Add files using upload-large-folder tool

config.json

@@ -0,0 +1,85 @@
+{
+  "activation_type": "silu",
+  "add_faster_video": false,
+  "add_time_instruction": false,
+  "alibi": false,
+  "alibi_bias_max": 8.0,
+  "architectures": [
+    "LlavaLladaForMaskedDiffusion"
+  ],
+  "attention_dropout": 0.0,
+  "attention_layer_norm": false,
+  "attention_layer_norm_with_affine": true,
+  "auto_map": {
+    "AutoConfig": "configuration_llada.LLaDAConfig",
+    "AutoModel": "modeling_llada.LLaDAModelLM",
+    "AutoModelForCausalLM": "modeling_llada.LLaDAModelLM"
+  },
+  "bias_for_layer_norm": false,
+  "block_group_size": 1,
+  "block_type": "llama",
+  "d_model": 4096,
+  "embedding_dropout": 0.0,
+  "embedding_size": 126464,
+  "eos_token_id": 126081,
+  "faster_token_stride": 10,
+  "flash_attention": false,
+  "force_sample": false,
+  "image_aspect_ratio": "square",
+  "image_crop_resolution": null,
+  "image_grid_pinpoints": null,
+  "image_split_resolution": null,
+  "include_bias": false,
+  "include_qkv_bias": false,
+  "init_cutoff_factor": null,
+  "init_device": "meta",
+  "init_fn": "mitchell",
+  "init_std": 0.02,
+  "input_emb_norm": false,
+  "layer_norm_type": "rms",
+  "layer_norm_with_affine": true,
+  "mask_token_id": 126336,
+  "max_sequence_length": 4096,
+  "mlp_hidden_size": 12288,
+  "mlp_ratio": 4,
+  "mm_hidden_size": 1152,
+  "mm_newline_position": "grid",
+  "mm_patch_merge_type": "spatial_unpad",
+  "mm_pooler_ratio": 2,
+  "mm_projector_lr": null,
+  "mm_projector_type": "mlp2x_gelu",
+  "mm_spatial_pool_mode": "bilinear",
+  "mm_spatial_pool_stride": null,
+  "mm_tunable_parts": "mm_vision_tower,mm_mlp_adapter,mm_language_model",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "mm_vision_select_feature": "patch",
+  "mm_vision_select_layer": -2,
+  "mm_vision_tower": "/data/siglip-so400m-patch14-384",
+  "mm_vision_tower_lr": 2e-06,
+  "model_type": "llada",
+  "multi_query_attention": null,
+  "n_heads": 32,
+  "n_kv_heads": 32,
+  "n_layers": 32,
+  "pad_token_id": 126081,
+  "pos_skipping_range": 4096,
+  "precision": "amp_bf16",
+  "resampler_type": null,
+  "residual_dropout": 0.0,
+  "rms_norm_eps": 1e-05,
+  "rope": true,
+  "rope_full_precision": true,
+  "rope_theta": 500000.0,
+  "scale_logits": false,
+  "tokenizer_model_max_length": 2048,
+  "tokenizer_padding_side": "right",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.50.3",
+  "use_cache": false,
+  "use_mm_proj": true,
+  "use_pos_skipping": false,
+  "vision_tower_pretrained": null,
+  "vocab_size": 126464,
+  "weight_tying": false
+}

