bias": "model-00015-of-00037.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.30.self_attn.q_proj.bias": "model-00015-of-00037.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.30.self_attn.v_proj.bias": "model-00015-of-00037.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00015-of-00037.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00015-of-00037.safetensors",
+    "model.layers.31.self_attn.k_proj.bias": "model-00015-of-00037.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.31.self_attn.q_proj.bias": "model-00015-of-00037.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.31.self_attn.v_proj.bias": "model-00015-of-00037.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00015-of-00037.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00015-of-00037.safetensors",
+    "model.layers.32.self_attn.k_proj.bias": "model-00015-of-00037.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.32.self_attn.q_proj.bias": "model-00015-of-00037.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.32.self_attn.v_proj.bias": "model-00015-of-00037.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00015-of-00037.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00016-of-00037.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00016-of-00037.safetensors",
+    "model.layers.33.self_attn.k_proj.bias": "model-00016-of-00037.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.33.self_attn.q_proj.bias": "model-00016-of-00037.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.33.self_attn.v_proj.bias": "model-00016-of-00037.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00016-of-00037.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00016-of-00037.safetensors",
+    "model.layers.34.self_attn.k_proj.bias": "model-00016-of-00037.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.34.self_attn.q_proj.bias": "model-00016-of-00037.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.34.self_attn.v_proj.bias": "model-00016-of-00037.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00016-of-00037.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00017-of-00037.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00017-of-00037.safetensors",
+    "model.layers.35.self_attn.k_proj.bias": "model-00017-of-00037.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.35.self_attn.q_proj.bias": "model-00017-of-00037.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.35.self_attn.v_proj.bias": "model-00017-of-00037.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00017-of-00037.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00017-of-00037.safetensors",
+    "model.layers.36.self_attn.k_proj.bias": "model-00017-of-00037.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.36.self_attn.q_proj.bias": "model-00017-of-00037.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.36.self_attn.v_proj.bias": "model-00017-of-00037.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00017-of-00037.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00018-of-00037.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00018-of-00037.safetensors",
+    "model.layers.37.self_attn.k_proj.bias": "model-00018-of-00037.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.37.self_attn.q_proj.bias": "model-00018-of-00037.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.37.self_attn.v_proj.bias": "model-00018-of-00037.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00018-of-00037.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00018-of-00037.safetensors",
+    "model.layers.38.self_attn.k_proj.bias": "model-00018-of-00037.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.38.self_attn.q_proj.bias": "model-00018-of-00037.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.38.self_attn.v_proj.bias": "model-00018-of-00037.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00018-of-00037.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00018-of-00037.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00019-of-00037.safetensors",
+    "model.layers.39.self_attn.k_proj.bias": "model-00019-of-00037.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.39.self_attn.q_proj.bias": "model-00019-of-00037.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.39.self_attn.v_proj.bias": "model-00019-of-00037.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00003-of-00037.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00003-of-00037.safetensors",
+    "model.layers.4.self_attn.k_proj.bias": "model-00003-of-00037.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.4.self_attn.q_proj.bias": "model-00003-of-00037.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.4.self_attn.v_proj.bias": "model-00003-of-00037.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00019-of-00037.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00019-of-00037.safetensors",
+    "model.layers.40.self_attn.k_proj.bias": "model-00019-of-00037.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.40.self_attn.q_proj.bias": "model-00019-of-00037.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.40.self_attn.v_proj.bias": "model-00019-of-00037.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00019-of-00037.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00019-of-00037.safetensors",
+    "model.layers.41.self_attn.k_proj.bias": "model-00019-of-00037.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.41.self_attn.q_proj.bias": "model-00019-of-00037.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.41.self_attn.v_proj.bias": "model-00019-of-00037.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00019-of-00037.safetensors",
+    "model.layers.42.input_layernorm.weight": "model-00020-of-00037.safetensors",
+    "model.layers.42.mlp.down_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.42.mlp.gate_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.42.mlp.up_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.42.post_attention_layernorm.weight": "model-00020-of-00037.safetensors",
+    "model.layers.42.self_attn.k_proj.bias": "model-00020-of-00037.safetensors",
+    "model.layers.42.self_attn.k_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.42.self_attn.o_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.42.self_attn.q_proj.bias": "model-00020-of-00037.safetensors",
+    "model.layers.42.self_attn.q_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.42.self_attn.v_proj.bias": "model-00020-of-00037.safetensors",
+    "model.layers.42.self_attn.v_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.43.input_layernorm.weight": "model-00020-of-00037.safetensors",
+    "model.layers.43.mlp.down_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.43.mlp.gate_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.43.mlp.up_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.43.post_attention_layernorm.weight": "model-00020-of-00037.safetensors",
+    "model.layers.43.self_attn.k_proj.bias": "model-00020-of-00037.safetensors",
+    "model.layers.43.self_attn.k_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.43.self_attn.o_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.43.self_attn.q_proj.bias": "model-00020-of-00037.safetensors",
+    "model.layers.43.self_attn.q_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.43.self_attn.v_proj.bias": "model-00020-of-00037.safetensors",
+    "model.layers.43.self_attn.v_proj.weight": "model-00020-of-00037.safetensors",
+    "model.layers.44.input_layernorm.weight": "model-00021-of-00037.safetensors",
+    "model.layers.44.mlp.down_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.44.mlp.gate_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.44.mlp.up_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.44.post_attention_layernorm.weight": "model-00021-of-00037.safetensors",
+    "model.layers.44.self_attn.k_proj.bias": "model-00021-of-00037.safetensors",
+    "model.layers.44.self_attn.k_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.44.self_attn.o_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.44.self_attn.q_proj.bias": "model-00021-of-00037.safetensors",
+    "model.layers.44.self_attn.q_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.44.self_attn.v_proj.bias": "model-00021-of-00037.safetensors",
+    "model.layers.44.self_attn.v_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.45.input_layernorm.weight": "model-00021-of-00037.safetensors",
+    "model.layers.45.mlp.down_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.45.mlp.gate_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.45.mlp.up_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.45.post_attention_layernorm.weight": "model-00021-of-00037.safetensors",
+    "model.layers.45.self_attn.k_proj.bias": "model-00021-of-00037.safetensors",
+    "model.layers.45.self_attn.k_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.45.self_attn.o_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.45.self_attn.q_proj.bias": "model-00021-of-00037.safetensors",
+    "model.layers.45.self_attn.q_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.45.self_attn.v_proj.bias": "model-00021-of-00037.safetensors",
+    "model.layers.45.self_attn.v_proj.weight": "model-00021-of-00037.safetensors",
+    "model.layers.46.input_layernorm.weight": "model-00022-of-00037.safetensors",
+    "model.layers.46.mlp.down_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.46.mlp.gate_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.46.mlp.up_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.46.post_attention_layernorm.weight": "model-00022-of-00037.safetensors",
+    "model.layers.46.self_attn.k_proj.bias": "model-00022-of-00037.safetensors",
+    "model.layers.46.self_attn.k_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.46.self_attn.o_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.46.self_attn.q_proj.bias": "model-00022-of-00037.safetensors",
+    "model.layers.46.self_attn.q_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.46.self_attn.v_proj.bias": "model-00022-of-00037.safetensors",
+    "model.layers.46.self_attn.v_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.47.input_layernorm.weight": "model-00022-of-00037.safetensors",
+    "model.layers.47.mlp.down_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.47.mlp.gate_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.47.mlp.up_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.47.post_attention_layernorm.weight": "model-00022-of-00037.safetensors",
+    "model.layers.47.self_attn.k_proj.bias": "model-00022-of-00037.safetensors",
+    "model.layers.47.self_attn.k_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.47.self_attn.o_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.47.self_attn.q_proj.bias": "model-00022-of-00037.safetensors",
+    "model.layers.47.self_attn.q_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.47.self_attn.v_proj.bias": "model-00022-of-00037.safetensors",
+    "model.layers.47.self_attn.v_proj.weight": "model-00022-of-00037.safetensors",
+    "model.layers.48.input_layernorm.weight": "model-00022-of-00037.safetensors",
+    "model.layers.48.mlp.down_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.48.mlp.gate_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.48.mlp.up_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.48.post_attention_layernorm.weight": "model-00023-of-00037.safetensors",
+    "model.layers.48.self_attn.k_proj.bias": "model-00023-of-00037.safetensors",
+    "model.layers.48.self_attn.k_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.48.self_attn.o_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.48.self_attn.q_proj.bias": "model-00023-of-00037.safetensors",
+    "model.layers.48.self_attn.q_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.48.self_attn.v_proj.bias": "model-00023-of-00037.safetensors",
+    "model.layers.48.self_attn.v_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.49.input_layernorm.weight": "model-00023-of-00037.safetensors",
+    "model.layers.49.mlp.down_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.49.mlp.gate_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.49.mlp.up_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.49.post_attention_layernorm.weight": "model-00023-of-00037.safetensors",
+    "model.layers.49.self_attn.k_proj.bias": "model-00023-of-00037.safetensors",
+    "model.layers.49.self_attn.k_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.49.self_attn.o_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.49.self_attn.q_proj.bias": "model-00023-of-00037.safetensors",
+    "model.layers.49.self_attn.q_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.49.self_attn.v_proj.bias": "model-00023-of-00037.safetensors",
+    "model.layers.49.self_attn.v_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00003-of-00037.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00003-of-00037.safetensors",
+    "model.layers.5.self_attn.k_proj.bias": "model-00003-of-00037.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.5.self_attn.q_proj.bias": "model-00003-of-00037.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.5.self_attn.v_proj.bias": "model-00003-of-00037.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00003-of-00037.safetensors",
+    "model.layers.50.input_layernorm.weight": "model-00023-of-00037.safetensors",
+    "model.layers.50.mlp.down_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.50.mlp.gate_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.50.mlp.up_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.50.post_attention_layernorm.weight": "model-00023-of-00037.safetensors",
+    "model.layers.50.self_attn.k_proj.bias": "model-00023-of-00037.safetensors",
+    "model.layers.50.self_attn.k_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.50.self_attn.o_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.50.self_attn.q_proj.bias": "model-00023-of-00037.safetensors",
+    "model.layers.50.self_attn.q_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.50.self_attn.v_proj.bias": "model-00023-of-00037.safetensors",
+    "model.layers.50.self_attn.v_proj.weight": "model-00023-of-00037.safetensors",
