initial commit

config.json

@@ -0,0 +1,58 @@
+{
+  "architectures": [
+    "DeepseekV3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "auto_map": {
+    "AutoConfig": "configuration_deepseek.DeepseekV3Config",
+    "AutoModel": "modeling_deepseek.DeepseekV3Model",
+    "AutoModelForCausalLM": "modeling_deepseek.DeepseekV3ForCausalLM"
+  },
+  "bos_token_id": 0,
+  "eos_token_id": 1,
+  "ep_size": 1,
+  "first_k_dense_replace": 3,
+  "hidden_act": "silu",
+  "hidden_size": 7168,
+  "initializer_range": 0.02,
+  "intermediate_size": 18432,
+  "kv_lora_rank": 512,
+  "max_position_embeddings": 163840,
+  "model_type": "deepseek_v3",
+  "moe_intermediate_size": 2048,
+  "moe_layer_freq": 1,
+  "n_group": 8,
+  "n_routed_experts": 256,
+  "n_shared_experts": 1,
+  "norm_topk_prob": true,
+  "num_attention_heads": 128,
+  "num_experts_per_tok": 8,
+  "num_hidden_layers": 61,
+  "num_key_value_heads": 128,
+  "num_nextn_predict_layers": 1,
+  "q_lora_rank": 1536,
+  "qk_nope_head_dim": 128,
+  "qk_rope_head_dim": 64,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": {
+    "beta_fast": 32,
+    "beta_slow": 1,
+    "factor": 40,
+    "mscale": 1.0,
+    "mscale_all_dim": 1.0,
+    "original_max_position_embeddings": 4096,
+    "type": "yarn"
+  },
+  "rope_theta": 10000,
+  "routed_scaling_factor": 2.5,
+  "scoring_func": "sigmoid",
+  "tie_word_embeddings": false,
+  "topk_group": 4,
+  "topk_method": "noaux_tc",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.44.2",
+  "use_cache": true,
+  "v_head_dim": 128,
+  "vocab_size": 129280
+}

configuration_deepseek.py

@@ -0,0 +1,199 @@
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+logger = logging.get_logger(__name__)
+
+DEEPSEEK_PRETRAINED_CONFIG_ARCHIVE_MAP = {}
+class DeepseekV3Config(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`DeepseekV3Model`]. It is used to instantiate an DeepSeek
+    model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
+    defaults will yield a similar configuration to that of the DeepSeek-V3.
+
+    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 129280):
+            Vocabulary size of the Deep model. Defines the number of different tokens that can be represented by the
+            `inputs_ids` passed when calling [`DeepseekV3Model`]
+        hidden_size (`int`, *optional*, defaults to 4096):
+            Dimension of the hidden representations.
+        intermediate_size (`int`, *optional*, defaults to 11008):
+            Dimension of the MLP representations.
+        moe_intermediate_size (`int`, *optional*, defaults to 1407):
+            Dimension of the MoE representations.
+        num_hidden_layers (`int`, *optional*, defaults to 32):
+            Number of hidden layers in the Transformer decoder.
+        num_nextn_predict_layers (`int`, *optional*, defaults to 1):
+            Number of nextn predict layers in the DeepSeekV3 Model.
+        num_attention_heads (`int`, *optional*, defaults to 32):
+            Number of attention heads for each attention layer in the Transformer decoder.
+        n_shared_experts (`int`, *optional*, defaults to None):
+            Number of shared experts, None means dense model.
+        n_routed_experts (`int`, *optional*, defaults to None):
+            Number of routed experts, None means dense model.
+        routed_scaling_factor (`float`, *optional*, defaults to 1.0):
+            Scaling factor or routed experts.
+        topk_method (`str`, *optional*, defaults to `gready`):
+            Topk method used in routed gate.
+        n_group (`int`, *optional*, defaults to None):
+            Number of groups for routed experts.