configuration_llada.py

@@ -0,0 +1,463 @@
+"""
+LLaDA configuration
+"""
+from transformers import AutoConfig, PretrainedConfig
+
+from enum import Enum
+from os import PathLike
+from typing import Union
+from dataclasses import asdict, dataclass, field
+from glob import glob
+from pathlib import Path
+from typing import (
+    Any,
+    Dict,
+    Iterable,
+    List,
+    Optional,
+    Tuple,
+    Type,
+    TypeVar,
+    Union,
+    cast,
+)
+
+
+__all__ = [
+    "ActivationType",
+    "ActivationCheckpointingStrategy",
+    "BlockType",
+    "LayerNormType",
+    "InitFnType",
+    "ModelConfig",
+]
+
+PathOrStr = Union[str, PathLike]
+
+
+class StrEnum(str, Enum):
+    """
+    This is equivalent to Python's :class:`enum.StrEnum` since version 3.11.
+    We include this here for compatibility with older version of Python.
+    """
+
+    def __str__(self) -> str:
+        return self.value
+
+    def __repr__(self) -> str:
+        return f"'{str(self)}'"
+
+
+class LayerNormType(StrEnum):
+    default = "default"
+    """
+    The default LayerNorm implementation, equivalent to PyTorch's built-in version.
+    """
+
+    low_precision = "low_precision"
+    """
+    A low-precision version of the default LayerNorm.
+    """
+
+    rms = "rms"
+    """
+    An RMSNorm implementation. When using ``torch.compile`` this is
+    probably the fastest implementation.
+    """
+
+    gemma_rms = "gemma_rms"
+    """
+    An RMSNorm implementation by gemmma. When using ``torch.compile`` this is
+    probably the fastest implementation.
+    """
+
+    amd_compatible = "amd_compatible"
+    """
+    LayerNorm implemented manually to work around an issue with ROCm.
+    """
+
+
+class ActivationType(StrEnum):
+    gelu = "gelu"
+    relu = "relu"
+    silu = "silu"
+    swiglu = "swiglu"
+
+
+class BlockType(StrEnum):
+    sequential = "sequential"
+    parallel = "parallel"
+
+    llama = "llama"
+    """
+    A block similar to the sequential block with slightly different
+    implementations of operations like attention to imitate the behavior of Llama.
+    """
+
+
+class InitFnType(StrEnum):
+    mitchell = "mitchell"
+    """
+    The strategy suggested to us by Mitchell Wortsman from UW.
+    This uses a truncated normal distribution with an adaptive standard deviation that depends
+    on the size of the weights as well as the depth of the layer.
+    """
+
+    normal = "normal"
+    """
+    All weights are initialized from the same normal distribution.
+    """
+
+    kaiming_normal = "kaiming_normal"
+    """
+    All weights are initialized with the Kaiming method from a normal distribution.
+    Note this currently won't work with FSDP.
+    """
+
+    fan_in = "fan_in"
+    """
+    "Fan-in variance scaling", i.e. normal with a standard deviation of ``1/sqrt(d_in)`` where ``d_in``
+    is the input dimensionality of the kernel.
+    """
+
+    full_megatron = "full_megatron"
+    """
+    This is what metaseq calls "full megatron init". It is the init used for Llama 2.
+    """
+
+
+@dataclass
+class ModelConfig():
+    """
+    LLaDA (model) configuration.
+    """
+
+    # Note that the defaults for these attributes are equivalent to the base GPT2 model.
+
+    d_model: int = 768
+    """
+    The hidden size of the model.
+    """
+
+    n_heads: int = 12
+    """
+    The number of self-attention heads.
+    """
+
+    n_kv_heads: Optional[int] = None
+    """
+    The number of heads to use for keys and values. Defaults to `n_heads`.
+    Set this to ``None`` or ``n_heads`` for normal multi-head attention.
+    Set this to 1 for multi-query attention.
+    Set it to some in-between value for Llama2-style grouped query attention.
+    """
+
+    n_layers: int = 12
+    """
+    The number of layers/blocks.
+    """
+
+    mlp_ratio: int = 4
+    """
+    The ratio of the inner MLP dimensionality to ``d_model``.
+    This is only used when ``mlp_hidden_size`` is not set.
+    """
+
+    mlp_hidden_size: Optional[int] = None
+    """
+    Set the exact hidden size for the MLP. Otherwise the inner MLP hidden size will be set to `mlp_ratio * d_model`.
+    """
+
+    activation_type: ActivationType = ActivationType.swiglu
+    """
+    The activation function to use within the MLP layers.
+    """
+
+    block_type: BlockType = BlockType.sequential
+    """
+    The transformer block implementation.
+    """
+
+    block_group_size: int = 1
+    """
+    The number of blocks to group together into a single parent block.
+    This has no affect on the number of parameters in the model and is only used to wrap groups
+    of blocks together with a single FSDP wrapper during training.
+    """
+
+    alibi: bool = False
+    """
+    If ``True``, use ALiBi embeddings. Mutually exclusive with ``rope``.
+    """
+
+    alibi_bias_max: float = 8.0
+    """
+    Maximum absolute value of ALiBi bias.
+    """
+
+    rope: bool = False
+    """
+    Use rotary positional embeddings (RoPE). Mutually exclusive with ``alibi``.
+    """
+
+    rope_full_precision: bool = True
+    """
+    If ``True``, apply RoPE embeddings at full precision regardless of the input type. Otherwise,
+    apply RoPE at the precision of the input.
+    """
+
+    flash_attention: bool = False
+    """
+    If ``True``, use ``FlashAttention``.
+    """
+
+    attention_dropout: float = 0.1
+    """
+    The dropout probability within the attention modules.
+    """
+
+    multi_query_attention: Optional[bool] = None
+    """
+    Use the Multi-Query formulation of attention used in PaLM. This reduces the number of parameters
+    and is more efficient during inference.
+    """
+
+    attention_layer_norm: bool = False
+    """
+    Apply layer norm to the keys and queries within the attention mechanism.
+    This can help stabilize training.
+    """
+
+    residual_dropout: float = 0.1
+    """
+    The dropout probability for the MLP and attention output within each block.
+    """
+
+    embedding_dropout: float = 0.1
+    """
+    The dropout probability for embeddings.
+    """
+
+    input_emb_norm: bool = False
+    """
+    An input hidden_states norm implementation by gemmma.
+    """
+
+    layer_norm_type: LayerNormType = LayerNormType.default
+    """
+    The layernorm implementation to use.
+    """
+
+    layer_norm_with_affine: bool = True
+    """
+    Whether to include bias and weight parameters for the layer norms.
+    This only affects layer norms that are immediately followed by a linear layer in the forward pass,
+    so everything except QK-norms. To turn off affines for QK norms as well, set :attr:`attention_layer_norm_with_affine`
+    to ``False``.
+    """
+
+    rms_norm_eps: float = 1e-05
+    """
+    The rms layernorm eps param.
+    """
+
+    attention_layer_norm_with_affine: bool = True
+    """
+    Toggle affine transform for the QK norms.
+    """
+
+    max_sequence_length: int = 1024
+    """
+    The maximum input sequence length supported by the model.
+    """
+
+    rope_theta: float = 10000.0
+    """
+    The rope base param.
+    """
+
+    include_qkv_bias: Optional[bool] = False
+    """
+    Whether or not to include bias parameters in qkv linear layers.
+    """
+
+    include_bias: bool = False
+    """
+    Whether or not to include bias parameters in linear layers.
+    In PaLM, they got rid of all bias terms because they found that large
+    models tend to have near 0 bias terms anyway.
+    """
+
+    bias_for_layer_norm: Optional[bool] = None
+    """
+    Whether or not to include bias parameters in layer norm.
+    This is separate from the include_bias parameter, because of a ROCm crash when biases are disabled in
+    layer norm.
+    When this is None (the default), it inherits the setting from include_bias.
+    """
+
+    scale_logits: bool = False
+    """
+    If ``True``, scale the output logits by ``1 / sqrt(d_model)``.
+    """
+
+    vocab_size: int = 50257
+    """
+    Vocabulary size of the model.
+    """
+
+    embedding_size: Optional[int] = 50304
+    """
+    The number of embeddings, i.e. the number of tokens. If set to ``None`` it will default
+    to ``vocab_size``. If ``vocab_size`` is not a multiple of 128, setting this to the
+    next multiple of 128 that's greater than ``vocab_size`` can improve throughput
+    substantially.
+    """
+
+    weight_tying: bool = True
+    """
+    Whether to tie output linear weights to the input embedding.
+    """
+
+    eos_token_id: int = 50256
+    """
+    The ID of the end-of-sentence special token.
+    """
+
+    pad_token_id: int = 50256
+    """
+    The ID of the token to use for padding. Defaults to the ID of the EOS token.
+    """
+
+    mask_token_id: Optional[int] = 50256
+    """
+    The ID of the token to use for mask token. Defaults to the ID of the EOS token.
+    """
+
+    init_device: Optional[str] = None
+    """
+    The torch device to use when initializing the model parameters, e.g. "cpu", "cuda:0", "meta".
+    """
+
+    init_fn: InitFnType = InitFnType.normal
+    """
+    The weight initialization strategy.
+    """
+
+    init_std: float = 0.02
+    """
+    The standard deviation to use when initializing weights with a "fixed distribution" ``init_fn``, such
+    as "normal".
+    """
+
+    init_cutoff_factor: Optional[float] = None
+    """
+    A positive factor used to scale the cutoff values when initializing weights with a "fixed distribution" ``init_fn``, such
+    as "normal". Setting this to None means values are not cutoff.
+    """
+
+    precision: Optional[str] = None
+    """
+    Precision used to train/evaluate with. You shouldn't set this directly.
+    See :data:`TrainConfig.precision` instead.
+    """
+
+    @property
+    def effective_n_kv_heads(self) -> int:
+        if self.n_kv_heads is None:
+            if self.multi_query_attention is True:
+                return 1
+            else:
+                return self.n_heads
+        else:
+            if self.multi_query_attention is None:
+                return self.n_kv_heads
+            if self.multi_query_attention:
+                n_kv_heads_should_be = 1
+            else:
+                n_kv_heads_should_be = self.n_heads
+            if self.n_kv_heads == n_kv_heads_should_be:
+                return n_kv_heads_should_be
+            else:
+                raise Exception(
+                    "You can't set `multi_query_attention` and `n_kv_heads` at the same time."
+                )
+
+class ActivationCheckpointingStrategy(StrEnum):
+    whole_layer = "whole_layer"
+    """
+    Checkpoint every transformer layer.
+    """
+
+    one_in_two = "one_in_two"
+    """
+    Checkpoint one in two transformer layers.
+    """
+
+    one_in_three = "one_in_three"
+    """
+    Checkpoint one in three transformer layers.
+    """
+
+    one_in_four = "one_in_four"
+    """
+    Checkpoint one in four transformer layers.
+    """
+    
+    two_in_three = "two_in_three"
+    """
+    Checkpoint two out of every three transformer layers.
+    """
+
+    three_in_four = "three_in_four"
+    """
+    Checkpoint three out of four of every transformer layers.
+    """
+
+    four_in_five = "four_in_five"
+    """
+    Checkpoint four out of five of every transformer layers.
+    """
+
+    nine_in_ten = "nine_in_ten"
+    """
+    Checkpoint nine out of ten of every transformer layers.
+    """
+
+    fine_grained = "fine_grained"
+    """
+    Focus checkpointing on where it is cheap to recompute and saves most memory.
+    """
+
+
+class LLaDAConfig(PretrainedConfig):
+    model_type = "llada"
+    keys_to_ignore_at_inference = ["past_key_values"]  # TODO: confirm
+
+    def __init__(self, use_cache: bool = False, **kwargs):
+        model_config = ModelConfig()
+        all_kwargs = model_config.__dict__
+        all_kwargs.update(kwargs)
+        all_kwargs.update({"use_cache": use_cache})
+        all_kwargs.update(
+            {
+                "architectures": all_kwargs.get("architectures", ["LLaDAModelLM"])
+            }
+        )
+        super().__init__(**all_kwargs)
+
+    @property
+    def num_attention_heads(self):
+        return self.n_heads
+
+    @property
+    def num_hidden_layers(self):
+        return self.n_layers
+
+    @property
+    def hidden_size(self):
+        return self.d_model
+
+
+# Register the config class so that it is available for transformer pipelines, auto-loading etc.
+AutoConfig.register("llada", LLaDAConfig)

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 126080,
+  "eos_token_id": 126081,
+  "transformers_version": "4.50.3"
+}

latest

@@ -0,0 +1 @@
+global_step321

model-00001-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c4f50fa160f55a9193d227afc2c3b9a2e903e2a87325c84754db3bb18e1a1b3
+size 4995589944

model-00002-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c3663d9900d79a708b0f9ce3554e86186aae472de271031f45c7f95e3dc8701
+size 4999819552

model-00003-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cabea9e9815b1c574328cd84ee375cbc77e61dc9239da4f78dc5e15be77bd5b5
+size 4999802728

model-00004-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f5e7f9011f0620d1fea7b106334d6210f54f957985dad2680a5a76aebe50428
+size 1874563264

model.safetensors.index.json

@@ -0,0 +1,724 @@
+{
+  "metadata": {
+    "total_size": 16869674048
+  },
+  "weight_map": {
+    "model.image_newline": "model-00001-of-00004.safetensors",
+    "model.mm_projector.0.bias": "model-00004-of-00004.safetensors",
+    "model.mm_projector.0.weight": "model-00004-of-00004.safetensors",
+    "model.mm_projector.2.bias": "model-00004-of-00004.safetensors",
+    "model.mm_projector.2.weight": "model-00004-of-00004.safetensors",
+    "model.transformer.blocks.0.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.10.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.2.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.20.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.20.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.20.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.20.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.20.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.20.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.20.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.20.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.20.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.21.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.3.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.30.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.4.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.9.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.9.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.ff_out.weight": "model-00004-of-00004.safetensors",
+    "model.transformer.ln_f.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.wte.weight": "model-00001-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.embeddings.patch_embedding.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.embeddings.patch_embedding.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.embeddings.position_embedding.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.24.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.25.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.layer_norm1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.layer_norm1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.layer_norm2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.layer_norm2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.mlp.fc1.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.mlp.fc1.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.mlp.fc2.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.mlp.fc2.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.k_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.out_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.out_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.q_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.v_proj.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.post_layernorm.bias": "model-00004-of-00004.safetensors",
+    "model.vision_tower.vision_tower.vision_model.post_layernorm.weight": "model-00004-of-00004.safetensors"
+  }
+}

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2b64c779616b4447b27dc70d231d96bd46384376b4aff094da11d9056000068
+size 1064

special_tokens_map.json

@@ -0,0 +1,38 @@
+{
+  "additional_special_tokens": [
+    "<role>",
+    "</role>",
+    "<|arithmetic_start|>",
+    "<|arithmetic_end|>",
+    "<|number_start|>",
+    "<|number_end|>"
+  ],
+  "bos_token": {
+    "content": "<|startoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "cls_token": {
+    "content": "[CLS]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer_config.json