+    "model.layers.51.input_layernorm.weight": "model-00024-of-00037.safetensors",
+    "model.layers.51.mlp.down_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.51.mlp.gate_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.51.mlp.up_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.51.post_attention_layernorm.weight": "model-00024-of-00037.safetensors",
+    "model.layers.51.self_attn.k_proj.bias": "model-00024-of-00037.safetensors",
+    "model.layers.51.self_attn.k_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.51.self_attn.o_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.51.self_attn.q_proj.bias": "model-00024-of-00037.safetensors",
+    "model.layers.51.self_attn.q_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.51.self_attn.v_proj.bias": "model-00024-of-00037.safetensors",
+    "model.layers.51.self_attn.v_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.52.input_layernorm.weight": "model-00024-of-00037.safetensors",
+    "model.layers.52.mlp.down_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.52.mlp.gate_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.52.mlp.up_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.52.post_attention_layernorm.weight": "model-00024-of-00037.safetensors",
+    "model.layers.52.self_attn.k_proj.bias": "model-00024-of-00037.safetensors",
+    "model.layers.52.self_attn.k_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.52.self_attn.o_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.52.self_attn.q_proj.bias": "model-00024-of-00037.safetensors",
+    "model.layers.52.self_attn.q_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.52.self_attn.v_proj.bias": "model-00024-of-00037.safetensors",
+    "model.layers.52.self_attn.v_proj.weight": "model-00024-of-00037.safetensors",
+    "model.layers.53.input_layernorm.weight": "model-00025-of-00037.safetensors",
+    "model.layers.53.mlp.down_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.53.mlp.gate_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.53.mlp.up_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.53.post_attention_layernorm.weight": "model-00025-of-00037.safetensors",
+    "model.layers.53.self_attn.k_proj.bias": "model-00025-of-00037.safetensors",
+    "model.layers.53.self_attn.k_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.53.self_attn.o_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.53.self_attn.q_proj.bias": "model-00025-of-00037.safetensors",
+    "model.layers.53.self_attn.q_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.53.self_attn.v_proj.bias": "model-00025-of-00037.safetensors",
+    "model.layers.53.self_attn.v_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.54.input_layernorm.weight": "model-00025-of-00037.safetensors",
+    "model.layers.54.mlp.down_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.54.mlp.gate_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.54.mlp.up_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.54.post_attention_layernorm.weight": "model-00025-of-00037.safetensors",
+    "model.layers.54.self_attn.k_proj.bias": "model-00025-of-00037.safetensors",
+    "model.layers.54.self_attn.k_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.54.self_attn.o_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.54.self_attn.q_proj.bias": "model-00025-of-00037.safetensors",
+    "model.layers.54.self_attn.q_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.54.self_attn.v_proj.bias": "model-00025-of-00037.safetensors",
+    "model.layers.54.self_attn.v_proj.weight": "model-00025-of-00037.safetensors",
+    "model.layers.55.input_layernorm.weight": "model-00026-of-00037.safetensors",
+    "model.layers.55.mlp.down_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.55.mlp.gate_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.55.mlp.up_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.55.post_attention_layernorm.weight": "model-00026-of-00037.safetensors",
+    "model.layers.55.self_attn.k_proj.bias": "model-00026-of-00037.safetensors",
+    "model.layers.55.self_attn.k_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.55.self_attn.o_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.55.self_attn.q_proj.bias": "model-00026-of-00037.safetensors",
+    "model.layers.55.self_attn.q_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.55.self_attn.v_proj.bias": "model-00026-of-00037.safetensors",
+    "model.layers.55.self_attn.v_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.56.input_layernorm.weight": "model-00026-of-00037.safetensors",
+    "model.layers.56.mlp.down_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.56.mlp.gate_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.56.mlp.up_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.56.post_attention_layernorm.weight": "model-00026-of-00037.safetensors",
+    "model.layers.56.self_attn.k_proj.bias": "model-00026-of-00037.safetensors",
+    "model.layers.56.self_attn.k_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.56.self_attn.o_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.56.self_attn.q_proj.bias": "model-00026-of-00037.safetensors",
+    "model.layers.56.self_attn.q_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.56.self_attn.v_proj.bias": "model-00026-of-00037.safetensors",
+    "model.layers.56.self_attn.v_proj.weight": "model-00026-of-00037.safetensors",
+    "model.layers.57.input_layernorm.weight": "model-00026-of-00037.safetensors",
+    "model.layers.57.mlp.down_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.57.mlp.gate_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.57.mlp.up_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.57.post_attention_layernorm.weight": "model-00027-of-00037.safetensors",
+    "model.layers.57.self_attn.k_proj.bias": "model-00027-of-00037.safetensors",
+    "model.layers.57.self_attn.k_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.57.self_attn.o_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.57.self_attn.q_proj.bias": "model-00027-of-00037.safetensors",
+    "model.layers.57.self_attn.q_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.57.self_attn.v_proj.bias": "model-00027-of-00037.safetensors",
+    "model.layers.57.self_attn.v_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.58.input_layernorm.weight": "model-00027-of-00037.safetensors",
+    "model.layers.58.mlp.down_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.58.mlp.gate_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.58.mlp.up_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.58.post_attention_layernorm.weight": "model-00027-of-00037.safetensors",
+    "model.layers.58.self_attn.k_proj.bias": "model-00027-of-00037.safetensors",
+    "model.layers.58.self_attn.k_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.58.self_attn.o_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.58.self_attn.q_proj.bias": "model-00027-of-00037.safetensors",
+    "model.layers.58.self_attn.q_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.58.self_attn.v_proj.bias": "model-00027-of-00037.safetensors",
+    "model.layers.58.self_attn.v_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.59.input_layernorm.weight": "model-00027-of-00037.safetensors",
+    "model.layers.59.mlp.down_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.59.mlp.gate_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.59.mlp.up_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.59.post_attention_layernorm.weight": "model-00027-of-00037.safetensors",
+    "model.layers.59.self_attn.k_proj.bias": "model-00027-of-00037.safetensors",
+    "model.layers.59.self_attn.k_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.59.self_attn.o_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.59.self_attn.q_proj.bias": "model-00027-of-00037.safetensors",
+    "model.layers.59.self_attn.q_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.59.self_attn.v_proj.bias": "model-00027-of-00037.safetensors",
+    "model.layers.59.self_attn.v_proj.weight": "model-00027-of-00037.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00004-of-00037.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00004-of-00037.safetensors",
+    "model.layers.6.self_attn.k_proj.bias": "model-00004-of-00037.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.6.self_attn.q_proj.bias": "model-00004-of-00037.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.6.self_attn.v_proj.bias": "model-00004-of-00037.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.60.input_layernorm.weight": "model-00028-of-00037.safetensors",
+    "model.layers.60.mlp.down_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.60.mlp.gate_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.60.mlp.up_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.60.post_attention_layernorm.weight": "model-00028-of-00037.safetensors",
+    "model.layers.60.self_attn.k_proj.bias": "model-00028-of-00037.safetensors",
+    "model.layers.60.self_attn.k_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.60.self_attn.o_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.60.self_attn.q_proj.bias": "model-00028-of-00037.safetensors",
+    "model.layers.60.self_attn.q_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.60.self_attn.v_proj.bias": "model-00028-of-00037.safetensors",
+    "model.layers.60.self_attn.v_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.61.input_layernorm.weight": "model-00028-of-00037.safetensors",
+    "model.layers.61.mlp.down_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.61.mlp.gate_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.61.mlp.up_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.61.post_attention_layernorm.weight": "model-00028-of-00037.safetensors",
+    "model.layers.61.self_attn.k_proj.bias": "model-00028-of-00037.safetensors",
+    "model.layers.61.self_attn.k_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.61.self_attn.o_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.61.self_attn.q_proj.bias": "model-00028-of-00037.safetensors",
+    "model.layers.61.self_attn.q_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.61.self_attn.v_proj.bias": "model-00028-of-00037.safetensors",
+    "model.layers.61.self_attn.v_proj.weight": "model-00028-of-00037.safetensors",
+    "model.layers.62.input_layernorm.weight": "model-00029-of-00037.safetensors",
+    "model.layers.62.mlp.down_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.62.mlp.gate_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.62.mlp.up_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.62.post_attention_layernorm.weight": "model-00029-of-00037.safetensors",
+    "model.layers.62.self_attn.k_proj.bias": "model-00029-of-00037.safetensors",
+    "model.layers.62.self_attn.k_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.62.self_attn.o_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.62.self_attn.q_proj.bias": "model-00029-of-00037.safetensors",
+    "model.layers.62.self_attn.q_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.62.self_attn.v_proj.bias": "model-00029-of-00037.safetensors",
+    "model.layers.62.self_attn.v_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.63.input_layernorm.weight": "model-00029-of-00037.safetensors",
+    "model.layers.63.mlp.down_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.63.mlp.gate_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.63.mlp.up_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.63.post_attention_layernorm.weight": "model-00029-of-00037.safetensors",
+    "model.layers.63.self_attn.k_proj.bias": "model-00029-of-00037.safetensors",
+    "model.layers.63.self_attn.k_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.63.self_attn.o_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.63.self_attn.q_proj.bias": "model-00029-of-00037.safetensors",
+    "model.layers.63.self_attn.q_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.63.self_attn.v_proj.bias": "model-00029-of-00037.safetensors",
+    "model.layers.63.self_attn.v_proj.weight": "model-00029-of-00037.safetensors",
+    "model.layers.64.input_layernorm.weight": "model-00030-of-00037.safetensors",
+    "model.layers.64.mlp.down_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.64.mlp.gate_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.64.mlp.up_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.64.post_attention_layernorm.weight": "model-00030-of-00037.safetensors",
+    "model.layers.64.self_attn.k_proj.bias": "model-00030-of-00037.safetensors",
+    "model.layers.64.self_attn.k_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.64.self_attn.o_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.64.self_attn.q_proj.bias": "model-00030-of-00037.safetensors",
+    "model.layers.64.self_attn.q_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.64.self_attn.v_proj.bias": "model-00030-of-00037.safetensors",
+    "model.layers.64.self_attn.v_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.65.input_layernorm.weight": "model-00030-of-00037.safetensors",
+    "model.layers.65.mlp.down_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.65.mlp.gate_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.65.mlp.up_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.65.post_attention_layernorm.weight": "model-00030-of-00037.safetensors",
+    "model.layers.65.self_attn.k_proj.bias": "model-00030-of-00037.safetensors",
+    "model.layers.65.self_attn.k_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.65.self_attn.o_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.65.self_attn.q_proj.bias": "model-00030-of-00037.safetensors",
+    "model.layers.65.self_attn.q_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.65.self_attn.v_proj.bias": "model-00030-of-00037.safetensors",
+    "model.layers.65.self_attn.v_proj.weight": "model-00030-of-00037.safetensors",
+    "model.layers.66.input_layernorm.weight": "model-00030-of-00037.safetensors",
+    "model.layers.66.mlp.down_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.66.mlp.gate_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.66.mlp.up_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.66.post_attention_layernorm.weight": "model-00031-of-00037.safetensors",