+        topk_group (`int`, *optional*, defaults to None):
+            Number of selected groups for each token(for each token, ensuring the selected experts is only within `topk_group` groups).
+        num_experts_per_tok (`int`, *optional*, defaults to None):
+            Number of selected experts, None means dense model.
+        moe_layer_freq (`int`, *optional*, defaults to 1):
+            The frequency of the MoE layer: one expert layer for every `moe_layer_freq - 1` dense layers.
+        first_k_dense_replace (`int`, *optional*, defaults to 0):
+            Number of dense layers in shallow layers(embed->dense->dense->...->dense->moe->moe...->lm_head).
+                                                            \--k dense layers--/
+        norm_topk_prob (`bool`, *optional*, defaults to False):
+            Whether to normalize the weights of the routed experts.
+        scoring_func (`str`, *optional*, defaults to 'softmax'):
+            Method of computing expert weights.
+        aux_loss_alpha (`float`, *optional*, defaults to 0.001):
+            Auxiliary loss weight coefficient.
+        seq_aux = (`bool`, *optional*, defaults to True):
+            Whether to compute the auxiliary loss for each individual sample.
+        num_key_value_heads (`int`, *optional*):
+            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
+            `num_attention_heads`.
+        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 2048):
+            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`.
+        pad_token_id (`int`, *optional*):
+            Padding token id.
+        bos_token_id (`int`, *optional*, defaults to 1):
+            Beginning of stream token id.
+        eos_token_id (`int`, *optional*, defaults to 2):
+            End of stream token id.
+        tie_word_embeddings (`bool`, *optional*, defaults to `False`):
+            Whether to tie weight embeddings
+        rope_theta (`float`, *optional*, defaults to 10000.0):
+            The base period of the RoPE embeddings.
+        rope_scaling (`Dict`, *optional*):
+            Dictionary containing the scaling configuration for the RoPE embeddings. Currently supports two scaling
+            strategies: linear and dynamic. Their scaling factor must be a float greater than 1. The expected format is
+            `{"type": strategy name, "factor": scaling factor}`. When using this flag, don't update
+            `max_position_embeddings` to the expected new maximum.
+        attention_bias (`bool`, defaults to `False`, *optional*, defaults to `False`):
+            Whether to use a bias in the query, key, value and output projection layers during self-attention.
+        attention_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio for the attention probabilities.
+
+    ```python
+    >>> from transformers import DeepseekV3Model, DeepseekV3Config
+
+    >>> # Initializing a Deepseek-V3 style configuration
+    >>> configuration = DeepseekV3Config()
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "deepseek_v3"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=129280,
+        hidden_size=7168,
+        intermediate_size=18432,
+        moe_intermediate_size = 2048,
+        num_hidden_layers=61,
+        num_nextn_predict_layers=1,
+        num_attention_heads=128,
+        num_key_value_heads=128,
+        n_shared_experts = 1,
+        n_routed_experts = 256,
+        ep_size = 1,
+        routed_scaling_factor = 2.5,
+        kv_lora_rank = 512,
+        q_lora_rank = 1536,
+        qk_rope_head_dim = 64,
+        v_head_dim = 128,
+        qk_nope_head_dim = 128,
+        topk_method = 'noaux_tc',
+        n_group = 8,