@@ -0,0 +1,2184 @@
+{
+  "add_bos_token": false,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "126080": {
+      "content": "<|startoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126081": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126082": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126083": {
+      "content": "[gMASK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126084": {
+      "content": "<|reserved_token_0|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126085": {
+      "content": "<|reserved_token_1|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126086": {
+      "content": "<|reserved_token_2|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126087": {
+      "content": "<|reserved_token_3|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126088": {
+      "content": "<|reserved_token_4|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126089": {
+      "content": "<|reserved_token_5|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126090": {
+      "content": "<|reserved_token_6|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126091": {
+      "content": "<|reserved_token_7|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126092": {
+      "content": "<|reserved_token_8|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126093": {
+      "content": "<|reserved_token_9|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126094": {
+      "content": "<|reserved_token_10|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126095": {
+      "content": "<|reserved_token_11|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126096": {
+      "content": "<|reserved_token_12|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126097": {
+      "content": "<|reserved_token_13|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126098": {
+      "content": "<|reserved_token_14|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126099": {
+      "content": "<|reserved_token_15|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126100": {
+      "content": "<|reserved_token_16|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126101": {
+      "content": "<|reserved_token_17|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126102": {
+      "content": "<|reserved_token_18|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126103": {
+      "content": "<|reserved_token_19|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126104": {
+      "content": "<|reserved_token_20|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126105": {
+      "content": "<|reserved_token_21|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126106": {
+      "content": "<|reserved_token_22|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126107": {
+      "content": "<|reserved_token_23|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126108": {
+      "content": "<|reserved_token_24|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126109": {
+      "content": "<|reserved_token_25|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126110": {
+      "content": "<|reserved_token_26|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126111": {
+      "content": "<|reserved_token_27|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126112": {
+      "content": "<|reserved_token_28|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126113": {
+      "content": "<|reserved_token_29|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126114": {
+      "content": "<|reserved_token_30|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126115": {
+      "content": "<|reserved_token_31|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126116": {
+      "content": "<|reserved_token_32|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126117": {
+      "content": "<|reserved_token_33|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126118": {
+      "content": "<|reserved_token_34|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126119": {
+      "content": "<|reserved_token_35|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126120": {
+      "content": "<|reserved_token_36|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126121": {
+      "content": "<|reserved_token_37|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126122": {
+      "content": "<|reserved_token_38|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126123": {
+      "content": "<|reserved_token_39|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126124": {
+      "content": "<|reserved_token_40|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126125": {
+      "content": "<|reserved_token_41|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126126": {
+      "content": "<|reserved_token_42|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126127": {
+      "content": "<|reserved_token_43|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126128": {
+      "content": "<|reserved_token_44|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126129": {
+      "content": "<|reserved_token_45|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126130": {
+      "content": "<|reserved_token_46|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126131": {
+      "content": "<|reserved_token_47|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126132": {
+      "content": "<|reserved_token_48|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126133": {
+      "content": "<|reserved_token_49|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126134": {
+      "content": "<|reserved_token_50|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126135": {
+      "content": "<|reserved_token_51|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126136": {
+      "content": "<|reserved_token_52|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126137": {
+      "content": "<|reserved_token_53|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126138": {
+      "content": "<|reserved_token_54|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126139": {
+      "content": "<|reserved_token_55|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126140": {
+      "content": "<|reserved_token_56|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126141": {
+      "content": "<|reserved_token_57|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126142": {
+      "content": "<|reserved_token_58|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126143": {
+      "content": "<|reserved_token_59|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126144": {
+      "content": "<|reserved_token_60|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126145": {
+      "content": "<|reserved_token_61|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126146": {
+      "content": "<|reserved_token_62|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126147": {
+      "content": "<|reserved_token_63|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126148": {
+      "content": "<|reserved_token_64|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126149": {
+      "content": "<|reserved_token_65|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126150": {
+      "content": "<|reserved_token_66|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126151": {
+      "content": "<|reserved_token_67|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126152": {
+      "content": "<|reserved_token_68|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126153": {
+      "content": "<|reserved_token_69|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126154": {
+      "content": "<|reserved_token_70|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126155": {
+      "content": "<|reserved_token_71|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126156": {
+      "content": "<|reserved_token_72|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126157": {
+      "content": "<|reserved_token_73|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126158": {
+      "content": "<|reserved_token_74|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126159": {
+      "content": "<|reserved_token_75|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126160": {
+      "content": "<|reserved_token_76|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126161": {
+      "content": "<|reserved_token_77|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126162": {
+      "content": "<|reserved_token_78|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126163": {
+      "content": "<|reserved_token_79|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126164": {
+      "content": "<|reserved_token_80|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126165": {
+      "content": "<|reserved_token_81|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126166": {
+      "content": "<|reserved_token_82|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126167": {
+      "content": "<|reserved_token_83|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126168": {
+      "content": "<|reserved_token_84|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126169": {
+      "content": "<|reserved_token_85|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126170": {
+      "content": "<|reserved_token_86|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126171": {
+      "content": "<|reserved_token_87|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126172": {
+      "content": "<|reserved_token_88|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126173": {
+      "content": "<|reserved_token_89|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126174": {
+      "content": "<|reserved_token_90|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126175": {
+      "content": "<|reserved_token_91|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126176": {
+      "content": "<|reserved_token_92|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126177": {
+      "content": "<|reserved_token_93|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126178": {
+      "content": "<|reserved_token_94|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126179": {
+      "content": "<|reserved_token_95|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126180": {
+      "content": "<|reserved_token_96|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126181": {
+      "content": "<|reserved_token_97|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126182": {
+      "content": "<|reserved_token_98|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126183": {
+      "content": "<|reserved_token_99|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126184": {
+      "content": "<|reserved_token_100|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126185": {
+      "content": "<|reserved_token_101|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126186": {
+      "content": "<|reserved_token_102|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126187": {
+      "content": "<|reserved_token_103|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126188": {
+      "content": "<|reserved_token_104|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126189": {
+      "content": "<|reserved_token_105|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126190": {
+      "content": "<|reserved_token_106|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126191": {
+      "content": "<|reserved_token_107|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126192": {
+      "content": "<|reserved_token_108|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126193": {
+      "content": "<|reserved_token_109|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126194": {
+      "content": "<|reserved_token_110|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126195": {
+      "content": "<|reserved_token_111|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126196": {
+      "content": "<|reserved_token_112|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126197": {
+      "content": "<|reserved_token_113|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126198": {
+      "content": "<|reserved_token_114|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126199": {
+      "content": "<|reserved_token_115|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126200": {
+      "content": "<|reserved_token_116|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126201": {
+      "content": "<|reserved_token_117|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126202": {
+      "content": "<|reserved_token_118|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126203": {
+      "content": "<|reserved_token_119|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126204": {
+      "content": "<|reserved_token_120|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126205": {
+      "content": "<|reserved_token_121|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126206": {
+      "content": "<|reserved_token_122|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126207": {
+      "content": "<|reserved_token_123|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126208": {
+      "content": "<|reserved_token_124|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126209": {
+      "content": "<|reserved_token_125|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126210": {
+      "content": "<|reserved_token_126|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126211": {
+      "content": "<|reserved_token_127|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126212": {
+      "content": "<|reserved_token_128|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126213": {
+      "content": "<|reserved_token_129|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126214": {
+      "content": "<|reserved_token_130|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126215": {
+      "content": "<|reserved_token_131|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126216": {
+      "content": "<|reserved_token_132|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126217": {
+      "content": "<|reserved_token_133|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126218": {
+      "content": "<|reserved_token_134|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126219": {