+    "model.layers.66.self_attn.k_proj.bias": "model-00031-of-00037.safetensors",
+    "model.layers.66.self_attn.k_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.66.self_attn.o_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.66.self_attn.q_proj.bias": "model-00031-of-00037.safetensors",
+    "model.layers.66.self_attn.q_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.66.self_attn.v_proj.bias": "model-00031-of-00037.safetensors",
+    "model.layers.66.self_attn.v_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.67.input_layernorm.weight": "model-00031-of-00037.safetensors",
+    "model.layers.67.mlp.down_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.67.mlp.gate_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.67.mlp.up_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.67.post_attention_layernorm.weight": "model-00031-of-00037.safetensors",
+    "model.layers.67.self_attn.k_proj.bias": "model-00031-of-00037.safetensors",
+    "model.layers.67.self_attn.k_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.67.self_attn.o_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.67.self_attn.q_proj.bias": "model-00031-of-00037.safetensors",
+    "model.layers.67.self_attn.q_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.67.self_attn.v_proj.bias": "model-00031-of-00037.safetensors",
+    "model.layers.67.self_attn.v_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.68.input_layernorm.weight": "model-00031-of-00037.safetensors",
+    "model.layers.68.mlp.down_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.68.mlp.gate_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.68.mlp.up_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.68.post_attention_layernorm.weight": "model-00031-of-00037.safetensors",
+    "model.layers.68.self_attn.k_proj.bias": "model-00031-of-00037.safetensors",
+    "model.layers.68.self_attn.k_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.68.self_attn.o_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.68.self_attn.q_proj.bias": "model-00031-of-00037.safetensors",
+    "model.layers.68.self_attn.q_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.68.self_attn.v_proj.bias": "model-00031-of-00037.safetensors",
+    "model.layers.68.self_attn.v_proj.weight": "model-00031-of-00037.safetensors",
+    "model.layers.69.input_layernorm.weight": "model-00032-of-00037.safetensors",
+    "model.layers.69.mlp.down_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.69.mlp.gate_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.69.mlp.up_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.69.post_attention_layernorm.weight": "model-00032-of-00037.safetensors",
+    "model.layers.69.self_attn.k_proj.bias": "model-00032-of-00037.safetensors",
+    "model.layers.69.self_attn.k_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.69.self_attn.o_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.69.self_attn.q_proj.bias": "model-00032-of-00037.safetensors",
+    "model.layers.69.self_attn.q_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.69.self_attn.v_proj.bias": "model-00032-of-00037.safetensors",
+    "model.layers.69.self_attn.v_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00004-of-00037.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00004-of-00037.safetensors",
+    "model.layers.7.self_attn.k_proj.bias": "model-00004-of-00037.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.7.self_attn.q_proj.bias": "model-00004-of-00037.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.7.self_attn.v_proj.bias": "model-00004-of-00037.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00004-of-00037.safetensors",
+    "model.layers.70.input_layernorm.weight": "model-00032-of-00037.safetensors",
+    "model.layers.70.mlp.down_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.70.mlp.gate_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.70.mlp.up_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.70.post_attention_layernorm.weight": "model-00032-of-00037.safetensors",
+    "model.layers.70.self_attn.k_proj.bias": "model-00032-of-00037.safetensors",
+    "model.layers.70.self_attn.k_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.70.self_attn.o_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.70.self_attn.q_proj.bias": "model-00032-of-00037.safetensors",
+    "model.layers.70.self_attn.q_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.70.self_attn.v_proj.bias": "model-00032-of-00037.safetensors",
+    "model.layers.70.self_attn.v_proj.weight": "model-00032-of-00037.safetensors",
+    "model.layers.71.input_layernorm.weight": "model-00033-of-00037.safetensors",
+    "model.layers.71.mlp.down_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.71.mlp.gate_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.71.mlp.up_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.71.post_attention_layernorm.weight": "model-00033-of-00037.safetensors",
+    "model.layers.71.self_attn.k_proj.bias": "model-00033-of-00037.safetensors",
+    "model.layers.71.self_attn.k_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.71.self_attn.o_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.71.self_attn.q_proj.bias": "model-00033-of-00037.safetensors",
+    "model.layers.71.self_attn.q_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.71.self_attn.v_proj.bias": "model-00033-of-00037.safetensors",
+    "model.layers.71.self_attn.v_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.72.input_layernorm.weight": "model-00033-of-00037.safetensors",
+    "model.layers.72.mlp.down_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.72.mlp.gate_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.72.mlp.up_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.72.post_attention_layernorm.weight": "model-00033-of-00037.safetensors",
+    "model.layers.72.self_attn.k_proj.bias": "model-00033-of-00037.safetensors",
+    "model.layers.72.self_attn.k_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.72.self_attn.o_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.72.self_attn.q_proj.bias": "model-00033-of-00037.safetensors",
+    "model.layers.72.self_attn.q_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.72.self_attn.v_proj.bias": "model-00033-of-00037.safetensors",
+    "model.layers.72.self_attn.v_proj.weight": "model-00033-of-00037.safetensors",
+    "model.layers.73.input_layernorm.weight": "model-00034-of-00037.safetensors",
+    "model.layers.73.mlp.down_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.73.mlp.gate_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.73.mlp.up_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.73.post_attention_layernorm.weight": "model-00034-of-00037.safetensors",
+    "model.layers.73.self_attn.k_proj.bias": "model-00034-of-00037.safetensors",
+    "model.layers.73.self_attn.k_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.73.self_attn.o_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.73.self_attn.q_proj.bias": "model-00034-of-00037.safetensors",
+    "model.layers.73.self_attn.q_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.73.self_attn.v_proj.bias": "model-00034-of-00037.safetensors",
+    "model.layers.73.self_attn.v_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.74.input_layernorm.weight": "model-00034-of-00037.safetensors",
+    "model.layers.74.mlp.down_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.74.mlp.gate_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.74.mlp.up_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.74.post_attention_layernorm.weight": "model-00034-of-00037.safetensors",
+    "model.layers.74.self_attn.k_proj.bias": "model-00034-of-00037.safetensors",
+    "model.layers.74.self_attn.k_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.74.self_attn.o_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.74.self_attn.q_proj.bias": "model-00034-of-00037.safetensors",
+    "model.layers.74.self_attn.q_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.74.self_attn.v_proj.bias": "model-00034-of-00037.safetensors",
+    "model.layers.74.self_attn.v_proj.weight": "model-00034-of-00037.safetensors",
+    "model.layers.75.input_layernorm.weight": "model-00034-of-00037.safetensors",
+    "model.layers.75.mlp.down_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.75.mlp.gate_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.75.mlp.up_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.75.post_attention_layernorm.weight": "model-00035-of-00037.safetensors",
+    "model.layers.75.self_attn.k_proj.bias": "model-00035-of-00037.safetensors",
+    "model.layers.75.self_attn.k_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.75.self_attn.o_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.75.self_attn.q_proj.bias": "model-00035-of-00037.safetensors",
+    "model.layers.75.self_attn.q_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.75.self_attn.v_proj.bias": "model-00035-of-00037.safetensors",
+    "model.layers.75.self_attn.v_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.76.input_layernorm.weight": "model-00035-of-00037.safetensors",
+    "model.layers.76.mlp.down_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.76.mlp.gate_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.76.mlp.up_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.76.post_attention_layernorm.weight": "model-00035-of-00037.safetensors",
+    "model.layers.76.self_attn.k_proj.bias": "model-00035-of-00037.safetensors",
+    "model.layers.76.self_attn.k_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.76.self_attn.o_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.76.self_attn.q_proj.bias": "model-00035-of-00037.safetensors",
+    "model.layers.76.self_attn.q_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.76.self_attn.v_proj.bias": "model-00035-of-00037.safetensors",
+    "model.layers.76.self_attn.v_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.77.input_layernorm.weight": "model-00035-of-00037.safetensors",
+    "model.layers.77.mlp.down_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.77.mlp.gate_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.77.mlp.up_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.77.post_attention_layernorm.weight": "model-00035-of-00037.safetensors",
+    "model.layers.77.self_attn.k_proj.bias": "model-00035-of-00037.safetensors",
+    "model.layers.77.self_attn.k_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.77.self_attn.o_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.77.self_attn.q_proj.bias": "model-00035-of-00037.safetensors",
+    "model.layers.77.self_attn.q_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.77.self_attn.v_proj.bias": "model-00035-of-00037.safetensors",
+    "model.layers.77.self_attn.v_proj.weight": "model-00035-of-00037.safetensors",
+    "model.layers.78.input_layernorm.weight": "model-00036-of-00037.safetensors",
+    "model.layers.78.mlp.down_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.78.mlp.gate_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.78.mlp.up_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.78.post_attention_layernorm.weight": "model-00036-of-00037.safetensors",
+    "model.layers.78.self_attn.k_proj.bias": "model-00036-of-00037.safetensors",
+    "model.layers.78.self_attn.k_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.78.self_attn.o_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.78.self_attn.q_proj.bias": "model-00036-of-00037.safetensors",
+    "model.layers.78.self_attn.q_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.78.self_attn.v_proj.bias": "model-00036-of-00037.safetensors",
+    "model.layers.78.self_attn.v_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.79.input_layernorm.weight": "model-00036-of-00037.safetensors",
+    "model.layers.79.mlp.down_proj.weight": "model-00037-of-00037.safetensors",
+    "model.layers.79.mlp.gate_proj.weight": "model-00037-of-00037.safetensors",
+    "model.layers.79.mlp.up_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.79.post_attention_layernorm.weight": "model-00036-of-00037.safetensors",
+    "model.layers.79.self_attn.k_proj.bias": "model-00036-of-00037.safetensors",
+    "model.layers.79.self_attn.k_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.79.self_attn.o_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.79.self_attn.q_proj.bias": "model-00036-of-00037.safetensors",
+    "model.layers.79.self_attn.q_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.79.self_attn.v_proj.bias": "model-00036-of-00037.safetensors",
+    "model.layers.79.self_attn.v_proj.weight": "model-00036-of-00037.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00005-of-00037.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00005-of-00037.safetensors",
+    "model.layers.8.self_attn.k_proj.bias": "model-00005-of-00037.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.8.self_attn.q_proj.bias": "model-00005-of-00037.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.8.self_attn.v_proj.bias": "model-00005-of-00037.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00005-of-00037.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00006-of-00037.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00005-of-00037.safetensors",
+    "model.layers.9.self_attn.k_proj.bias": "model-00005-of-00037.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.9.self_attn.q_proj.bias": "model-00005-of-00037.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00005-of-00037.safetensors",
+    "model.layers.9.self_attn.v_proj.bias": "model-00005-of-00037.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00005-of-00037.safetensors",
+    "model.norm.weight": "model-00037-of-00037.safetensors",
+    "score.0.bias": "model-00001-of-00037.safetensors",
+    "score.0.weight": "model-00001-of-00037.safetensors",
+    "score.2.bias": "model-00001-of-00037.safetensors",
+    "score.2.weight": "model-00001-of-00037.safetensors"
+  }
+}

modeling_qwen2_rm.py

@@ -0,0 +1,1549 @@
+# coding=utf-8
+# Copyright 2024 The Qwen team, Alibaba Group and the HuggingFace Inc. team. All rights reserved.