+        topk_group = 4,
+        num_experts_per_tok = 8,
+        moe_layer_freq = 1,
+        first_k_dense_replace = 3,
+        norm_topk_prob = True,
+        scoring_func = 'sigmoid',
+        hidden_act="silu",
+        max_position_embeddings=4096,
+        initializer_range=0.02,
+        rms_norm_eps=1e-6,
+        use_cache=True,
+        pad_token_id=None,
+        bos_token_id=0,
+        eos_token_id=1,
+        tie_word_embeddings=False,
+        rope_theta=10000.0,
+        rope_scaling=None,
+        attention_bias=False,
+        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.moe_intermediate_size = moe_intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_nextn_predict_layers = num_nextn_predict_layers
+        self.num_attention_heads = num_attention_heads
+        self.n_shared_experts = n_shared_experts
+        self.n_routed_experts = n_routed_experts
+        self.ep_size = ep_size
+        self.routed_scaling_factor = routed_scaling_factor
+        self.kv_lora_rank = kv_lora_rank
+        self.q_lora_rank = q_lora_rank
+        self.qk_rope_head_dim = qk_rope_head_dim
+        self.v_head_dim = v_head_dim
+        self.qk_nope_head_dim = qk_nope_head_dim
+        self.topk_method = topk_method
+        self.n_group = n_group
+        self.topk_group = topk_group
+        self.num_experts_per_tok = num_experts_per_tok
+        self.moe_layer_freq = moe_layer_freq
+        self.first_k_dense_replace = first_k_dense_replace
+        self.norm_topk_prob = norm_topk_prob
+        self.scoring_func = scoring_func
+        # 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.rope_scaling = rope_scaling
+        self.attention_bias = attention_bias
+        self.attention_dropout = attention_dropout
+
+        super().__init__(
+            pad_token_id=pad_token_id,
+            bos_token_id=bos_token_id,
+            eos_token_id=eos_token_id,
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )

generation_config.json

@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 0,
+  "eos_token_id": 1,
+  "do_sample": true,
+  "temperature": 0.6,
+  "top_p": 0.95,
+  "transformers_version": "4.46.3"
+}

hf_quant_config.json

@@ -0,0 +1,259 @@
+{
+    "producer": {
+        "name": "modelopt",
+        "version": "0.22.1.dev750+g59a703703"
+    },
+    "quantization": {
+        "quant_algo": "NVFP4",
+        "kv_cache_quant_algo": "FP8",
+        "group_size": 16,
+        "exclude_modules": [
+            "lm_head",
+            "model.layers.0.self_attn.kv_a_proj_with_mqa",
+            "model.layers.0.self_attn.kv_b_proj",
+            "model.layers.0.self_attn.q_a_proj",
+            "model.layers.0.self_attn.q_b_proj",
+            "model.layers.1.self_attn.kv_a_proj_with_mqa",
+            "model.layers.1.self_attn.kv_b_proj",
+            "model.layers.1.self_attn.q_a_proj",
+            "model.layers.1.self_attn.q_b_proj",
+            "model.layers.10.self_attn.kv_a_proj_with_mqa",
+            "model.layers.10.self_attn.kv_b_proj",
+            "model.layers.10.self_attn.q_a_proj",
+            "model.layers.10.self_attn.q_b_proj",
+            "model.layers.11.self_attn.kv_a_proj_with_mqa",
+            "model.layers.11.self_attn.kv_b_proj",
+            "model.layers.11.self_attn.q_a_proj",
+            "model.layers.11.self_attn.q_b_proj",
+            "model.layers.12.self_attn.kv_a_proj_with_mqa",
+            "model.layers.12.self_attn.kv_b_proj",
+            "model.layers.12.self_attn.q_a_proj",
+            "model.layers.12.self_attn.q_b_proj",
+            "model.layers.13.self_attn.kv_a_proj_with_mqa",
+            "model.layers.13.self_attn.kv_b_proj",