+      "content": "<|reserved_token_135|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126220": {
+      "content": "<|reserved_token_136|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126221": {
+      "content": "<|reserved_token_137|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126222": {
+      "content": "<|reserved_token_138|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126223": {
+      "content": "<|reserved_token_139|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126224": {
+      "content": "<|reserved_token_140|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126225": {
+      "content": "<|reserved_token_141|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126226": {
+      "content": "<|reserved_token_142|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126227": {
+      "content": "<|reserved_token_143|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126228": {
+      "content": "<|reserved_token_144|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126229": {
+      "content": "<|reserved_token_145|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126230": {
+      "content": "<|reserved_token_146|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126231": {
+      "content": "<|reserved_token_147|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126232": {
+      "content": "<|reserved_token_148|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126233": {
+      "content": "<|reserved_token_149|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126234": {
+      "content": "<|reserved_token_150|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126235": {
+      "content": "<|reserved_token_151|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126236": {
+      "content": "<|reserved_token_152|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126237": {
+      "content": "<|reserved_token_153|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126238": {
+      "content": "<|reserved_token_154|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126239": {
+      "content": "<|reserved_token_155|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126240": {
+      "content": "<|reserved_token_156|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126241": {
+      "content": "<|reserved_token_157|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126242": {
+      "content": "<|reserved_token_158|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126243": {
+      "content": "<|reserved_token_159|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126244": {
+      "content": "<|reserved_token_160|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126245": {
+      "content": "<|reserved_token_161|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126246": {
+      "content": "<|reserved_token_162|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126247": {
+      "content": "<|reserved_token_163|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126248": {
+      "content": "<|reserved_token_164|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126249": {
+      "content": "<|reserved_token_165|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126250": {
+      "content": "<|reserved_token_166|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126251": {
+      "content": "<|reserved_token_167|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126252": {
+      "content": "<|reserved_token_168|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126253": {
+      "content": "<|reserved_token_169|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126254": {
+      "content": "<|reserved_token_170|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126255": {
+      "content": "<|reserved_token_171|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126256": {
+      "content": "<|reserved_token_172|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126257": {
+      "content": "<|reserved_token_173|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126258": {
+      "content": "<|reserved_token_174|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126259": {
+      "content": "<|reserved_token_175|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126260": {
+      "content": "<|reserved_token_176|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126261": {
+      "content": "<|reserved_token_177|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126262": {
+      "content": "<|reserved_token_178|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126263": {
+      "content": "<|reserved_token_179|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126264": {
+      "content": "<|reserved_token_180|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126265": {
+      "content": "<|reserved_token_181|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126266": {
+      "content": "<|reserved_token_182|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126267": {
+      "content": "<|reserved_token_183|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126268": {
+      "content": "<|reserved_token_184|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126269": {
+      "content": "<|reserved_token_185|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126270": {
+      "content": "<|reserved_token_186|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126271": {
+      "content": "<|reserved_token_187|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126272": {
+      "content": "<|reserved_token_188|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126273": {
+      "content": "<|reserved_token_189|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126274": {
+      "content": "<|reserved_token_190|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126275": {
+      "content": "<|reserved_token_191|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126276": {
+      "content": "<|reserved_token_192|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126277": {
+      "content": "<|reserved_token_193|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126278": {
+      "content": "<|reserved_token_194|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126279": {
+      "content": "<|reserved_token_195|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126280": {
+      "content": "<|reserved_token_196|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126281": {
+      "content": "<|reserved_token_197|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126282": {
+      "content": "<|reserved_token_198|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126283": {
+      "content": "<|reserved_token_199|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126284": {
+      "content": "<|reserved_token_200|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126285": {
+      "content": "<|reserved_token_201|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126286": {
+      "content": "<|reserved_token_202|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126287": {
+      "content": "<|reserved_token_203|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126288": {
+      "content": "<|reserved_token_204|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126289": {
+      "content": "<|reserved_token_205|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126290": {
+      "content": "<|reserved_token_206|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126291": {
+      "content": "<|reserved_token_207|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126292": {
+      "content": "<|reserved_token_208|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126293": {
+      "content": "<|reserved_token_209|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126294": {
+      "content": "<|reserved_token_210|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126295": {
+      "content": "<|reserved_token_211|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126296": {
+      "content": "<|reserved_token_212|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126297": {
+      "content": "<|reserved_token_213|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126298": {
+      "content": "<|reserved_token_214|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126299": {
+      "content": "<|reserved_token_215|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126300": {
+      "content": "<|reserved_token_216|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126301": {
+      "content": "<|reserved_token_217|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126302": {
+      "content": "<|reserved_token_218|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126303": {
+      "content": "<|reserved_token_219|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126304": {
+      "content": "<|reserved_token_220|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126305": {
+      "content": "<|reserved_token_221|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126306": {
+      "content": "<|reserved_token_222|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126307": {
+      "content": "<|reserved_token_223|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126308": {
+      "content": "<|reserved_token_224|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126309": {
+      "content": "<|reserved_token_225|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126310": {
+      "content": "<|reserved_token_226|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126311": {
+      "content": "<|reserved_token_227|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126312": {
+      "content": "<|reserved_token_228|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126313": {
+      "content": "<|reserved_token_229|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126314": {
+      "content": "<|reserved_token_230|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126315": {
+      "content": "<|reserved_token_231|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126316": {
+      "content": "<|reserved_token_232|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126317": {
+      "content": "<|reserved_token_233|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126318": {
+      "content": "<|reserved_token_234|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126319": {
+      "content": "<|reserved_token_235|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126320": {
+      "content": "<|reserved_token_236|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126321": {
+      "content": "<|reserved_token_237|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126322": {
+      "content": "<|reserved_token_238|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126323": {
+      "content": "<|reserved_token_239|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126324": {
+      "content": "<|reserved_token_240|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126325": {
+      "content": "<|reserved_token_241|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126326": {
+      "content": "<|reserved_token_242|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126327": {
+      "content": "<|reserved_token_243|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126328": {
+      "content": "<|reserved_token_244|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126329": {
+      "content": "<|reserved_token_245|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126330": {
+      "content": "<|reserved_token_246|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126331": {
+      "content": "<|reserved_token_247|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126332": {
+      "content": "<|reserved_token_248|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126333": {
+      "content": "<|reserved_token_249|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126334": {
+      "content": "<|reserved_token_250|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126335": {
+      "content": "<|reserved_token_251|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126336": {
+      "content": "<|mdm_mask|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126337": {
+      "content": "<|reserved_token_253|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126338": {
+      "content": "<|reserved_token_254|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126339": {
+      "content": "<|reserved_token_255|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126340": {
+      "content": "<role>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126341": {
+      "content": "</role>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126342": {
+      "content": "<|arithmetic_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126343": {
+      "content": "<|arithmetic_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126344": {
+      "content": "<|number_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126345": {
+      "content": "<|number_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126346": {
+      "content": "<|start_header_id|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126347": {
+      "content": "<|end_header_id|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126348": {
+      "content": "<|eot_id|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<role>",
+    "</role>",
+    "<|arithmetic_start|>",
+    "<|arithmetic_end|>",
+    "<|number_start|>",
+    "<|number_end|>"
+  ],
+  "bos_token": "<|startoftext|>",
+  "chat_template": "{% set loop_messages = messages %}{% for message in loop_messages %}{% set content = '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim + '<|eot_id|>' %}{% if loop.index0 == 0 %}{% set content = bos_token + content %}{% endif %}{{ content }}{% endfor %}{{ '<|start_header_id|>assistant<|end_header_id|>\n\n' }}",
+  "clean_up_tokenization_spaces": false,
+  "cls_token": "[CLS]",
+  "eos_token": "<|endoftext|>",
+  "extra_special_tokens": {},
+  "fast_tokenizer": true,
+  "gmask_token": "[gMASK]",
+  "merges_file": null,
+  "model_input_names": [
+    "input_ids",
+    "attention_mask"
+  ],
+  "model_max_length": 2048,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "tokenizer_class": "PreTrainedTokenizer",
+  "trust_remote_code": true
+}