+#
+# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
+# and OPT implementations in this library. It has been modified from its
+# original forms to accommodate minor architectural differences compared
+# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""PyTorch Qwen2 model."""
+
+import math
+from typing import List, Optional, Tuple, Union
+
+import torch
+import torch.utils.checkpoint
+from torch import nn
+from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss
+
+from transformers.activations import ACT2FN
+from transformers.cache_utils import Cache, DynamicCache#, StaticCache
+from transformers.modeling_attn_mask_utils import AttentionMaskConverter
+from transformers.modeling_outputs import (
+    BaseModelOutputWithPast,
+    CausalLMOutputWithPast,
+    SequenceClassifierOutputWithPast,
+    TokenClassifierOutput,
+)
+from transformers.modeling_utils import PreTrainedModel
+from transformers.utils import (
+    add_start_docstrings,
+    add_start_docstrings_to_model_forward,
+    is_flash_attn_2_available,
+    is_flash_attn_greater_or_equal_2_10,
+    logging,
+    replace_return_docstrings,
+)
+from configuration_qwen2_rm import Qwen2RMConfig as Qwen2Config
+
+
+# if is_flash_attn_2_available():
+#     from transformers.modeling_flash_attention_utils import _flash_attention_forward
+
+
+logger = logging.get_logger(__name__)
+
+
+_CHECKPOINT_FOR_DOC = "Qwen/Qwen2-7B-beta"
+_CONFIG_FOR_DOC = "Qwen2Config"
+
+
+# Copied from transformers.models.llama.modeling_llama._prepare_4d_causal_attention_mask_with_cache_position
+def _prepare_4d_causal_attention_mask_with_cache_position(
+    attention_mask: torch.Tensor,
+    sequence_length: int,
+    target_length: int,
+    dtype: torch.dtype,
+    device: torch.device,
+    min_dtype: float,
+    cache_position: torch.Tensor,
+    batch_size: int,
+):
+    """
+    Creates a causal 4D mask of shape `(batch_size, 1, query_length, key_value_length)` from a 2D mask of shape
+    `(batch_size, key_value_length)`, or if the input `attention_mask` is already 4D, do nothing.
+
+    Args:
+        attention_mask (`torch.Tensor`):
+            A 2D attention mask of shape `(batch_size, key_value_length)` or a 4D attention mask of shape `(batch_size, 1, query_length, key_value_length)`.
+        sequence_length (`int`):
+            The sequence length being processed.
+        target_length (`int`):
+            The target length: when generating with static cache, the mask should be as long as the static cache, to account for the 0 padding, the part of the cache that is not filled yet.
+        dtype (`torch.dtype`):
+            The dtype to use for the 4D attention mask.
+        device (`torch.device`):
+            The device to plcae the 4D attention mask on.
+        min_dtype (`float`):
+            The minimum value representable with the dtype `dtype`.
+        cache_position (`torch.Tensor`):
+            Indices depicting the position of the input sequence tokens in the sequence.
+        batch_size (`torch.Tensor`):
+            Batch size.
+    """
+    if attention_mask is not None and attention_mask.dim() == 4:
+        # In this case we assume that the mask comes already in inverted form and requires no inversion or slicing.
+        causal_mask = attention_mask
+    else:
+        causal_mask = torch.full((sequence_length, target_length), fill_value=min_dtype, dtype=dtype, device=device)
+        if sequence_length != 1:
+            causal_mask = torch.triu(causal_mask, diagonal=1)
+        causal_mask *= torch.arange(target_length, device=device) > cache_position.reshape(-1, 1)
+        causal_mask = causal_mask[None, None, :, :].expand(batch_size, 1, -1, -1)
+        if attention_mask is not None:
+            causal_mask = causal_mask.clone()  # copy to contiguous memory for in-place edit
+            mask_length = attention_mask.shape[-1]
+            padding_mask = causal_mask[:, :, :, :mask_length] + attention_mask[:, None, None, :]
+            padding_mask = padding_mask == 0
+            causal_mask[:, :, :, :mask_length] = causal_mask[:, :, :, :mask_length].masked_fill(
+                padding_mask, min_dtype
+            )
+
+    return causal_mask
+
+
+# Copied from transformers.models.llama.modeling_llama.LlamaRMSNorm with Llama->Qwen2
+class Qwen2RMSNorm(nn.Module):
+    def __init__(self, hidden_size, eps=1e-6):
+        """
+        Qwen2RMSNorm is equivalent to T5LayerNorm
+        """
+        super().__init__()
+        self.weight = nn.Parameter(torch.ones(hidden_size))
+        self.variance_epsilon = eps
+
+    def forward(self, hidden_states):
+        input_dtype = hidden_states.dtype
+        hidden_states = hidden_states.to(torch.float32)
+        variance = hidden_states.pow(2).mean(-1, keepdim=True)
+        hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon)
+        return self.weight * hidden_states.to(input_dtype)
+
+    def extra_repr(self):
+        return f"{tuple(self.weight.shape)}, eps={self.variance_epsilon}"
+
+
+# Copied from transformers.models.mixtral.modeling_mixtral.MixtralRotaryEmbedding with Mixtral->Qwen2
+class Qwen2RotaryEmbedding(nn.Module):
+    def __init__(self, dim, max_position_embeddings=2048, base=10000, device=None):
+        super().__init__()
+
+        self.dim = dim
+        self.max_position_embeddings = max_position_embeddings
+        self.base = base
+        inv_freq = 1.0 / (self.base ** (torch.arange(0, self.dim, 2, dtype=torch.int64).float().to(device) / self.dim))
+        self.register_buffer("inv_freq", inv_freq, persistent=False)
+
+        # Build here to make `torch.jit.trace` work.
+        self._set_cos_sin_cache(
+            seq_len=max_position_embeddings, device=self.inv_freq.device, dtype=torch.get_default_dtype()
+        )
+
+    def _set_cos_sin_cache(self, seq_len, device, dtype):
+        self.max_seq_len_cached = seq_len
+        t = torch.arange(self.max_seq_len_cached, device=device, dtype=torch.int64).type_as(self.inv_freq)
+
+        freqs = torch.outer(t, self.inv_freq)
+        # Different from paper, but it uses a different permutation in order to obtain the same calculation
+        emb = torch.cat((freqs, freqs), dim=-1)
+        self.register_buffer("cos_cached", emb.cos().to(dtype), persistent=False)
+        self.register_buffer("sin_cached", emb.sin().to(dtype), persistent=False)
+
+    def forward(self, x, seq_len=None):
+        # x: [bs, num_attention_heads, seq_len, head_size]
+        if seq_len > self.max_seq_len_cached:
+            self._set_cos_sin_cache(seq_len=seq_len, device=x.device, dtype=x.dtype)
+
+        return (
+            self.cos_cached[:seq_len].to(dtype=x.dtype),
+            self.sin_cached[:seq_len].to(dtype=x.dtype),
+        )
+
+
+# Copied from transformers.models.llama.modeling_llama.rotate_half
+def rotate_half(x):
+    """Rotates half the hidden dims of the input."""
+    x1 = x[..., : x.shape[-1] // 2]
+    x2 = x[..., x.shape[-1] // 2 :]
+    return torch.cat((-x2, x1), dim=-1)
+
+
+# Copied from transformers.models.mixtral.modeling_mixtral.apply_rotary_pos_emb
+def apply_rotary_pos_emb(q, k, cos, sin, position_ids, unsqueeze_dim=1):
+    """Applies Rotary Position Embedding to the query and key tensors.
+
+    Args:
+        q (`torch.Tensor`): The query tensor.
+        k (`torch.Tensor`): The key tensor.
+        cos (`torch.Tensor`): The cosine part of the rotary embedding.
+        sin (`torch.Tensor`): The sine part of the rotary embedding.
+        position_ids (`torch.Tensor`):
+            The position indices of the tokens corresponding to the query and key tensors. For example, this can be
+            used to pass offsetted position ids when working with a KV-cache.
+        unsqueeze_dim (`int`, *optional*, defaults to 1):
+            The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and
+            sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note
+            that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and
+            k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes
+            cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have
+            the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2.
+    Returns:
+        `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding.
+    """
+    cos = cos[position_ids].unsqueeze(unsqueeze_dim)
+    sin = sin[position_ids].unsqueeze(unsqueeze_dim)
+    q_embed = (q * cos) + (rotate_half(q) * sin)
+    k_embed = (k * cos) + (rotate_half(k) * sin)
+    return q_embed, k_embed
+
+
+# Copied from transformers.models.mistral.modeling_mistral.MistralMLP with Mistral->Qwen2
+class Qwen2MLP(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.hidden_size = config.hidden_size
+        self.intermediate_size = config.intermediate_size
+        self.gate_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=False)
+        self.up_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=False)
+        self.down_proj = nn.Linear(self.intermediate_size, self.hidden_size, bias=False)
+        self.act_fn = ACT2FN[config.hidden_act]
+
+    def forward(self, hidden_state):
+        return self.down_proj(self.act_fn(self.gate_proj(hidden_state)) * self.up_proj(hidden_state))
+
+
+# Copied from transformers.models.llama.modeling_llama.repeat_kv
+def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
+    """
+    This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch,
+    num_key_value_heads, seqlen, head_dim) to (batch, num_attention_heads, seqlen, head_dim)
+    """
+    batch, num_key_value_heads, slen, head_dim = hidden_states.shape
+    if n_rep == 1:
+        return hidden_states
+    hidden_states = hidden_states[:, :, None, :, :].expand(batch, num_key_value_heads, n_rep, slen, head_dim)
+    return hidden_states.reshape(batch, num_key_value_heads * n_rep, slen, head_dim)
+
+
+class Qwen2Attention(nn.Module):
+    """
+    Multi-headed attention from 'Attention Is All You Need' paper. Modified to use sliding window attention: Longformer
+    and "Generating Long Sequences with Sparse Transformers".