+            "model.layers.13.self_attn.q_a_proj",
+            "model.layers.13.self_attn.q_b_proj",
+            "model.layers.14.self_attn.kv_a_proj_with_mqa",
+            "model.layers.14.self_attn.kv_b_proj",
+            "model.layers.14.self_attn.q_a_proj",
+            "model.layers.14.self_attn.q_b_proj",
+            "model.layers.15.self_attn.kv_a_proj_with_mqa",
+            "model.layers.15.self_attn.kv_b_proj",
+            "model.layers.15.self_attn.q_a_proj",
+            "model.layers.15.self_attn.q_b_proj",
+            "model.layers.16.self_attn.kv_a_proj_with_mqa",
+            "model.layers.16.self_attn.kv_b_proj",
+            "model.layers.16.self_attn.q_a_proj",
+            "model.layers.16.self_attn.q_b_proj",
+            "model.layers.17.self_attn.kv_a_proj_with_mqa",
+            "model.layers.17.self_attn.kv_b_proj",
+            "model.layers.17.self_attn.q_a_proj",
+            "model.layers.17.self_attn.q_b_proj",
+            "model.layers.18.self_attn.kv_a_proj_with_mqa",
+            "model.layers.18.self_attn.kv_b_proj",
+            "model.layers.18.self_attn.q_a_proj",
+            "model.layers.18.self_attn.q_b_proj",
+            "model.layers.19.self_attn.kv_a_proj_with_mqa",
+            "model.layers.19.self_attn.kv_b_proj",
+            "model.layers.19.self_attn.q_a_proj",
+            "model.layers.19.self_attn.q_b_proj",
+            "model.layers.2.self_attn.kv_a_proj_with_mqa",
+            "model.layers.2.self_attn.kv_b_proj",
+            "model.layers.2.self_attn.q_a_proj",
+            "model.layers.2.self_attn.q_b_proj",
+            "model.layers.20.self_attn.kv_a_proj_with_mqa",
+            "model.layers.20.self_attn.kv_b_proj",
+            "model.layers.20.self_attn.q_a_proj",
+            "model.layers.20.self_attn.q_b_proj",
+            "model.layers.21.self_attn.kv_a_proj_with_mqa",
+            "model.layers.21.self_attn.kv_b_proj",
+            "model.layers.21.self_attn.q_a_proj",
+            "model.layers.21.self_attn.q_b_proj",
+            "model.layers.22.self_attn.kv_a_proj_with_mqa",
+            "model.layers.22.self_attn.kv_b_proj",
+            "model.layers.22.self_attn.q_a_proj",
+            "model.layers.22.self_attn.q_b_proj",
+            "model.layers.23.self_attn.kv_a_proj_with_mqa",
+            "model.layers.23.self_attn.kv_b_proj",
+            "model.layers.23.self_attn.q_a_proj",
+            "model.layers.23.self_attn.q_b_proj",
+            "model.layers.24.self_attn.kv_a_proj_with_mqa",
+            "model.layers.24.self_attn.kv_b_proj",
+            "model.layers.24.self_attn.q_a_proj",
+            "model.layers.24.self_attn.q_b_proj",
+            "model.layers.25.self_attn.kv_a_proj_with_mqa",
+            "model.layers.25.self_attn.kv_b_proj",
+            "model.layers.25.self_attn.q_a_proj",
+            "model.layers.25.self_attn.q_b_proj",
+            "model.layers.26.self_attn.kv_a_proj_with_mqa",
+            "model.layers.26.self_attn.kv_b_proj",
+            "model.layers.26.self_attn.q_a_proj",
+            "model.layers.26.self_attn.q_b_proj",
+            "model.layers.27.self_attn.kv_a_proj_with_mqa",
+            "model.layers.27.self_attn.kv_b_proj",
+            "model.layers.27.self_attn.q_a_proj",
+            "model.layers.27.self_attn.q_b_proj",
+            "model.layers.28.self_attn.kv_a_proj_with_mqa",
+            "model.layers.28.self_attn.kv_b_proj",
+            "model.layers.28.self_attn.q_a_proj",
+            "model.layers.28.self_attn.q_b_proj",
+            "model.layers.29.self_attn.kv_a_proj_with_mqa",
+            "model.layers.29.self_attn.kv_b_proj",
+            "model.layers.29.self_attn.q_a_proj",