trainer_state.json

@@ -0,0 +1,2309 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 325,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01556420233463035,
+      "grad_norm": 32.460037697908916,
+      "learning_rate": 1.25e-06,
+      "loss": 2.0427,
+      "step": 1
+    },
+    {
+      "epoch": 0.0311284046692607,
+      "grad_norm": 14.314919163607806,
+      "learning_rate": 2.5e-06,
+      "loss": 1.8704,
+      "step": 2
+    },
+    {
+      "epoch": 0.04669260700389105,
+      "grad_norm": 27.137073787955334,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 2.009,
+      "step": 3
+    },
+    {
+      "epoch": 0.0622568093385214,
+      "grad_norm": 22.855686347208184,
+      "learning_rate": 5e-06,
+      "loss": 2.2908,
+      "step": 4
+    },
+    {
+      "epoch": 0.07782101167315175,
+      "grad_norm": 17.021052812960836,
+      "learning_rate": 6.25e-06,
+      "loss": 1.6062,
+      "step": 5
+    },
+    {
+      "epoch": 0.0933852140077821,
+      "grad_norm": 8.04115371265731,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.6328,
+      "step": 6
+    },
+    {
+      "epoch": 0.10894941634241245,
+      "grad_norm": 8.684988862656194,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 1.5791,
+      "step": 7
+    },
+    {
+      "epoch": 0.1245136186770428,
+      "grad_norm": 10.489704115967008,
+      "learning_rate": 1e-05,
+      "loss": 1.437,
+      "step": 8
+    },
+    {
+      "epoch": 0.14007782101167315,
+      "grad_norm": 9.345852547867393,
+      "learning_rate": 1.125e-05,
+      "loss": 1.4409,
+      "step": 9
+    },
+    {
+      "epoch": 0.1556420233463035,
+      "grad_norm": 12.566418385513145,
+      "learning_rate": 1.25e-05,
+      "loss": 1.9836,
+      "step": 10
+    },
+    {
+      "epoch": 0.17120622568093385,
+      "grad_norm": 9.458631541156663,
+      "learning_rate": 1.375e-05,
+      "loss": 1.7625,
+      "step": 11
+    },
+    {
+      "epoch": 0.1867704280155642,
+      "grad_norm": 4.775804904067164,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.3358,
+      "step": 12
+    },
+    {
+      "epoch": 0.20233463035019456,
+      "grad_norm": 7.063116954030999,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 1.5127,
+      "step": 13
+    },
+    {
+      "epoch": 0.2178988326848249,
+      "grad_norm": 20.081862895733657,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 1.6345,
+      "step": 14
+    },
+    {
+      "epoch": 0.23346303501945526,
+      "grad_norm": 7.4857791672008664,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 1.6543,
+      "step": 15
+    },
+    {
+      "epoch": 0.2490272373540856,
+      "grad_norm": 5.930935178930693,
+      "learning_rate": 2e-05,
+      "loss": 1.3915,
+      "step": 16
+    },
+    {
+      "epoch": 0.26459143968871596,
+      "grad_norm": 4.891425231471704,
+      "learning_rate": 1.9999819470801393e-05,
+      "loss": 1.4524,
+      "step": 17
+    },
+    {
+      "epoch": 0.2801556420233463,
+      "grad_norm": 8.526785651246044,
+      "learning_rate": 1.999927789044796e-05,
+      "loss": 1.3936,
+      "step": 18
+    },
+    {
+      "epoch": 0.29571984435797666,
+      "grad_norm": 8.794855084727713,
+      "learning_rate": 1.9998375280666606e-05,
+      "loss": 1.4946,
+      "step": 19
+    },
+    {
+      "epoch": 0.311284046692607,
+      "grad_norm": 5.5605631475331,
+      "learning_rate": 1.9997111677667875e-05,
+      "loss": 1.5168,
+      "step": 20
+    },
+    {
+      "epoch": 0.32684824902723736,
+      "grad_norm": 6.109413863257978,
+      "learning_rate": 1.999548713214448e-05,
+      "loss": 1.3959,
+      "step": 21
+    },
+    {
+      "epoch": 0.3424124513618677,
+      "grad_norm": 8.216498243755654,
+      "learning_rate": 1.9993501709269297e-05,
+      "loss": 1.5457,
+      "step": 22
+    },
+    {
+      "epoch": 0.35797665369649806,
+      "grad_norm": 4.575471414036173,
+      "learning_rate": 1.9991155488692714e-05,
+      "loss": 1.3057,
+      "step": 23
+    },
+    {
+      "epoch": 0.3735408560311284,
+      "grad_norm": 7.713781786096019,
+      "learning_rate": 1.9988448564539475e-05,
+      "loss": 1.52,
+      "step": 24
+    },
+    {
+      "epoch": 0.38910505836575876,
+      "grad_norm": 9.333764622766337,
+      "learning_rate": 1.998538104540488e-05,
+      "loss": 1.2957,
+      "step": 25
+    },
+    {
+      "epoch": 0.4046692607003891,
+      "grad_norm": 6.153987069403218,
+      "learning_rate": 1.9981953054350436e-05,
+      "loss": 1.2496,
+      "step": 26
+    },
+    {
+      "epoch": 0.42023346303501946,
+      "grad_norm": 7.204269939964425,
+      "learning_rate": 1.997816472889891e-05,
+      "loss": 1.2556,
+      "step": 27
+    },
+    {
+      "epoch": 0.4357976653696498,
+      "grad_norm": 9.356556657699878,
+      "learning_rate": 1.9974016221028825e-05,
+      "loss": 1.3693,
+      "step": 28
+    },
+    {
+      "epoch": 0.45136186770428016,
+      "grad_norm": 18.674995074165327,
+      "learning_rate": 1.9969507697168372e-05,
+      "loss": 1.3901,
+      "step": 29
+    },
+    {
+      "epoch": 0.4669260700389105,
+      "grad_norm": 8.743718086448856,
+      "learning_rate": 1.996463933818869e-05,
+      "loss": 1.4637,
+      "step": 30
+    },
+    {
+      "epoch": 0.48249027237354086,
+      "grad_norm": 6.036142649788235,
+      "learning_rate": 1.9959411339396667e-05,
+      "loss": 1.446,
+      "step": 31
+    },
+    {
+      "epoch": 0.4980544747081712,
+      "grad_norm": 11.118098752442311,
+      "learning_rate": 1.9953823910527057e-05,
+      "loss": 1.2325,
+      "step": 32
+    },
+    {
+      "epoch": 0.5136186770428015,
+      "grad_norm": 8.532563003893063,
+      "learning_rate": 1.9947877275734103e-05,
+      "loss": 1.4182,
+      "step": 33
+    },
+    {
+      "epoch": 0.5291828793774319,
+      "grad_norm": 8.44093940432976,
+      "learning_rate": 1.9941571673582517e-05,
+      "loss": 1.519,
+      "step": 34
+    },
+    {
+      "epoch": 0.5447470817120622,
+      "grad_norm": 10.678108587266333,
+      "learning_rate": 1.9934907357037913e-05,
+      "loss": 1.4811,
+      "step": 35
+    },
+    {
+      "epoch": 0.5603112840466926,
+      "grad_norm": 12.279223122144039,
+      "learning_rate": 1.992788459345669e-05,
+      "loss": 1.5383,
+      "step": 36
+    },
+    {
+      "epoch": 0.5758754863813229,
+      "grad_norm": 9.6225514739576,
+      "learning_rate": 1.9920503664575252e-05,
+      "loss": 1.634,
+      "step": 37
+    },
+    {
+      "epoch": 0.5914396887159533,
+      "grad_norm": 6.816063682489534,
+      "learning_rate": 1.991276486649876e-05,
+      "loss": 1.4922,
+      "step": 38
+    },
+    {
+      "epoch": 0.6070038910505836,
+      "grad_norm": 8.462345745450083,
+      "learning_rate": 1.990466850968921e-05,
+      "loss": 1.4553,
+      "step": 39
+    },
+    {
+      "epoch": 0.622568093385214,
+      "grad_norm": 7.511393821964174,
+      "learning_rate": 1.9896214918953003e-05,
+      "loss": 1.613,
+      "step": 40
+    },
+    {
+      "epoch": 0.6381322957198443,
+      "grad_norm": 12.76100865490119,
+      "learning_rate": 1.9887404433427917e-05,
+      "loss": 1.3829,
+      "step": 41
+    },
+    {
+      "epoch": 0.6536964980544747,
+      "grad_norm": 9.953708209183798,
+      "learning_rate": 1.9878237406569476e-05,
+      "loss": 1.2018,
+      "step": 42
+    },
+    {
+      "epoch": 0.669260700389105,
+      "grad_norm": 17.646408078968005,
+      "learning_rate": 1.9868714206136787e-05,
+      "loss": 1.2234,
+      "step": 43
+    },
+    {
+      "epoch": 0.6848249027237354,
+      "grad_norm": 6.687546724266615,
+      "learning_rate": 1.985883521417781e-05,
+      "loss": 1.3619,
+      "step": 44
+    },
+    {
+      "epoch": 0.7003891050583657,
+      "grad_norm": 7.265965326318142,
+      "learning_rate": 1.9848600827013976e-05,
+      "loss": 1.3766,
+      "step": 45
+    },
+    {
+      "epoch": 0.7159533073929961,
+      "grad_norm": 10.47550352191459,
+      "learning_rate": 1.983801145522434e-05,
+      "loss": 1.4978,
+      "step": 46
+    },
+    {
+      "epoch": 0.7315175097276264,