+    """
+
+    def __init__(self, config: Qwen2Config, layer_idx: Optional[int] = None):
+        super().__init__()
+        self.config = config
+        self.layer_idx = layer_idx
+        if layer_idx is None:
+            logger.warning_once(
+                f"Instantiating {self.__class__.__name__} without passing `layer_idx` is not recommended and will "
+                "to errors during the forward call, if caching is used. Please make sure to provide a `layer_idx` "
+                "when creating this class."
+            )
+
+        self.hidden_size = config.hidden_size
+        self.num_heads = config.num_attention_heads
+        self.head_dim = self.hidden_size // self.num_heads
+        self.num_key_value_heads = config.num_key_value_heads
+        self.num_key_value_groups = self.num_heads // self.num_key_value_heads
+        self.max_position_embeddings = config.max_position_embeddings
+        self.rope_theta = config.rope_theta
+        self.is_causal = True
+        self.attention_dropout = config.attention_dropout
+
+        if (self.head_dim * self.num_heads) != self.hidden_size:
+            raise ValueError(
+                f"hidden_size must be divisible by num_heads (got `hidden_size`: {self.hidden_size}"
+                f" and `num_heads`: {self.num_heads})."
+            )
+        self.q_proj = nn.Linear(self.hidden_size, self.num_heads * self.head_dim, bias=True)
+        self.k_proj = nn.Linear(self.hidden_size, self.num_key_value_heads * self.head_dim, bias=True)
+        self.v_proj = nn.Linear(self.hidden_size, self.num_key_value_heads * self.head_dim, bias=True)
+        self.o_proj = nn.Linear(self.num_heads * self.head_dim, self.hidden_size, bias=False)
+
+        self.rotary_emb = Qwen2RotaryEmbedding(
+            self.head_dim,
+            max_position_embeddings=self.max_position_embeddings,
+            base=self.rope_theta,
+        )
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Cache] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+        cache_position: Optional[torch.LongTensor] = None,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states)
+        value_states = self.v_proj(hidden_states)
+
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        kv_seq_len = key_states.shape[-2]
+        if past_key_value is not None:
+            if self.layer_idx is None:
+                raise ValueError(
+                    f"The cache structure has changed since version v4.36. If you are using {self.__class__.__name__} "
+                    "for auto-regressive decoding with k/v caching, please make sure to initialize the attention class "
+                    "with a layer index."
+                )
+            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
+        cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
+        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
+
+        if past_key_value is not None:
+            cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position}  # Specific to RoPE models
+            key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
+
+        # repeat k/v heads if n_kv_heads < n_heads
+        key_states = repeat_kv(key_states, self.num_key_value_groups)
+        value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+        attn_weights = torch.matmul(query_states, key_states.transpose(2, 3)) / math.sqrt(self.head_dim)
+
+        if attn_weights.size() != (bsz, self.num_heads, q_len, kv_seq_len):
+            raise ValueError(
+                f"Attention weights should be of size {(bsz, self.num_heads, q_len, kv_seq_len)}, but is"
+                f" {attn_weights.size()}"
+            )
+
+        if attention_mask is not None:  # no matter the length, we just slice it
+            causal_mask = attention_mask[:, :, :, : key_states.shape[-2]]
+            attn_weights = attn_weights + causal_mask
+
+        # upcast attention to fp32
+        attn_weights = nn.functional.softmax(attn_weights, dim=-1, dtype=torch.float32).to(query_states.dtype)
+        attn_weights = nn.functional.dropout(attn_weights, p=self.attention_dropout, training=self.training)
+        attn_output = torch.matmul(attn_weights, value_states)
+
+        if attn_output.size() != (bsz, self.num_heads, q_len, self.head_dim):
+            raise ValueError(
+                f"`attn_output` should be of size {(bsz, self.num_heads, q_len, self.head_dim)}, but is"
+                f" {attn_output.size()}"
+            )
+
+        attn_output = attn_output.transpose(1, 2).contiguous()
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
+
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+
+
+class Qwen2FlashAttention2(Qwen2Attention):
+    """
+    Qwen2 flash attention module, following Qwen2 attention module. This module inherits from `Qwen2Attention`
+    as the weights of the module stays untouched. The only required change would be on the forward pass
+    where it needs to correctly call the public API of flash attention and deal with padding tokens
+    in case the input contains any of them. Additionally, for sliding window attention, we apply SWA only to the bottom
+    config.max_window_layers layers.
+    """
+
+    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+
+        # TODO: Should be removed once Flash Attention for RoCm is bumped to 2.1.
+        # flash_attn<2.1 generates top-left aligned causal mask, while what is needed here is bottom-right alignement, that was made default for flash_attn>=2.1. This attribute is used to handle this difference. Reference: https://github.com/Dao-AILab/flash-attention/releases/tag/v2.1.0.
+        # Beware that with flash_attn<2.1, using q_seqlen != k_seqlen (except for the case q_seqlen == 1) produces a wrong mask (top-left).
+        self._flash_attn_uses_top_left_mask = not is_flash_attn_greater_or_equal_2_10()
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Cache] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+        cache_position: Optional[torch.LongTensor] = None,
+    ):
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states)
+        value_states = self.v_proj(hidden_states)
+
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        kv_seq_len = key_states.shape[-2]
+        if past_key_value is not None:
+            if self.layer_idx is None:
+                raise ValueError(
+                    f"The cache structure has changed since version v4.36. If you are using {self.__class__.__name__} "
+                    "for auto-regressive decoding with k/v caching, please make sure to initialize the attention class "
+                    "with a layer index."
+                )
+            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
+
+        # Because the input can be padded, the absolute sequence length depends on the max position id.
+        rotary_seq_len = (
+            max(kv_seq_len, position_ids[:, -1].max().item() + 1) if position_ids is not None else kv_seq_len
+        )
+
+        cos, sin = self.rotary_emb(value_states, seq_len=rotary_seq_len)
+
+        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
+
+        if past_key_value is not None:
+            # Activate slicing cache only if the config has a value `sliding_windows` attribute
+            cache_has_contents = past_key_value.get_seq_length(self.layer_idx) > 0
+            if (
+                getattr(self.config, "sliding_window", None) is not None
+                and kv_seq_len > self.config.sliding_window
+                and cache_has_contents
+            ):
+                slicing_tokens = 1 - self.config.sliding_window
+
+                past_key = past_key_value[self.layer_idx][0]
+                past_value = past_key_value[self.layer_idx][1]
+
+                past_key = past_key[:, :, slicing_tokens:, :].contiguous()
+                past_value = past_value[:, :, slicing_tokens:, :].contiguous()
+
+                if past_key.shape[-2] != self.config.sliding_window - 1:
+                    raise ValueError(
+                        f"past key must have a shape of (`batch_size, num_heads, self.config.sliding_window-1, head_dim`), got"
+                        f" {past_key.shape}"
+                    )
+
+                if attention_mask is not None:
+                    attention_mask = attention_mask[:, slicing_tokens:]
+                    attention_mask = torch.cat([attention_mask, torch.ones_like(attention_mask[:, -1:])], dim=-1)
+
+            cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position}  # Specific to RoPE models
+            key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
+
+        # repeat k/v heads if n_kv_heads < n_heads
+        key_states = repeat_kv(key_states, self.num_key_value_groups)
+        value_states = repeat_kv(value_states, self.num_key_value_groups)
+        dropout_rate = 0.0 if not self.training else self.attention_dropout
+
+        # In PEFT, usually we cast the layer norms in float32 for training stability reasons
+        # therefore the input hidden states gets silently casted in float32. Hence, we need
+        # cast them back in float16 just to be sure everything works as expected.
+        input_dtype = query_states.dtype
+        if input_dtype == torch.float32:
+            if torch.is_autocast_enabled():
+                target_dtype = torch.get_autocast_gpu_dtype()
+            # Handle the case where the model is quantized
+            elif hasattr(self.config, "_pre_quantization_dtype"):
+                target_dtype = self.config._pre_quantization_dtype
+            else:
+                target_dtype = self.q_proj.weight.dtype
+
+            logger.warning_once(
+                f"The input hidden states seems to be silently casted in float32, this might be related to"
+                f" the fact you have upcasted embedding or layer norm layers in float32. We will cast back the input in"
+                f" {target_dtype}."
+            )
+
+            query_states = query_states.to(target_dtype)
+            key_states = key_states.to(target_dtype)
+            value_states = value_states.to(target_dtype)
+
+        # Reashape to the expected shape for Flash Attention
+        query_states = query_states.transpose(1, 2)
+        key_states = key_states.transpose(1, 2)
+        value_states = value_states.transpose(1, 2)
+
+        if (
+            self.config.use_sliding_window
+            and getattr(self.config, "sliding_window", None) is not None
+            and self.layer_idx >= self.config.max_window_layers
+        ):
+            sliding_window = self.config.sliding_window
+        else:
+            sliding_window = None
+
+        attn_output = _flash_attention_forward(
+            query_states,
+            key_states,
+            value_states,
+            attention_mask,
+            q_len,
+            position_ids=position_ids,
+            dropout=dropout_rate,
+            sliding_window=sliding_window,
+            is_causal=self.is_causal,
+            use_top_left_mask=self._flash_attn_uses_top_left_mask,
+        )
+
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size).contiguous()
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+
+
+# Copied from transformers.models.mixtral.modeling_mixtral.MixtralSdpaAttention with Mixtral->Qwen2
+class Qwen2SdpaAttention(Qwen2Attention):
+    """
+    Qwen2 attention module using torch.nn.functional.scaled_dot_product_attention. This module inherits from
+    `Qwen2Attention` as the weights of the module stays untouched. The only changes are on the forward pass to adapt to
+    SDPA API.