+            "model.layers.29.self_attn.q_b_proj",
+            "model.layers.3.self_attn.kv_a_proj_with_mqa",
+            "model.layers.3.self_attn.kv_b_proj",
+            "model.layers.3.self_attn.q_a_proj",
+            "model.layers.3.self_attn.q_b_proj",
+            "model.layers.30.self_attn.kv_a_proj_with_mqa",
+            "model.layers.30.self_attn.kv_b_proj",
+            "model.layers.30.self_attn.q_a_proj",
+            "model.layers.30.self_attn.q_b_proj",
+            "model.layers.31.self_attn.kv_a_proj_with_mqa",
+            "model.layers.31.self_attn.kv_b_proj",
+            "model.layers.31.self_attn.q_a_proj",
+            "model.layers.31.self_attn.q_b_proj",
+            "model.layers.32.self_attn.kv_a_proj_with_mqa",
+            "model.layers.32.self_attn.kv_b_proj",
+            "model.layers.32.self_attn.q_a_proj",
+            "model.layers.32.self_attn.q_b_proj",
+            "model.layers.33.self_attn.kv_a_proj_with_mqa",
+            "model.layers.33.self_attn.kv_b_proj",
+            "model.layers.33.self_attn.q_a_proj",
+            "model.layers.33.self_attn.q_b_proj",
+            "model.layers.34.self_attn.kv_a_proj_with_mqa",
+            "model.layers.34.self_attn.kv_b_proj",
+            "model.layers.34.self_attn.q_a_proj",
+            "model.layers.34.self_attn.q_b_proj",
+            "model.layers.35.self_attn.kv_a_proj_with_mqa",
+            "model.layers.35.self_attn.kv_b_proj",
+            "model.layers.35.self_attn.q_a_proj",
+            "model.layers.35.self_attn.q_b_proj",
+            "model.layers.36.self_attn.kv_a_proj_with_mqa",
+            "model.layers.36.self_attn.kv_b_proj",
+            "model.layers.36.self_attn.q_a_proj",
+            "model.layers.36.self_attn.q_b_proj",
+            "model.layers.37.self_attn.kv_a_proj_with_mqa",
+            "model.layers.37.self_attn.kv_b_proj",
+            "model.layers.37.self_attn.q_a_proj",
+            "model.layers.37.self_attn.q_b_proj",
+            "model.layers.38.self_attn.kv_a_proj_with_mqa",
+            "model.layers.38.self_attn.kv_b_proj",
+            "model.layers.38.self_attn.q_a_proj",
+            "model.layers.38.self_attn.q_b_proj",
+            "model.layers.39.self_attn.kv_a_proj_with_mqa",
+            "model.layers.39.self_attn.kv_b_proj",
+            "model.layers.39.self_attn.q_a_proj",
+            "model.layers.39.self_attn.q_b_proj",
+            "model.layers.4.self_attn.kv_a_proj_with_mqa",
+            "model.layers.4.self_attn.kv_b_proj",
+            "model.layers.4.self_attn.q_a_proj",
+            "model.layers.4.self_attn.q_b_proj",
+            "model.layers.40.self_attn.kv_a_proj_with_mqa",
+            "model.layers.40.self_attn.kv_b_proj",
+            "model.layers.40.self_attn.q_a_proj",
+            "model.layers.40.self_attn.q_b_proj",
+            "model.layers.41.self_attn.kv_a_proj_with_mqa",
+            "model.layers.41.self_attn.kv_b_proj",
+            "model.layers.41.self_attn.q_a_proj",
+            "model.layers.41.self_attn.q_b_proj",
+            "model.layers.42.self_attn.kv_a_proj_with_mqa",
+            "model.layers.42.self_attn.kv_b_proj",
+            "model.layers.42.self_attn.q_a_proj",
+            "model.layers.42.self_attn.q_b_proj",
+            "model.layers.43.self_attn.kv_a_proj_with_mqa",
+            "model.layers.43.self_attn.kv_b_proj",
+            "model.layers.43.self_attn.q_a_proj",
+            "model.layers.43.self_attn.q_b_proj",
+            "model.layers.44.self_attn.kv_a_proj_with_mqa",
+            "model.layers.44.self_attn.kv_b_proj",
+            "model.layers.44.self_attn.q_a_proj",
+            "model.layers.44.self_attn.q_b_proj",