+      "grad_norm": 7.719719591865551,
+      "learning_rate": 1.9827067523629075e-05,
+      "loss": 1.5026,
+      "step": 47
+    },
+    {
+      "epoch": 0.7470817120622568,
+      "grad_norm": 6.653131519406193,
+      "learning_rate": 1.981576947127245e-05,
+      "loss": 1.3853,
+      "step": 48
+    },
+    {
+      "epoch": 0.7626459143968871,
+      "grad_norm": 8.90485792562694,
+      "learning_rate": 1.9804117751405213e-05,
+      "loss": 1.4688,
+      "step": 49
+    },
+    {
+      "epoch": 0.7782101167315175,
+      "grad_norm": 7.409558877721469,
+      "learning_rate": 1.9792112831466385e-05,
+      "loss": 1.2318,
+      "step": 50
+    },
+    {
+      "epoch": 0.7937743190661478,
+      "grad_norm": 10.548282670671158,
+      "learning_rate": 1.9779755193064545e-05,
+      "loss": 1.1853,
+      "step": 51
+    },
+    {
+      "epoch": 0.8093385214007782,
+      "grad_norm": 8.195819265190133,
+      "learning_rate": 1.9767045331958486e-05,
+      "loss": 1.3867,
+      "step": 52
+    },
+    {
+      "epoch": 0.8249027237354085,
+      "grad_norm": 7.195118313650143,
+      "learning_rate": 1.9753983758037324e-05,
+      "loss": 1.3728,
+      "step": 53
+    },
+    {
+      "epoch": 0.8404669260700389,
+      "grad_norm": 6.865553942117972,
+      "learning_rate": 1.9740570995300054e-05,
+      "loss": 1.4971,
+      "step": 54
+    },
+    {
+      "epoch": 0.8560311284046692,
+      "grad_norm": 13.711210401882385,
+      "learning_rate": 1.9726807581834522e-05,
+      "loss": 1.5261,
+      "step": 55
+    },
+    {
+      "epoch": 0.8715953307392996,
+      "grad_norm": 5.9709078674205776,
+      "learning_rate": 1.971269406979584e-05,
+      "loss": 1.1547,
+      "step": 56
+    },
+    {
+      "epoch": 0.8871595330739299,
+      "grad_norm": 7.305358527853627,
+      "learning_rate": 1.9698231025384234e-05,
+      "loss": 1.2522,
+      "step": 57
+    },
+    {
+      "epoch": 0.9027237354085603,
+      "grad_norm": 7.144056189221613,
+      "learning_rate": 1.9683419028822333e-05,
+      "loss": 1.2921,
+      "step": 58
+    },
+    {
+      "epoch": 0.9182879377431906,
+      "grad_norm": 6.9606269877032005,
+      "learning_rate": 1.9668258674331882e-05,
+      "loss": 1.2708,
+      "step": 59
+    },
+    {
+      "epoch": 0.933852140077821,
+      "grad_norm": 9.722878510396836,
+      "learning_rate": 1.9652750570109914e-05,
+      "loss": 1.4427,
+      "step": 60
+    },
+    {
+      "epoch": 0.9494163424124513,
+      "grad_norm": 9.963909818863343,
+      "learning_rate": 1.9636895338304347e-05,
+      "loss": 1.3381,
+      "step": 61
+    },
+    {
+      "epoch": 0.9649805447470817,
+      "grad_norm": 6.446954560778193,
+      "learning_rate": 1.9620693614989024e-05,
+      "loss": 1.4269,
+      "step": 62
+    },
+    {
+      "epoch": 0.980544747081712,
+      "grad_norm": 7.603247343231994,
+      "learning_rate": 1.9604146050138194e-05,
+      "loss": 1.4298,
+      "step": 63
+    },
+    {
+      "epoch": 0.9961089494163424,
+      "grad_norm": 6.349264800380979,
+      "learning_rate": 1.958725330760044e-05,
+      "loss": 1.0356,
+      "step": 64
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.349264800380979,
+      "learning_rate": 1.9570016065072047e-05,
+      "loss": 0.3835,
+      "step": 65
+    },
+    {
+      "epoch": 1.0155642023346303,
+      "grad_norm": 8.467306321422342,
+      "learning_rate": 1.9552435014069805e-05,
+      "loss": 1.329,
+      "step": 66
+    },
+    {
+      "epoch": 1.0311284046692606,
+      "grad_norm": 8.408957527442329,
+      "learning_rate": 1.953451085990329e-05,
+      "loss": 1.2351,
+      "step": 67
+    },
+    {
+      "epoch": 1.046692607003891,
+      "grad_norm": 13.406884070709882,
+      "learning_rate": 1.9516244321646533e-05,
+      "loss": 1.6201,
+      "step": 68
+    },
+    {
+      "epoch": 1.0622568093385214,
+      "grad_norm": 6.37594555257866,
+      "learning_rate": 1.9497636132109208e-05,
+      "loss": 1.3477,
+      "step": 69
+    },
+    {
+      "epoch": 1.0778210116731517,
+      "grad_norm": 6.084035911845763,
+      "learning_rate": 1.9478687037807215e-05,
+      "loss": 1.1285,
+      "step": 70
+    },
+    {
+      "epoch": 1.0933852140077822,
+      "grad_norm": 5.938233548564638,
+      "learning_rate": 1.9459397798932732e-05,
+      "loss": 1.2728,
+      "step": 71
+    },
+    {
+      "epoch": 1.1089494163424125,
+      "grad_norm": 4.999727686110884,
+      "learning_rate": 1.9439769189323727e-05,
+      "loss": 1.2983,
+      "step": 72
+    },
+    {
+      "epoch": 1.1245136186770428,
+      "grad_norm": 7.546246025502454,
+      "learning_rate": 1.9419801996432896e-05,
+      "loss": 1.2559,
+      "step": 73
+    },
+    {
+      "epoch": 1.140077821011673,
+      "grad_norm": 5.50857701371075,
+      "learning_rate": 1.9399497021296094e-05,
+      "loss": 1.2612,
+      "step": 74
+    },
+    {
+      "epoch": 1.1556420233463034,
+      "grad_norm": 19.755888394433814,
+      "learning_rate": 1.937885507850018e-05,
+      "loss": 1.3324,
+      "step": 75
+    },
+    {
+      "epoch": 1.171206225680934,
+      "grad_norm": 7.035191415935576,
+      "learning_rate": 1.935787699615036e-05,
+      "loss": 1.3674,
+      "step": 76
+    },
+    {
+      "epoch": 1.1867704280155642,
+      "grad_norm": 6.756030563749288,
+      "learning_rate": 1.933656361583694e-05,
+      "loss": 1.183,
+      "step": 77
+    },
+    {
+      "epoch": 1.2023346303501945,
+      "grad_norm": 8.841867408302369,
+      "learning_rate": 1.931491579260158e-05,
+      "loss": 1.2485,
+      "step": 78
+    },
+    {
+      "epoch": 1.217898832684825,
+      "grad_norm": 5.387317154730147,
+      "learning_rate": 1.9292934394902992e-05,
+      "loss": 1.5021,
+      "step": 79
+    },
+    {
+      "epoch": 1.2334630350194553,
+      "grad_norm": 14.871501309985415,
+      "learning_rate": 1.9270620304582077e-05,
+      "loss": 1.5676,
+      "step": 80
+    },
+    {
+      "epoch": 1.2490272373540856,
+      "grad_norm": 10.997862717881436,
+      "learning_rate": 1.9247974416826585e-05,
+      "loss": 1.3707,
+      "step": 81
+    },
+    {
+      "epoch": 1.264591439688716,
+      "grad_norm": 7.2507989168336735,
+      "learning_rate": 1.922499764013518e-05,
+      "loss": 1.5842,
+      "step": 82
+    },
+    {
+      "epoch": 1.2801556420233462,
+      "grad_norm": 9.537768676771838,
+      "learning_rate": 1.920169089628099e-05,
+      "loss": 1.4067,
+      "step": 83
+    },
+    {
+      "epoch": 1.2957198443579767,
+      "grad_norm": 8.163551805381035,
+      "learning_rate": 1.9178055120274625e-05,
+      "loss": 1.2465,
+      "step": 84
+    },
+    {
+      "epoch": 1.311284046692607,
+      "grad_norm": 6.640431551160505,
+      "learning_rate": 1.9154091260326698e-05,
+      "loss": 1.3956,
+      "step": 85
+    },
+    {
+      "epoch": 1.3268482490272373,
+      "grad_norm": 26.20900323475498,
+      "learning_rate": 1.9129800277809742e-05,
+      "loss": 1.2936,
+      "step": 86
+    },
+    {
+      "epoch": 1.3424124513618678,
+      "grad_norm": 9.219414651764641,
+      "learning_rate": 1.910518314721967e-05,
+      "loss": 1.5675,
+      "step": 87
+    },
+    {
+      "epoch": 1.3579766536964981,
+      "grad_norm": 6.169958007250397,
+      "learning_rate": 1.9080240856136675e-05,
+      "loss": 1.51,
+      "step": 88
+    },
+    {
+      "epoch": 1.3735408560311284,
+      "grad_norm": 7.775254406585112,
+      "learning_rate": 1.9054974405185605e-05,
+      "loss": 1.5837,
+      "step": 89
+    },
+    {
+      "epoch": 1.3891050583657587,
+      "grad_norm": 6.000524431156958,
+      "learning_rate": 1.902938480799583e-05,
+      "loss": 1.239,
+      "step": 90
+    },
+    {
+      "epoch": 1.404669260700389,
+      "grad_norm": 6.445928086057122,
+      "learning_rate": 1.9003473091160557e-05,
+      "loss": 1.2915,
+      "step": 91
+    },
+    {
+      "epoch": 1.4202334630350195,
+      "grad_norm": 8.175071297255242,
+      "learning_rate": 1.8977240294195676e-05,
+      "loss": 1.5306,
+      "step": 92
+    },
+    {
+      "epoch": 1.4357976653696498,