+    """
+
+    # Adapted from Qwen2Attention.forward
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Cache] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+        cache_position: Optional[torch.LongTensor] = None,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        if output_attentions:
+            # TODO: Improve this warning with e.g. `model.config.attn_implementation = "manual"` once this is implemented.
+            logger.warning_once(
+                "Qwen2Model is using Qwen2SdpaAttention, but `torch.nn.functional.scaled_dot_product_attention` does not support `output_attentions=True`. Falling back to the manual attention implementation, "
+                'but specifying the manual implementation will be required from Transformers version v5.0.0 onwards. This warning can be removed using the argument `attn_implementation="eager"` when loading the model.'
+            )
+            return super().forward(
+                hidden_states=hidden_states,
+                attention_mask=attention_mask,
+                position_ids=position_ids,
+                past_key_value=past_key_value,
+                output_attentions=output_attentions,
+                use_cache=use_cache,
+            )
+
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states)
+        value_states = self.v_proj(hidden_states)
+
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        kv_seq_len = key_states.shape[-2]
+        if past_key_value is not None:
+            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
+        cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
+
+        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
+
+        if past_key_value is not None:
+            cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position}  # Specific to RoPE models
+            key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
+
+        key_states = repeat_kv(key_states, self.num_key_value_groups)
+        value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+        causal_mask = attention_mask
+        if attention_mask is not None:  # no matter the length, we just slice it
+            causal_mask = attention_mask[:, :, :, : key_states.shape[-2]]
+
+        # SDPA with memory-efficient backend is currently (torch==2.1.2) bugged with non-contiguous inputs with custom attn_mask,
+        # Reference: https://github.com/pytorch/pytorch/issues/112577.
+        if query_states.device.type == "cuda" and attention_mask is not None:
+            query_states = query_states.contiguous()
+            key_states = key_states.contiguous()
+            value_states = value_states.contiguous()
+
+        # We dispatch to SDPA's Flash Attention or Efficient kernels via this `is_causal` if statement instead of an inline conditional assignment
+        # in SDPA to support both torch.compile's dynamic shapes and full graph options. An inline conditional prevents dynamic shapes from compiling.
+        # The q_len > 1 is necessary to match with AttentionMaskConverter.to_causal_4d that does not create a causal mask in case q_len == 1.
+        is_causal = True if causal_mask is None and q_len > 1 else False
+
+        attn_output = torch.nn.functional.scaled_dot_product_attention(
+            query_states,
+            key_states,
+            value_states,
+            attn_mask=causal_mask,
+            dropout_p=self.attention_dropout if self.training else 0.0,
+            is_causal=is_causal,
+        )
+
+        attn_output = attn_output.transpose(1, 2).contiguous()
+        attn_output = attn_output.view(bsz, q_len, self.hidden_size)
+
+        attn_output = self.o_proj(attn_output)
+
+        return attn_output, None, past_key_value
+
+
+QWEN2_ATTENTION_CLASSES = {
+    "eager": Qwen2Attention,
+    "flash_attention_2": Qwen2FlashAttention2,
+    "sdpa": Qwen2SdpaAttention,
+}
+
+
+class Qwen2DecoderLayer(nn.Module):
+    def __init__(self, config: Qwen2Config, layer_idx: int):
+        super().__init__()
+        self.hidden_size = config.hidden_size
+
+        if config.sliding_window and config._attn_implementation != "flash_attention_2":
+            logger.warning_once(
+                f"Sliding Window Attention is enabled but not implemented for `{config._attn_implementation}`; "
+                "unexpected results may be encountered."
+            )
+        self.self_attn = QWEN2_ATTENTION_CLASSES[config._attn_implementation](config, layer_idx)
+
+        self.mlp = Qwen2MLP(config)
+        self.input_layernorm = Qwen2RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+        self.post_attention_layernorm = Qwen2RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Tuple[torch.Tensor]] = None,
+        output_attentions: Optional[bool] = False,
+        use_cache: Optional[bool] = False,
+        cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
+    ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
+        """
+        Args:
+            hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
+            attention_mask (`torch.FloatTensor`, *optional*): attention mask of size
+                `(batch, sequence_length)` where padding elements are indicated by 0.
+            output_attentions (`bool`, *optional*):
+                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
+                returned tensors for more detail.
+            use_cache (`bool`, *optional*):
+                If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
+                (see `past_key_values`).
+            past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
+            cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
+                Indices depicting the position of the input sequence tokens in the sequence.
+            kwargs (`dict`, *optional*):
+                Arbitrary kwargs to be ignored, used for FSDP and other methods that injects code
+                into the model
+        """
+
+        residual = hidden_states
+
+        hidden_states = self.input_layernorm(hidden_states)
+
+        # Self Attention
+        hidden_states, self_attn_weights, present_key_value = self.self_attn(
+            hidden_states=hidden_states,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_value=past_key_value,
+            output_attentions=output_attentions,
+            use_cache=use_cache,
+            cache_position=cache_position,
+        )
+        hidden_states = residual + hidden_states
+
+        # Fully Connected
+        residual = hidden_states
+        hidden_states = self.post_attention_layernorm(hidden_states)
+        hidden_states = self.mlp(hidden_states)
+        hidden_states = residual + hidden_states
+
+        outputs = (hidden_states,)
+
+        if output_attentions:
+            outputs += (self_attn_weights,)
+
+        if use_cache:
+            outputs += (present_key_value,)
+
+        return outputs
+
+
+QWEN2_START_DOCSTRING = r"""
+    This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic methods the
+    library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
+    etc.)
+
+    This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass.
+    Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage
+    and behavior.
+
+    Parameters:
+        config ([`Qwen2Config`]):
+            Model configuration class with all the parameters of the model. Initializing with a config file does not
+            load the weights associated with the model, only the configuration. Check out the
+            [`~PreTrainedModel.from_pretrained`] method to load the model weights.
+"""
+
+
+@add_start_docstrings(
+    "The bare Qwen2 Model outputting raw hidden-states without any specific head on top.",
+    QWEN2_START_DOCSTRING,
+)
+class Qwen2PreTrainedModel(PreTrainedModel):
+    config_class = Qwen2Config
+    base_model_prefix = "model"
+    supports_gradient_checkpointing = True
+    _no_split_modules = ["Qwen2DecoderLayer"]
+    _skip_keys_device_placement = "past_key_values"
+    _supports_flash_attn_2 = True
+    _supports_sdpa = True
+    _supports_cache_class = True
+
+    def _init_weights(self, module):
+        std = self.config.initializer_range
+        if isinstance(module, nn.Linear):
+            module.weight.data.normal_(mean=0.0, std=std)
+            if module.bias is not None:
+                module.bias.data.zero_()
+        elif isinstance(module, nn.Embedding):
+            module.weight.data.normal_(mean=0.0, std=std)
+            if module.padding_idx is not None:
+                module.weight.data[module.padding_idx].zero_()
+
+
+QWEN2_INPUTS_DOCSTRING = r"""
+    Args:
+        input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
+            Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide
+            it.
+
+            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
+            [`PreTrainedTokenizer.__call__`] for details.
+
+            [What are input IDs?](../glossary#input-ids)
+        attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
+
+            - 1 for tokens that are **not masked**,
+            - 0 for tokens that are **masked**.
+
+            [What are attention masks?](../glossary#attention-mask)
+
+            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
+            [`PreTrainedTokenizer.__call__`] for details.
+
+            If `past_key_values` is used, optionally only the last `decoder_input_ids` have to be input (see
+            `past_key_values`).
+
+            If you want to change padding behavior, you should read [`modeling_opt._prepare_decoder_attention_mask`]
+            and modify to your needs. See diagram 1 in [the paper](https://arxiv.org/abs/1910.13461) for more
+            information on the default strategy.
+
+            - 1 indicates the head is **not masked**,
+            - 0 indicates the head is **masked**.
+        position_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0,
+            config.n_positions - 1]`.
+
+            [What are position IDs?](../glossary#position-ids)
+        past_key_values (`Cache` or `tuple(tuple(torch.FloatTensor))`, *optional*):
+            Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention
+            blocks) that can be used to speed up sequential decoding. This typically consists in the `past_key_values`
+            returned by the model at a previous stage of decoding, when `use_cache=True` or `config.use_cache=True`.
+
+            Two formats are allowed:
+            - a [`~cache_utils.Cache`] instance;
+            - Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of
+            shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`). This is also known as the legacy
+            cache format.
+
+            The model will output the same cache format that is fed as input. If no `past_key_values` are passed, the
+            legacy cache format will be returned.
+
+            If `past_key_values` are used, the user can optionally input only the last `input_ids` (those that don't
+            have their past key value states given to this model) of shape `(batch_size, 1)` instead of all `input_ids`
+            of shape `(batch_size, sequence_length)`.
+        inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
+            Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This
+            is useful if you want more control over how to convert `input_ids` indices into associated vectors than the
+            model's internal embedding lookup matrix.
+        use_cache (`bool`, *optional*):
+            If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see
+            `past_key_values`).
+        output_attentions (`bool`, *optional*):
+            Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
+            tensors for more detail.
+        output_hidden_states (`bool`, *optional*):
+            Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
+            more detail.
+        return_dict (`bool`, *optional*):
+            Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
+        cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
+            Indices depicting the position of the input sequence tokens in the sequence. Contrarily to `position_ids`,
+            this tensor is not affected by padding. It is used to update the cache in the correct position and to infer
+            the complete sequence length.