+            "model.layers.45.self_attn.kv_a_proj_with_mqa",
+            "model.layers.45.self_attn.kv_b_proj",
+            "model.layers.45.self_attn.q_a_proj",
+            "model.layers.45.self_attn.q_b_proj",
+            "model.layers.46.self_attn.kv_a_proj_with_mqa",
+            "model.layers.46.self_attn.kv_b_proj",
+            "model.layers.46.self_attn.q_a_proj",
+            "model.layers.46.self_attn.q_b_proj",
+            "model.layers.47.self_attn.kv_a_proj_with_mqa",
+            "model.layers.47.self_attn.kv_b_proj",
+            "model.layers.47.self_attn.q_a_proj",
+            "model.layers.47.self_attn.q_b_proj",
+            "model.layers.48.self_attn.kv_a_proj_with_mqa",
+            "model.layers.48.self_attn.kv_b_proj",
+            "model.layers.48.self_attn.q_a_proj",
+            "model.layers.48.self_attn.q_b_proj",
+            "model.layers.49.self_attn.kv_a_proj_with_mqa",
+            "model.layers.49.self_attn.kv_b_proj",
+            "model.layers.49.self_attn.q_a_proj",
+            "model.layers.49.self_attn.q_b_proj",
+            "model.layers.5.self_attn.kv_a_proj_with_mqa",
+            "model.layers.5.self_attn.kv_b_proj",
+            "model.layers.5.self_attn.q_a_proj",
+            "model.layers.5.self_attn.q_b_proj",
+            "model.layers.50.self_attn.kv_a_proj_with_mqa",
+            "model.layers.50.self_attn.kv_b_proj",
+            "model.layers.50.self_attn.q_a_proj",
+            "model.layers.50.self_attn.q_b_proj",
+            "model.layers.51.self_attn.kv_a_proj_with_mqa",
+            "model.layers.51.self_attn.kv_b_proj",
+            "model.layers.51.self_attn.q_a_proj",
+            "model.layers.51.self_attn.q_b_proj",
+            "model.layers.52.self_attn.kv_a_proj_with_mqa",
+            "model.layers.52.self_attn.kv_b_proj",
+            "model.layers.52.self_attn.q_a_proj",
+            "model.layers.52.self_attn.q_b_proj",
+            "model.layers.53.self_attn.kv_a_proj_with_mqa",
+            "model.layers.53.self_attn.kv_b_proj",
+            "model.layers.53.self_attn.q_a_proj",
+            "model.layers.53.self_attn.q_b_proj",
+            "model.layers.54.self_attn.kv_a_proj_with_mqa",
+            "model.layers.54.self_attn.kv_b_proj",
+            "model.layers.54.self_attn.q_a_proj",
+            "model.layers.54.self_attn.q_b_proj",
+            "model.layers.55.self_attn.kv_a_proj_with_mqa",
+            "model.layers.55.self_attn.kv_b_proj",
+            "model.layers.55.self_attn.q_a_proj",
+            "model.layers.55.self_attn.q_b_proj",
+            "model.layers.56.self_attn.kv_a_proj_with_mqa",
+            "model.layers.56.self_attn.kv_b_proj",
+            "model.layers.56.self_attn.q_a_proj",
+            "model.layers.56.self_attn.q_b_proj",
+            "model.layers.57.self_attn.kv_a_proj_with_mqa",
+            "model.layers.57.self_attn.kv_b_proj",
+            "model.layers.57.self_attn.q_a_proj",
+            "model.layers.57.self_attn.q_b_proj",
+            "model.layers.58.self_attn.kv_a_proj_with_mqa",
+            "model.layers.58.self_attn.kv_b_proj",
+            "model.layers.58.self_attn.q_a_proj",
+            "model.layers.58.self_attn.q_b_proj",
+            "model.layers.59.self_attn.kv_a_proj_with_mqa",
+            "model.layers.59.self_attn.kv_b_proj",
+            "model.layers.59.self_attn.q_a_proj",
+            "model.layers.59.self_attn.q_b_proj",
+            "model.layers.6.self_attn.kv_a_proj_with_mqa",
+            "model.layers.6.self_attn.kv_b_proj",
+            "model.layers.6.self_attn.q_a_proj",
+            "model.layers.6.self_attn.q_b_proj",