+      "grad_norm": 5.720210778577455,
+      "learning_rate": 1.895068746949803e-05,
+      "loss": 1.3159,
+      "step": 93
+    },
+    {
+      "epoch": 1.45136186770428,
+      "grad_norm": 9.833046906184748,
+      "learning_rate": 1.8923815682303214e-05,
+      "loss": 1.553,
+      "step": 94
+    },
+    {
+      "epoch": 1.4669260700389106,
+      "grad_norm": 6.851775208018755,
+      "learning_rate": 1.8896626010642833e-05,
+      "loss": 1.5885,
+      "step": 95
+    },
+    {
+      "epoch": 1.482490272373541,
+      "grad_norm": 7.1890477833480295,
+      "learning_rate": 1.886911954530124e-05,
+      "loss": 1.5322,
+      "step": 96
+    },
+    {
+      "epoch": 1.4980544747081712,
+      "grad_norm": 8.094895085879692,
+      "learning_rate": 1.884129738977181e-05,
+      "loss": 1.326,
+      "step": 97
+    },
+    {
+      "epoch": 1.5136186770428015,
+      "grad_norm": 9.604855227288974,
+      "learning_rate": 1.8813160660212636e-05,
+      "loss": 1.4432,
+      "step": 98
+    },
+    {
+      "epoch": 1.5291828793774318,
+      "grad_norm": 10.150820230299194,
+      "learning_rate": 1.8784710485401775e-05,
+      "loss": 1.2174,
+      "step": 99
+    },
+    {
+      "epoch": 1.544747081712062,
+      "grad_norm": 9.238169258331935,
+      "learning_rate": 1.875594800669195e-05,
+      "loss": 1.2617,
+      "step": 100
+    },
+    {
+      "epoch": 1.5603112840466926,
+      "grad_norm": 9.441159749907698,
+      "learning_rate": 1.8726874377964764e-05,
+      "loss": 1.234,
+      "step": 101
+    },
+    {
+      "epoch": 1.575875486381323,
+      "grad_norm": 7.2474587450825,
+      "learning_rate": 1.869749076558442e-05,
+      "loss": 1.1766,
+      "step": 102
+    },
+    {
+      "epoch": 1.5914396887159534,
+      "grad_norm": 6.23841795541216,
+      "learning_rate": 1.8667798348350918e-05,
+      "loss": 1.2045,
+      "step": 103
+    },
+    {
+      "epoch": 1.6070038910505837,
+      "grad_norm": 6.554762465264489,
+      "learning_rate": 1.863779831745276e-05,
+      "loss": 1.1382,
+      "step": 104
+    },
+    {
+      "epoch": 1.622568093385214,
+      "grad_norm": 5.997892924979543,
+      "learning_rate": 1.8607491876419183e-05,
+      "loss": 1.4172,
+      "step": 105
+    },
+    {
+      "epoch": 1.6381322957198443,
+      "grad_norm": 6.371740787277364,
+      "learning_rate": 1.8576880241071852e-05,
+      "loss": 1.3726,
+      "step": 106
+    },
+    {
+      "epoch": 1.6536964980544746,
+      "grad_norm": 7.86316264854559,
+      "learning_rate": 1.8545964639476105e-05,
+      "loss": 1.3789,
+      "step": 107
+    },
+    {
+      "epoch": 1.669260700389105,
+      "grad_norm": 7.417730516932254,
+      "learning_rate": 1.851474631189167e-05,
+      "loss": 1.2634,
+      "step": 108
+    },
+    {
+      "epoch": 1.6848249027237354,
+      "grad_norm": 8.222005228354707,
+      "learning_rate": 1.848322651072291e-05,
+      "loss": 1.4934,
+      "step": 109
+    },
+    {
+      "epoch": 1.7003891050583657,
+      "grad_norm": 6.747214231413209,
+      "learning_rate": 1.8451406500468598e-05,
+      "loss": 1.2275,
+      "step": 110
+    },
+    {
+      "epoch": 1.7159533073929962,
+      "grad_norm": 7.522875546614491,
+      "learning_rate": 1.841928755767116e-05,
+      "loss": 1.4232,
+      "step": 111
+    },
+    {
+      "epoch": 1.7315175097276265,
+      "grad_norm": 8.395027504519643,
+      "learning_rate": 1.8386870970865488e-05,
+      "loss": 1.4617,
+      "step": 112
+    },
+    {
+      "epoch": 1.7470817120622568,
+      "grad_norm": 5.682565677158726,
+      "learning_rate": 1.835415804052724e-05,
+      "loss": 1.2782,
+      "step": 113
+    },
+    {
+      "epoch": 1.7626459143968871,
+      "grad_norm": 17.319126618522,
+      "learning_rate": 1.8321150079020656e-05,
+      "loss": 1.2252,
+      "step": 114
+    },
+    {
+      "epoch": 1.7782101167315174,
+      "grad_norm": 6.698269746955856,
+      "learning_rate": 1.8287848410545922e-05,
+      "loss": 1.2638,
+      "step": 115
+    },
+    {
+      "epoch": 1.7937743190661477,
+      "grad_norm": 5.389193784149641,
+      "learning_rate": 1.825425437108605e-05,
+      "loss": 1.2101,
+      "step": 116
+    },
+    {
+      "epoch": 1.8093385214007782,
+      "grad_norm": 7.556872728377763,
+      "learning_rate": 1.8220369308353255e-05,
+      "loss": 1.4187,
+      "step": 117
+    },
+    {
+      "epoch": 1.8249027237354085,
+      "grad_norm": 11.500517221563017,
+      "learning_rate": 1.8186194581734922e-05,
+      "loss": 1.2614,
+      "step": 118
+    },
+    {
+      "epoch": 1.840466926070039,
+      "grad_norm": 6.973436659657028,
+      "learning_rate": 1.815173156223906e-05,
+      "loss": 1.6287,
+      "step": 119
+    },
+    {
+      "epoch": 1.8560311284046693,
+      "grad_norm": 8.82430378736939,
+      "learning_rate": 1.811698163243929e-05,
+      "loss": 1.7683,
+      "step": 120
+    },
+    {
+      "epoch": 1.8715953307392996,
+      "grad_norm": 7.44523008567814,
+      "learning_rate": 1.8081946186419375e-05,
+      "loss": 1.3718,
+      "step": 121
+    },
+    {
+      "epoch": 1.88715953307393,
+      "grad_norm": 11.092035922405335,
+      "learning_rate": 1.804662662971732e-05,
+      "loss": 1.2244,
+      "step": 122
+    },
+    {
+      "epoch": 1.9027237354085602,
+      "grad_norm": 8.907024665470365,
+      "learning_rate": 1.801102437926896e-05,
+      "loss": 1.1516,
+      "step": 123
+    },
+    {
+      "epoch": 1.9182879377431905,
+      "grad_norm": 7.330064038520311,
+      "learning_rate": 1.797514086335113e-05,
+      "loss": 1.288,
+      "step": 124
+    },
+    {
+      "epoch": 1.933852140077821,
+      "grad_norm": 10.980569189715007,
+      "learning_rate": 1.7938977521524355e-05,
+      "loss": 1.3815,
+      "step": 125
+    },
+    {
+      "epoch": 1.9494163424124513,
+      "grad_norm": 8.792218570481833,
+      "learning_rate": 1.79025358045751e-05,
+      "loss": 1.5337,
+      "step": 126
+    },
+    {
+      "epoch": 1.9649805447470818,
+      "grad_norm": 8.759608210372287,
+      "learning_rate": 1.786581717445759e-05,
+      "loss": 1.4971,
+      "step": 127
+    },
+    {
+      "epoch": 1.9805447470817121,
+      "grad_norm": 6.719480961643205,
+      "learning_rate": 1.782882310423512e-05,
+      "loss": 1.3845,
+      "step": 128
+    },
+    {
+      "epoch": 1.9961089494163424,
+      "grad_norm": 11.98512915914497,
+      "learning_rate": 1.7791555078020992e-05,
+      "loss": 1.8083,
+      "step": 129
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 11.98512915914497,
+      "learning_rate": 1.7754014590918964e-05,
+      "loss": 0.3816,
+      "step": 130
+    },
+    {
+      "epoch": 2.0155642023346303,
+      "grad_norm": 6.076810203247505,
+      "learning_rate": 1.771620314896327e-05,
+      "loss": 1.1992,
+      "step": 131
+    },
+    {
+      "epoch": 2.0311284046692606,
+      "grad_norm": 5.799331945733901,
+      "learning_rate": 1.76781222690582e-05,
+      "loss": 0.9603,
+      "step": 132
+    },
+    {
+      "epoch": 2.046692607003891,
+      "grad_norm": 7.639810257162938,
+      "learning_rate": 1.763977347891725e-05,
+      "loss": 1.1609,
+      "step": 133
+    },
+    {
+      "epoch": 2.062256809338521,
+      "grad_norm": 4.641223961792758,
+      "learning_rate": 1.7601158317001835e-05,
+      "loss": 1.4175,
+      "step": 134
+    },
+    {
+      "epoch": 2.077821011673152,
+      "grad_norm": 6.543906827631292,
+      "learning_rate": 1.756227833245956e-05,
+      "loss": 1.1293,
+      "step": 135
+    },
+    {
+      "epoch": 2.093385214007782,
+      "grad_norm": 5.807214604460152,
+      "learning_rate": 1.752313508506208e-05,
+      "loss": 1.2959,
+      "step": 136
+    },
+    {
+      "epoch": 2.1089494163424125,
+      "grad_norm": 5.553563692538825,
+      "learning_rate": 1.748373014514253e-05,
+      "loss": 1.5164,
+      "step": 137
+    },
+    {
+      "epoch": 2.124513618677043,
+      "grad_norm": 5.570141549794821,
+      "learning_rate": 1.7444065093532507e-05,
+      "loss": 1.223,
+      "step": 138
+    },
+    {
+      "epoch": 2.140077821011673,