+"""
+
+
+@add_start_docstrings(
+    "The bare Qwen2 Model outputting raw hidden-states without any specific head on top.",
+    QWEN2_START_DOCSTRING,
+)
+class Qwen2Model(Qwen2PreTrainedModel):
+    """
+    Transformer decoder consisting of *config.num_hidden_layers* layers. Each layer is a [`Qwen2DecoderLayer`]
+
+    Args:
+        config: Qwen2Config
+    """
+
+    def __init__(self, config: Qwen2Config):
+        super().__init__(config)
+        self.padding_idx = config.pad_token_id
+        self.vocab_size = config.vocab_size
+
+        self.embed_tokens = nn.Embedding(config.vocab_size, config.hidden_size, self.padding_idx)
+        self.layers = nn.ModuleList(
+            [Qwen2DecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
+        )
+        self._attn_implementation = config._attn_implementation
+        self.norm = Qwen2RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+        self.gradient_checkpointing = False
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.embed_tokens
+
+    def set_input_embeddings(self, value):
+        self.embed_tokens = value
+
+    @add_start_docstrings_to_model_forward(QWEN2_INPUTS_DOCSTRING)
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+        cache_position: Optional[torch.LongTensor] = None,
+    ) -> Union[Tuple, BaseModelOutputWithPast]:
+        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        if (input_ids is None) ^ (inputs_embeds is not None):
+            raise ValueError(
+                "You cannot specify both input_ids and inputs_embeds at the same time, and must specify either one"
+            )
+
+        if self.gradient_checkpointing and self.training:
+            if use_cache:
+                logger.warning_once(
+                    "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
+                )
+                use_cache = False
+
+        use_legacy_cache = False
+        if use_cache and not isinstance(past_key_values, Cache) and not self.training:
+            use_legacy_cache = True
+            past_key_values = DynamicCache.from_legacy_cache(past_key_values)
+            logger.warning_once(
+                "We detected that you are passing `past_key_values` as a tuple and this is deprecated and will be removed in v4.43. "
+                "Please use an appropriate `Cache` class (https://huggingface.co/docs/transformers/v4.41.3/en/internal/generation_utils#transformers.Cache)"
+            )
+
+        if inputs_embeds is None:
+            inputs_embeds = self.embed_tokens(input_ids)
+
+        if cache_position is None:
+            past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0
+            cache_position = torch.arange(
+                past_seen_tokens, past_seen_tokens + inputs_embeds.shape[1], device=inputs_embeds.device
+            )
+        if position_ids is None:
+            position_ids = cache_position.unsqueeze(0)
+
+        causal_mask = self._update_causal_mask(
+            attention_mask, inputs_embeds, cache_position, past_key_values, output_attentions
+        )
+
+        hidden_states = inputs_embeds
+
+        # decoder layers
+        all_hidden_states = () if output_hidden_states else None
+        all_self_attns = () if output_attentions else None
+        next_decoder_cache = None
+
+        for decoder_layer in self.layers:
+            if output_hidden_states:
+                all_hidden_states += (hidden_states,)
+
+            if self.gradient_checkpointing and self.training:
+                layer_outputs = self._gradient_checkpointing_func(
+                    decoder_layer.__call__,
+                    hidden_states,
+                    causal_mask,
+                    position_ids,
+                    past_key_values,
+                    output_attentions,
+                    use_cache,
+                    cache_position,
+                )
+            else:
+                layer_outputs = decoder_layer(
+                    hidden_states,
+                    attention_mask=causal_mask,
+                    position_ids=position_ids,
+                    past_key_value=past_key_values,
+                    output_attentions=output_attentions,
+                    use_cache=use_cache,
+                    cache_position=cache_position,
+                )
+
+            hidden_states = layer_outputs[0]
+
+            if use_cache:
+                next_decoder_cache = layer_outputs[2 if output_attentions else 1]
+
+            if output_attentions:
+                all_self_attns += (layer_outputs[1],)
+
+        hidden_states = self.norm(hidden_states)
+
+        # add hidden states from the last decoder layer
+        if output_hidden_states:
+            all_hidden_states += (hidden_states,)
+
+        next_cache = None
+        if use_cache:
+            next_cache = next_decoder_cache.to_legacy_cache() if use_legacy_cache else next_decoder_cache
+
+        if not return_dict:
+            return tuple(v for v in [hidden_states, next_cache, all_hidden_states, all_self_attns] if v is not None)
+        return BaseModelOutputWithPast(
+            last_hidden_state=hidden_states,
+            past_key_values=next_cache,
+            hidden_states=all_hidden_states,
+            attentions=all_self_attns,
+        )
+
+    # Copied from transformers.models.llama.modeling_llama.LlamaModel._update_causal_mask
+    def _update_causal_mask(
+        self,
+        attention_mask: torch.Tensor,
+        input_tensor: torch.Tensor,
+        cache_position: torch.Tensor,
+        past_key_values: Cache,
+        output_attentions: bool,
+    ):
+        # TODO: As of torch==2.2.0, the `attention_mask` passed to the model in `generate` is 2D and of dynamic length even when the static
+        # KV cache is used. This is an issue for torch.compile which then recaptures cudagraphs at each decode steps due to the dynamic shapes.
+        # (`recording cudagraph tree for symint key 13`, etc.), which is VERY slow. A workaround is `@torch.compiler.disable`, but this prevents using
+        # `fullgraph=True`. See more context in https://github.com/huggingface/transformers/pull/29114
+
+        if self.config._attn_implementation == "flash_attention_2":
+            if attention_mask is not None and 0.0 in attention_mask:
+                return attention_mask
+            return None
+
+        # For SDPA, when possible, we will rely on its `is_causal` argument instead of its `attn_mask` argument, in
+        # order to dispatch on Flash Attention 2. This feature is not compatible with static cache, as SDPA will fail
+        # to infer the attention mask.
+        past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0
+        using_static_cache = False#isinstance(past_key_values, StaticCache)
+
+        # When output attentions is True, sdpa implementation's forward method calls the eager implementation's forward
+        if self.config._attn_implementation == "sdpa" and not using_static_cache and not output_attentions:
+            if AttentionMaskConverter._ignore_causal_mask_sdpa(
+                attention_mask,
+                inputs_embeds=input_tensor,
+                past_key_values_length=past_seen_tokens,
+                is_training=self.training,
+            ):
+                return None
+
+        dtype, device = input_tensor.dtype, input_tensor.device
+        min_dtype = torch.finfo(dtype).min
+        sequence_length = input_tensor.shape[1]
+        if using_static_cache:
+            target_length = past_key_values.get_max_length()
+        else:
+            target_length = (
+                attention_mask.shape[-1]
+                if isinstance(attention_mask, torch.Tensor)
+                else past_seen_tokens + sequence_length + 1
+            )
+
+        # In case the provided `attention` mask is 2D, we generate a causal mask here (4D).
+        causal_mask = _prepare_4d_causal_attention_mask_with_cache_position(
+            attention_mask,
+            sequence_length=sequence_length,
+            target_length=target_length,
+            dtype=dtype,
+            device=device,
+            min_dtype=min_dtype,
+            cache_position=cache_position,
+            batch_size=input_tensor.shape[0],
+        )
+
+        if (
+            self.config._attn_implementation == "sdpa"
+            and attention_mask is not None
+            and attention_mask.device.type == "cuda"
+            and not output_attentions
+        ):
+            # Attend to all tokens in fully masked rows in the causal_mask, for example the relevant first rows when
+            # using left padding. This is required by F.scaled_dot_product_attention memory-efficient attention path.
+            # Details: https://github.com/pytorch/pytorch/issues/110213
+            causal_mask = AttentionMaskConverter._unmask_unattended(causal_mask, min_dtype)
+
+        return causal_mask
+
+
+class Qwen2ForCausalLM(Qwen2PreTrainedModel):
+    _tied_weights_keys = ["lm_head.weight"]
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.model = Qwen2Model(config)
+        self.vocab_size = config.vocab_size
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.model.embed_tokens
+
+    def set_input_embeddings(self, value):
+        self.model.embed_tokens = value
+
+    def get_output_embeddings(self):
+        return self.lm_head
+
+    def set_output_embeddings(self, new_embeddings):
+        self.lm_head = new_embeddings
+
+    def set_decoder(self, decoder):
+        self.model = decoder
+
+    def get_decoder(self):
+        return self.model
+
+    @add_start_docstrings_to_model_forward(QWEN2_INPUTS_DOCSTRING)
+    @replace_return_docstrings(output_type=CausalLMOutputWithPast, config_class=_CONFIG_FOR_DOC)
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+        cache_position: Optional[torch.LongTensor] = None,
+    ) -> Union[Tuple, CausalLMOutputWithPast]:
+        r"""
+        Args:
+            labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+                Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
+                config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
+                (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
+
+        Returns:
+
+        Example:
+
+        ```python
+        >>> from transformers import AutoTokenizer, Qwen2ForCausalLM
+
+        >>> model = Qwen2ForCausalLM.from_pretrained(PATH_TO_CONVERTED_WEIGHTS)
+        >>> tokenizer = AutoTokenizer.from_pretrained(PATH_TO_CONVERTED_TOKENIZER)
+
+        >>> prompt = "Hey, are you conscious? Can you talk to me?"
+        >>> inputs = tokenizer(prompt, return_tensors="pt")
+
+        >>> # Generate
+        >>> generate_ids = model.generate(inputs.input_ids, max_length=30)
+        >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
+        "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
+        ```"""
+
+        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
+        outputs = self.model(
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+            cache_position=cache_position,
+        )
+
+        hidden_states = outputs[0]
+        logits = self.lm_head(hidden_states)
+        logits = logits.float()
+
+        loss = None
+        if labels is not None:
+            # Shift so that tokens < n predict n
+            shift_logits = logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss()
+            shift_logits = shift_logits.view(-1, self.config.vocab_size)
+            shift_labels = shift_labels.view(-1)
+            # Enable model parallelism
+            shift_labels = shift_labels.to(shift_logits.device)
+            loss = loss_fct(shift_logits, shift_labels)
+
+        if not return_dict:
+            output = (logits,) + outputs[1:]
+            return (loss,) + output if loss is not None else output
+
+        return CausalLMOutputWithPast(
+            loss=loss,
+            logits=logits,
+            past_key_values=outputs.past_key_values,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+    # Copied from transformers.models.llama.modeling_llama.LlamaForCausalLM.prepare_inputs_for_generation
+    def prepare_inputs_for_generation(
+        self,
+        input_ids,
+        past_key_values=None,
+        attention_mask=None,
+        inputs_embeds=None,
+        cache_position=None,
+        position_ids=None,
+        use_cache=True,
+        **kwargs,
+    ):
+        # If we have cache: let's slice `input_ids` through `cache_position`, to keep only the unprocessed tokens
+        # Exception 1: when passing input_embeds, input_ids may be missing entries
+        # Exception 2: some generation methods do special slicing of input_ids, so we don't need to do it here
+        if past_key_values is not None:
+            if inputs_embeds is not None:  # Exception 1
+                input_ids = input_ids[:, -cache_position.shape[0] :]
+            elif input_ids.shape[1] != cache_position.shape[0]:  # Default case (the "else", a no op, is Exception 2)
+                input_ids = input_ids[:, cache_position]
+
+        if attention_mask is not None and position_ids is None:
+            # create position_ids on the fly for batch generation
+            position_ids = attention_mask.long().cumsum(-1) - 1
+            position_ids.masked_fill_(attention_mask == 0, 1)
+            if past_key_values:
+                position_ids = position_ids[:, -input_ids.shape[1] :]
+
+                # This `clone` call is needed to avoid recapturing cuda graphs with `torch.compile`'s  `mode="reduce-overhead`, as otherwise the input `position_ids` would have various stride during the decoding. Here, simply using `.contiguous()` is not sufficient as in the batch size = 1 case, `position_ids` is already contiguous but with varying stride which retriggers a capture.