+            "model.layers.60.self_attn.kv_a_proj_with_mqa",
+            "model.layers.60.self_attn.kv_b_proj",
+            "model.layers.60.self_attn.q_a_proj",
+            "model.layers.60.self_attn.q_b_proj",
+            "model.layers.7.self_attn.kv_a_proj_with_mqa",
+            "model.layers.7.self_attn.kv_b_proj",
+            "model.layers.7.self_attn.q_a_proj",
+            "model.layers.7.self_attn.q_b_proj",
+            "model.layers.8.self_attn.kv_a_proj_with_mqa",
+            "model.layers.8.self_attn.kv_b_proj",
+            "model.layers.8.self_attn.q_a_proj",
+            "model.layers.8.self_attn.q_b_proj",
+            "model.layers.9.self_attn.kv_a_proj_with_mqa",
+            "model.layers.9.self_attn.kv_b_proj",
+            "model.layers.9.self_attn.q_a_proj",
+            "model.layers.9.self_attn.q_b_proj",
+            "model.layers.61*"
+        ]
+    }
+}

model-00001-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79afbda70fc76dfa8e62f6305529ab211d8e22b4da3cc5d3652b2741a28af07b
+size 4156845944

model-00002-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ce8085cff032e9905071c208474f0ad7021682bc0c860ce47a09f91c580585c
+size 2419597576

model-00003-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e92e0df7d80cda07586e0844d3124e40d4582d03104a16a1c3ea28d722b819fc
+size 2419598384

model-00004-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b9bd5357eff82ca15998d10a00c01ca835b8a0391c0b42998cdc58bb6578e48
+size 2521338792

model-00005-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:30788b22a4740aff70e41d8cf2bf9b4d78a8997f59febe355543873fca349430
+size 2419597944

model-00006-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6aab4d239d206d1ac1e412433dcda50c8b561ff12a77d1240d76ecf45c8ed429
+size 2558958464

model-00007-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:350a47546c1d0d39ba343b3df6360c476d7612696ed19939d496e9ad777d8ccb
+size 2423268408

model-00008-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97ca03f88ad84ddaa508a2796c6438ee5155a7e052a49d05850ae5aa350e753d
+size 2419598352

model-00009-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3c8ec18a19d75f5ee6c27db405ceeb2eb05a4a84fb5f92519dc833e9d87ba1c
+size 2521339176

model-00010-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4a070318dd61ff5e16021b75adb8653b476142187a8fa17e16c5b2854228d33
+size 2419597560

model-00011-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:44ac5e59b853b4a2a9d146b784dbd25fcfc8a30f4b78b29e45684314bda73585
+size 2419598384

model-00012-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0973f937413daf29176b9b52cb8b42b1274ccdc1b4d5b8894df3cd1fb711e168
+size 743249736

model-00013-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:defdb3703c7c50400774a01ed346e69f9a904f044d0efafa16ec680b3fac5aab
+size 2521309560

model-00014-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d128224429aede8c6b75a0b559048415d0f8856661588c295983b16a249158f8
+size 2419598296

model-00015-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18b847c10c28d3b0a71f2ddcd5530b5529829aa6d08a2e47cb587ef52cd75dee
+size 2521339232

model-00016-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc7ecc22a799fab2e518d3d7d6aec2899bf968a2de2e4787f080597cb073db3a
+size 2419597504

model-00017-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87590f2135ac8c93c0ed597f029aace6ce1e3cd62da120b4c97163162a1c8510
+size 2419598384

model-00018-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1f83af4fe6fccd7f2dec6e4608b3e97961bfd3b41981e77dcce00de2092149b
+size 2521338856

model-00019-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35f50844545ec1c8ef8c8a585b095c6a6a4f6087a6728f8aa578f0b3d82388be
+size 2419597880

