initial commit

config.json

@@ -0,0 +1,59 @@
+{
+  "architectures": [
+    "DeepseekV3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_module_list_cfg": [],
+  "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": "0.35.1"
+    },
+    "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:f86f4b3fceec6d6e6df7354e5c59a432e616eae7b0f125c8a836afbfc3d9febc
+size 4156949552

model-00002-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e8da07c613f208a177e2a0318a654c5483c634c76c696bff1220b7692973d4d0
+size 2419597576

model-00003-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:411777f83c65f0ff8b8a65865e9c4db2c0033be8aba3881fc27cc28bdb64676c
+size 2419598384

model-00004-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb17bd8c6c16854d8f280debebec0fd9b2cec5e4e72468cd90b9fe8c32f3766b
+size 2521373216

model-00005-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fcee02c1152e3eeda4402eccb6197c4637f57165af66b116556e14101e6fe03
+size 2419597944

model-00006-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad739cf199fbea8ec0422e0d9876e727deaa887d2dc2ec561245e1887556c3f9
+size 2558992920

model-00007-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b233b2d4646a70fa5b679f8e3f900cf2fc16b6e0e7c44af792a61d3559ffa4d
+size 2423268408

model-00008-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a665ed7c85cc40fbec006beabc0a8828dc743c0add02a26e9a2d9c2734f05f6c
+size 2419598352

model-00009-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78e2e69aa4c9d1f7556700a124fc7a95dcc513a7cae97538658614d32af47e2e
+size 2521373600

model-00010-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d173738485857701f910c4a0afeec5e9152cf6b3db7f5aae9fee7c7c7ad2756c
+size 2419597560

model-00011-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b9e7eebfff110727964d678ed3b00bef882c6fb1bac790a5ed472a8e57f931e
+size 2419598384

model-00012-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4d5cd0d3576a7c654354f0a4459bb4564af92340f486e86e3738d5e84d6b950
+size 743279184

model-00013-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9ce8e271d6ac3e40ab632a78adb2bbc617c03138aef4ac43b3699306ebc30c0
+size 2521313640

model-00014-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:012ad3971d27ae8aec1284b95b7812eb485d3230df77199dcc0ac0f5438dde1d
+size 2419598296

model-00015-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aefc7a8778d13ba21739b1c2527298cbb3de688564c03a59a0b1a59bbaedbbb3
+size 2521373664

model-00016-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0cf284c994aa3d9d21c10e5dfc92331a030966ab05ee539edabdcf0ac25dbf7d
+size 2419597504

model-00017-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55d4fc049bc4d1b2284ce7190eedb39133909198dc6b299827061ebe5c4fc9d5
+size 2419598384

model-00018-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91835f42bc23c65a70749e3c09d6b2784177abb2f3da53118e3e43e34a3ed595
+size 2521373280

model-00019-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f543e823fcd8d80985235e2a3007ee8b8022b4bb22bd346f4f777a13945142bb
+size 2419597880

model-00020-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad97c03ce378824d82acc3e6f8edb9b53b886a3547daed844fa3d5a8d9ca9e5d
+size 2419598392

model-00021-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0cc7c588416215030ff24b076e8ffc16324dc55d7a679501153515daebe0f91
+size 2521372352

model-00022-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:741e09286660e279dea021099a1f9a10610f7ba0edc3115f553188cf8518df8f
+size 2419599440

model-00023-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:028c065bb1b8e72a277e501ca6facf97811779304b451494fdf2e6df18a46e3e
+size 2521374824

model-00024-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0167483a59e3029585c03569ffcf06675b7a98560dce9763e293e8737f67448a
+size 2419598648

model-00025-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d3be930981cbb31968da8fa872c70d8a4f29df0bd597e683323645ec8bd9497
+size 2419599560

model-00026-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d42dc01c62be665a3ee087684683f684cc8864630e39490efb60f3fa1cbd074e
+size 2521374448

model-00027-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d14e584400d3fbd4bda36eded1af80e75c76c44b109a1cd01ba2f9def81a669
+size 2419599024

model-00028-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1c93b54d1c39116ad1ff2dc17b7e7bd6a040edbe3aa4d7acecb298fa628601a
+size 2419599560

model-00029-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6775b389f50d761bb7fd5a237f9ccca1664d7fa0e7310f5ea29cd1eeb1059391
+size 2521374072

model-00030-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0b0b2d6739e25d8b8f7c0aeb83f1397c0ae6cff6831cfc5bc0127a56cee3519
+size 2419599400

model-00031-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f9ba4d291a9f121893a6e665552dd68d773d7656f9d5a38cb110f496ddf9da2
+size 2521374864

model-00032-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb3fd7aef3652394b7f1b254e1cd53e6b144e6623944abdc9658b2e1b2c71e46
+size 2419598608

model-00033-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dcff6216ad1e6c61bdd2b17c3d247701e2bac7d1df92b5e98f51a60d4d48a086
+size 2419599568

model-00034-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd6b5878a60409d3edc8b5dfe0913eaf712e5a3c1a0c689765dd2a554801bff4
+size 982762216

model-00035-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d06c5a7f95a7ff69bae05922f877d48e156b39bb6673e0f0bd66b6a49c021139
+size 2521314768

model-00036-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4e7375d10626d0b0b154cc970349a20148d1da0a4d6c035f48151aa6af9de04
+size 2419599464

model-00037-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de98df022079723cebbef0af59e31c3561ebc765dd17192e12f09d039db1b893
+size 2521374800

model-00038-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9bdf0d207035e3503c840824fd768166f972f6b150dea115eac111dfae5c37a5
+size 2419598672

model-00039-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0582d3ce28f62408c6d27dbe519c5f471ee236e66858b1f61aeda7fe1a13061f
+size 2419599560

model-00040-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de1ed43195ae49ab9de8133fc18058798cbd1cabde2376f30bc624c868680d00
+size 2521374416

model-00041-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f73d04bcf5ec8c7cfc984682aa4b9e680d632ebcb94551b3fa7e9c171dfe37a8
+size 2419599056

model-00042-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2655bb4ac376bba01749ea9599a5763b9da18fce0cdfa7c01d07e7714eb943c1
+size 2419599568

model-00043-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d05a109bd9842476f00334b98c28bf83ddc6f00996c8fed80f50094f93048e99
+size 2521374032

model-00044-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08b96d23c3269825b61800e8bfa830a045e56036e13e242a661644349092cda0
+size 2419599440

model-00045-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:922df8a2dc58500f1cb98a3a4af4b0cc27945dd56933c99177ea1365451e94c9
+size 2521374824

model-00046-of-000163.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2462bcc5e730beac5c4c3b266210078cf56885bc5b9d5bfd421ea14785e1e113
+size 2419598648