+                position_ids = position_ids.clone(memory_format=torch.contiguous_format)
+
+        # if `inputs_embeds` are passed, we only want to use them in the 1st generation step
+        if inputs_embeds is not None and cache_position[0] == 0:
+            model_inputs = {"inputs_embeds": inputs_embeds}
+        else:
+            model_inputs = {"input_ids": input_ids}
+
+        if False and isinstance(past_key_values, StaticCache) and attention_mask.ndim == 2:
+            if inputs_embeds is not None:
+                batch_size, sequence_length = inputs_embeds.shape
+                device = inputs_embeds.device
+            else:
+                batch_size, sequence_length = input_ids.shape
+                device = input_ids.device
+
+            dtype = self.lm_head.weight.dtype
+            min_dtype = torch.finfo(dtype).min
+
+            attention_mask = _prepare_4d_causal_attention_mask_with_cache_position(
+                attention_mask,
+                sequence_length=sequence_length,
+                target_length=past_key_values.get_max_length(),
+                dtype=dtype,
+                device=device,
+                min_dtype=min_dtype,
+                cache_position=cache_position,
+                batch_size=batch_size,
+            )
+
+        model_inputs.update(
+            {
+                "position_ids": position_ids,
+                "cache_position": cache_position,
+                "past_key_values": past_key_values,
+                "use_cache": use_cache,
+                "attention_mask": attention_mask,
+            }
+        )
+        return model_inputs
+
+
+@add_start_docstrings(
+    """
+    The Qwen2 Model transformer with a sequence classification head on top (linear layer).
+
+    [`Qwen2ForSequenceClassification`] uses the last token in order to do the classification, as other causal models
+    (e.g. GPT-2) do.
+
+    Since it does classification on the last token, it requires to know the position of the last token. If a
+    `pad_token_id` is defined in the configuration, it finds the last token that is not a padding token in each row. If
+    no `pad_token_id` is defined, it simply takes the last value in each row of the batch. Since it cannot guess the
+    padding tokens when `inputs_embeds` are passed instead of `input_ids`, it does the same (take the last value in
+    each row of the batch).
+    """,
+    QWEN2_START_DOCSTRING,
+)
+class Qwen2ForSequenceClassification(Qwen2PreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+        self.model = Qwen2Model(config)
+        self.score = nn.Linear(config.hidden_size, self.num_labels, bias=False)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.model.embed_tokens
+
+    def set_input_embeddings(self, value):
+        self.model.embed_tokens = value
+
+    @add_start_docstrings_to_model_forward(QWEN2_INPUTS_DOCSTRING)
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, SequenceClassifierOutputWithPast]:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
+            Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
+            config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
+            `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
+        """
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        transformer_outputs = self.model(
+            input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+        hidden_states = transformer_outputs[0]
+        logits = self.score(hidden_states)
+
+        if input_ids is not None:
+            batch_size = input_ids.shape[0]
+        else:
+            batch_size = inputs_embeds.shape[0]
+
+        if self.config.pad_token_id is None and batch_size != 1:
+            raise ValueError("Cannot handle batch sizes > 1 if no padding token is defined.")
+        if self.config.pad_token_id is None:
+            sequence_lengths = -1
+        else:
+            if input_ids is not None:
+                # if no pad token found, use modulo instead of reverse indexing for ONNX compatibility
+                sequence_lengths = torch.eq(input_ids, self.config.pad_token_id).int().argmax(-1) - 1
+                sequence_lengths = sequence_lengths % input_ids.shape[-1]
+                sequence_lengths = sequence_lengths.to(logits.device)
+            else:
+                sequence_lengths = -1
+
+        pooled_logits = logits[torch.arange(batch_size, device=logits.device), sequence_lengths]
+
+        loss = None
+        if labels is not None:
+            labels = labels.to(logits.device)
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+
+            if self.config.problem_type == "regression":
+                loss_fct = MSELoss()
+                if self.num_labels == 1:
+                    loss = loss_fct(pooled_logits.squeeze(), labels.squeeze())
+                else:
+                    loss = loss_fct(pooled_logits, labels)
+            elif self.config.problem_type == "single_label_classification":
+                loss_fct = CrossEntropyLoss()
+                loss = loss_fct(pooled_logits.view(-1, self.num_labels), labels.view(-1))
+            elif self.config.problem_type == "multi_label_classification":
+                loss_fct = BCEWithLogitsLoss()
+                loss = loss_fct(pooled_logits, labels)
+        if not return_dict:
+            output = (pooled_logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return SequenceClassifierOutputWithPast(
+            loss=loss,
+            logits=pooled_logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )
+
+
+@add_start_docstrings(
+    """
+    The Qwen2 Model transformer with a token classification head on top (a linear layer on top of the hidden-states
+    output) e.g. for Named-Entity-Recognition (NER) tasks.
+    """,
+    QWEN2_START_DOCSTRING,
+)
+# Copied from transformers.models.llama.modeling_llama.LlamaForTokenClassification with Llama->Qwen2, LLAMA->QWEN2
+class Qwen2ForTokenClassification(Qwen2PreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+        self.model = Qwen2Model(config)
+        if getattr(config, "classifier_dropout", None) is not None:
+            classifier_dropout = config.classifier_dropout
+        elif getattr(config, "hidden_dropout", None) is not None:
+            classifier_dropout = config.hidden_dropout
+        else:
+            classifier_dropout = 0.1
+        self.dropout = nn.Dropout(classifier_dropout)
+        self.score = nn.Linear(config.hidden_size, config.num_labels)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.model.embed_tokens
+
+    def set_input_embeddings(self, value):
+        self.model.embed_tokens = value
+
+    @add_start_docstrings_to_model_forward(QWEN2_INPUTS_DOCSTRING)
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, TokenClassifierOutput]:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
+            Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
+            config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
+            `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
+        """
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        outputs = self.model(
+            input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+        sequence_output = outputs[0]
+        sequence_output = self.dropout(sequence_output)
+        logits = self.score(sequence_output)
+
+        loss = None
+        if labels is not None:
+            loss_fct = CrossEntropyLoss()
+            loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1))
+
+        if not return_dict:
+            output = (logits,) + outputs[2:]
+            return ((loss,) + output) if loss is not None else output
+
+        return TokenClassifierOutput(
+            loss=loss,
+            logits=logits,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+
+@add_start_docstrings(
+    """
+    The Qwen2 Model transformer with a sequence classification head on top (linear layer).
+
+    [`Qwen2ForSequenceClassification`] uses the last token in order to do the classification, as other causal models
+    (e.g. GPT-2) do.
+
+    Since it does classification on the last token, it requires to know the position of the last token. If a
+    `pad_token_id` is defined in the configuration, it finds the last token that is not a padding token in each row. If
+    no `pad_token_id` is defined, it simply takes the last value in each row of the batch. Since it cannot guess the
+    padding tokens when `inputs_embeds` are passed instead of `input_ids`, it does the same (take the last value in
+    each row of the batch).
+    """,
+    QWEN2_START_DOCSTRING,
+)
+class Qwen2ForRewardModel(Qwen2PreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+        self.num_labels = 1#config.num_labels
+        self.model = Qwen2Model(config)
+        self.score = nn.Sequential(
+            nn.Linear(config.hidden_size, config.hidden_size),
+            nn.ReLU(),
+            nn.Linear(config.hidden_size, self.num_labels)
+        )
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.model.embed_tokens
+
+    def set_input_embeddings(self, value):
+        self.model.embed_tokens = value
+
+    @add_start_docstrings_to_model_forward(QWEN2_INPUTS_DOCSTRING)
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, SequenceClassifierOutputWithPast]:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
+            Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
+            config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
+            `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
+        """
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        transformer_outputs = self.model(
+            input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+        hidden_states = transformer_outputs[0]
+        logits = self.score(hidden_states)
+
+        if input_ids is not None:
+            batch_size = input_ids.shape[0]
+        else:
+            batch_size = inputs_embeds.shape[0]
+
+        if self.config.pad_token_id is None and batch_size != 1:
+            raise ValueError("Cannot handle batch sizes > 1 if no padding token is defined.")
+        if self.config.pad_token_id is None:
+            sequence_lengths = -1
+        else:
+            if input_ids is not None:
+                # if no pad token found, use modulo instead of reverse indexing for ONNX compatibility
+                sequence_lengths = torch.eq(input_ids, self.config.pad_token_id).int().argmax(-1) - 1
+                sequence_lengths = sequence_lengths % input_ids.shape[-1]
+                sequence_lengths = sequence_lengths.to(logits.device)
+            else:
+                sequence_lengths = -1
+
+        pooled_logits = logits[torch.arange(batch_size, device=logits.device), sequence_lengths]
+
+        loss = None
+        if labels is not None:
+            labels = labels.to(logits.device)
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+
+            if self.config.problem_type == "regression":
+                loss_fct = MSELoss()
+                if self.num_labels == 1:
+                    loss = loss_fct(pooled_logits.squeeze(), labels.squeeze())
+                else:
+                    loss = loss_fct(pooled_logits, labels)
+            elif self.config.problem_type == "single_label_classification":
+                loss_fct = CrossEntropyLoss()
+                loss = loss_fct(pooled_logits.view(-1, self.num_labels), labels.view(-1))
+            elif self.config.problem_type == "multi_label_classification":
+                loss_fct = BCEWithLogitsLoss()
+                loss = loss_fct(pooled_logits, labels)
+        if not return_dict:
+            output = (pooled_logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return SequenceClassifierOutputWithPast(
+            loss=loss,
+            logits=pooled_logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )

tokenizer_config.json

@@ -0,0 +1,40 @@
+{
+  "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
+    }
+  },
+  "additional_special_tokens": ["<|im_start|>", "<|im_end|>"],
+  "bos_token": null,
+  "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}{% endif %}{% if loop.last %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>'}}{% else %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endif %}{% endfor %}{{ '<|endoftext|>' }}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}