model-00020-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d312e5cffb02edc2760295554d0be02c9732b95df11d11c435c2703c4c30e5d2
+size 2419598392

model-00021-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da8ad816b27adf76b329bdcc61cd73a4dc997f3c84c055f6b0abc92ddd2f1a4e
+size 2521339584

model-00022-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f7802d3b07ccfa804e9799cbc72240aaa2e8965cfbd5cfbe72f067056edbea2
+size 2419599440

model-00023-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e30802b8b244a28b4d8cd32175905e285cbc9f6ed5ad7e8462b8469fa787211c
+size 2521340392

model-00024-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f0651f23995adb80ec0215f5591a3763f617bec5ca686a7cbe837b86bddeaf1
+size 2419598648

model-00025-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fbf67e075f9872a2e6143a042cea7fdba12f3643cbcec34831d3fbc567a1b670
+size 2419599560

model-00026-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8156b6ff9b69d325c6f688483abd7dc4203171851821b945c65ed2ad62b858b8
+size 2521340016

model-00027-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1a1a137a83a6f52ef3bbc991cf3757a92f7c9b97faa21606884063d29afc3bd
+size 2419599024

model-00028-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d974765b482728b81c1428f90b1bbbfc5f308007c30fd2521b7ed86eafe74c8
+size 2419599560

model-00029-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a57a91a66e30fa6be8de9fb083a7951260c848e6cc7438e7d47b4e4645deceb9
+size 2521339640

model-00030-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5c694f21306957194732d107c7a009b3f3f135ea9a1a654f91077e2392ceda4
+size 2419599400

model-00031-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d85f244267e7499432f9fed8868d200d8b58ab11849f1d03281674c92a25b71
+size 2521340432

model-00032-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5844ba91a0672bd1840b9779e1d6436e9d36131c9b0d22d42874816246a55669
+size 2419598608

model-00033-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:918130fff9e9c47dd32f67e7c280c62dc934891cc66b4e45c28562ad800b78f5
+size 2419599568

model-00034-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d4d519c3ce6c3ffa7e6efa81cec10b8ef4abcd3ea64cd1e005ddbf33179299b
+size 982732544

model-00035-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:040e7375b0f88d890f02f84c4c57090d3170084b7edb44045cfcd24376185f39
+size 2521310688

model-00036-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f50447d6958c3d01e9efbb8414898c36e83515723778388ee0786f929b63f1b
+size 2419599464

model-00037-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:276b2e20ef350eaecb34620ce7cbbcafe09654fe4ec53cb9b38267dabae98197
+size 2521340368

model-00038-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b08f2f135cbfd8df122fc80f35979ba3f903056c1fc6b1b7b62b226155e2059
+size 2419598672

model-00039-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a44a45b85a0c7c5a76744bd8448c2b70ba97ea4d0671d302d6d6f5eda466dcf
+size 2419599560

model-00040-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c5c451e54d83a65dab43a972c0ef44bf2777a8c57b3b70f8d1c03edab38b7c9
+size 2521339984

model-00041-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ab3c5dbfe8766ac5b33ede9d2e142fd061def1431c511c33089a65a5e2570df
+size 2419599056

model-00042-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b00071ed8f884dc54d970c4aa3f0a3b165310339a76569c252cb26ca3cd7025
+size 2419599568

model-00043-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86fb1add3de7678b499dcb25ffe3a7cffc77529c322000bbd89fc17dc8203e28
+size 2521339600

model-00044-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec2c7905c746a20715a1222d33fc1e7fbd136641a5b04f2c37b6b8d774a6d997
+size 2419599440