+      "grad_norm": 10.218543215792467,
+      "learning_rate": 1.740414152149868e-05,
+      "loss": 1.2693,
+      "step": 139
+    },
+    {
+      "epoch": 2.1556420233463034,
+      "grad_norm": 6.654829769858401,
+      "learning_rate": 1.736396103067893e-05,
+      "loss": 1.2822,
+      "step": 140
+    },
+    {
+      "epoch": 2.1712062256809337,
+      "grad_norm": 7.063879942686053,
+      "learning_rate": 1.73235252330181e-05,
+      "loss": 1.2313,
+      "step": 141
+    },
+    {
+      "epoch": 2.1867704280155644,
+      "grad_norm": 8.26459321928057,
+      "learning_rate": 1.728283575070333e-05,
+      "loss": 1.3606,
+      "step": 142
+    },
+    {
+      "epoch": 2.2023346303501947,
+      "grad_norm": 8.158351915542648,
+      "learning_rate": 1.7241894216098995e-05,
+      "loss": 1.1179,
+      "step": 143
+    },
+    {
+      "epoch": 2.217898832684825,
+      "grad_norm": 6.025748582779371,
+      "learning_rate": 1.720070227168118e-05,
+      "loss": 1.2041,
+      "step": 144
+    },
+    {
+      "epoch": 2.2334630350194553,
+      "grad_norm": 7.30387722610904,
+      "learning_rate": 1.7159261569971828e-05,
+      "loss": 1.2661,
+      "step": 145
+    },
+    {
+      "epoch": 2.2490272373540856,
+      "grad_norm": 6.274862129099997,
+      "learning_rate": 1.7117573773472418e-05,
+      "loss": 1.3209,
+      "step": 146
+    },
+    {
+      "epoch": 2.264591439688716,
+      "grad_norm": 8.052280930110639,
+      "learning_rate": 1.7075640554597278e-05,
+      "loss": 1.1796,
+      "step": 147
+    },
+    {
+      "epoch": 2.280155642023346,
+      "grad_norm": 10.893409682259657,
+      "learning_rate": 1.703346359560651e-05,
+      "loss": 1.4143,
+      "step": 148
+    },
+    {
+      "epoch": 2.2957198443579765,
+      "grad_norm": 9.090803934496224,
+      "learning_rate": 1.6991044588538455e-05,
+      "loss": 1.1129,
+      "step": 149
+    },
+    {
+      "epoch": 2.311284046692607,
+      "grad_norm": 9.407206024176414,
+      "learning_rate": 1.694838523514187e-05,
+      "loss": 1.1586,
+      "step": 150
+    },
+    {
+      "epoch": 2.3268482490272375,
+      "grad_norm": 8.517506234140628,
+      "learning_rate": 1.690548724680761e-05,
+      "loss": 1.4231,
+      "step": 151
+    },
+    {
+      "epoch": 2.342412451361868,
+      "grad_norm": 6.1621127583650255,
+      "learning_rate": 1.6862352344500004e-05,
+      "loss": 1.3577,
+      "step": 152
+    },
+    {
+      "epoch": 2.357976653696498,
+      "grad_norm": 37.87140785888283,
+      "learning_rate": 1.681898225868779e-05,
+      "loss": 1.3384,
+      "step": 153
+    },
+    {
+      "epoch": 2.3735408560311284,
+      "grad_norm": 8.232248393098434,
+      "learning_rate": 1.677537872927471e-05,
+      "loss": 1.3798,
+      "step": 154
+    },
+    {
+      "epoch": 2.3891050583657587,
+      "grad_norm": 8.407799337670626,
+      "learning_rate": 1.673154350552971e-05,
+      "loss": 1.3535,
+      "step": 155
+    },
+    {
+      "epoch": 2.404669260700389,
+      "grad_norm": 11.052092401395148,
+      "learning_rate": 1.6687478346016736e-05,
+      "loss": 1.3003,
+      "step": 156
+    },
+    {
+      "epoch": 2.4202334630350193,
+      "grad_norm": 11.914207501283393,
+      "learning_rate": 1.6643185018524227e-05,
+      "loss": 1.2803,
+      "step": 157
+    },
+    {
+      "epoch": 2.43579766536965,
+      "grad_norm": 9.369362265914104,
+      "learning_rate": 1.6598665299994162e-05,
+      "loss": 1.3889,
+      "step": 158
+    },
+    {
+      "epoch": 2.4513618677042803,
+      "grad_norm": 6.155749688526481,
+      "learning_rate": 1.655392097645079e-05,
+      "loss": 1.3065,
+      "step": 159
+    },
+    {
+      "epoch": 2.4669260700389106,
+      "grad_norm": 7.610662694751711,
+      "learning_rate": 1.6508953842928966e-05,
+      "loss": 1.3677,
+      "step": 160
+    },
+    {
+      "epoch": 2.482490272373541,
+      "grad_norm": 8.874434851480999,
+      "learning_rate": 1.6463765703402154e-05,
+      "loss": 1.1467,
+      "step": 161
+    },
+    {
+      "epoch": 2.498054474708171,
+      "grad_norm": 6.048721329156399,
+      "learning_rate": 1.6418358370710048e-05,
+      "loss": 1.2878,
+      "step": 162
+    },
+    {
+      "epoch": 2.5136186770428015,
+      "grad_norm": 8.012765512165412,
+      "learning_rate": 1.6372733666485842e-05,
+      "loss": 1.2369,
+      "step": 163
+    },
+    {
+      "epoch": 2.529182879377432,
+      "grad_norm": 9.98813005959193,
+      "learning_rate": 1.6326893421083157e-05,
+      "loss": 1.3142,
+      "step": 164
+    },
+    {
+      "epoch": 2.544747081712062,
+      "grad_norm": 6.989195393880204,
+      "learning_rate": 1.6280839473502607e-05,
+      "loss": 1.2651,
+      "step": 165
+    },
+    {
+      "epoch": 2.5603112840466924,
+      "grad_norm": 13.328346097167426,
+      "learning_rate": 1.6234573671318027e-05,
+      "loss": 1.5734,
+      "step": 166
+    },
+    {
+      "epoch": 2.5758754863813227,
+      "grad_norm": 7.47939944848892,
+      "learning_rate": 1.6188097870602344e-05,
+      "loss": 1.1876,
+      "step": 167
+    },
+    {
+      "epoch": 2.5914396887159534,
+      "grad_norm": 9.813447558487908,
+      "learning_rate": 1.614141393585313e-05,
+      "loss": 1.2428,
+      "step": 168
+    },
+    {
+      "epoch": 2.6070038910505837,
+      "grad_norm": 8.170087254779242,
+      "learning_rate": 1.6094523739917797e-05,
+      "loss": 1.2607,
+      "step": 169
+    },
+    {
+      "epoch": 2.622568093385214,
+      "grad_norm": 6.143715398790098,
+      "learning_rate": 1.6047429163918444e-05,
+      "loss": 1.3353,
+      "step": 170
+    },
+    {
+      "epoch": 2.6381322957198443,
+      "grad_norm": 17.125656609647287,
+      "learning_rate": 1.600013209717642e-05,
+      "loss": 1.4865,
+      "step": 171
+    },
+    {
+      "epoch": 2.6536964980544746,
+      "grad_norm": 10.776858476646886,
+      "learning_rate": 1.5952634437136523e-05,
+      "loss": 1.5398,
+      "step": 172
+    },
+    {
+      "epoch": 2.669260700389105,
+      "grad_norm": 7.0818183172701605,
+      "learning_rate": 1.5904938089290864e-05,
+      "loss": 1.3059,
+      "step": 173
+    },
+    {
+      "epoch": 2.6848249027237356,
+      "grad_norm": 20.107104125600667,
+      "learning_rate": 1.5857044967102423e-05,
+      "loss": 1.325,
+      "step": 174
+    },
+    {
+      "epoch": 2.700389105058366,
+      "grad_norm": 9.685536590494742,
+      "learning_rate": 1.580895699192831e-05,
+      "loss": 1.2209,
+      "step": 175
+    },
+    {
+      "epoch": 2.7159533073929962,
+      "grad_norm": 14.272302800628733,
+      "learning_rate": 1.5760676092942663e-05,
+      "loss": 1.1283,
+      "step": 176
+    },
+    {
+      "epoch": 2.7315175097276265,
+      "grad_norm": 8.529426981120846,
+      "learning_rate": 1.571220420705926e-05,
+      "loss": 1.1967,
+      "step": 177
+    },
+    {
+      "epoch": 2.747081712062257,
+      "grad_norm": 9.25820017440293,
+      "learning_rate": 1.5663543278853818e-05,
+      "loss": 1.2561,
+      "step": 178
+    },
+    {
+      "epoch": 2.762645914396887,
+      "grad_norm": 6.7665962336727725,
+      "learning_rate": 1.5614695260485973e-05,
+      "loss": 1.1381,
+      "step": 179
+    },
+    {
+      "epoch": 2.7782101167315174,
+      "grad_norm": 11.677157670781277,
+      "learning_rate": 1.5565662111620967e-05,
+      "loss": 1.123,
+      "step": 180
+    },
+    {
+      "epoch": 2.7937743190661477,
+      "grad_norm": 10.612616450023706,
+      "learning_rate": 1.5516445799351046e-05,
+      "loss": 1.4241,
+      "step": 181
+    },
+    {
+      "epoch": 2.809338521400778,
+      "grad_norm": 5.785097669619218,
+      "learning_rate": 1.5467048298116516e-05,
+      "loss": 1.1191,
+      "step": 182
+    },
+    {
+      "epoch": 2.8249027237354083,
+      "grad_norm": 19.456658401658856,
+      "learning_rate": 1.5417471589626563e-05,
+      "loss": 1.0679,
+      "step": 183
+    },
+    {
+      "epoch": 2.840466926070039,
+      "grad_norm": 8.020315674017025,
+      "learning_rate": 1.5367717662779732e-05,
+      "loss": 1.2059,
+      "step": 184
+    },
+    {
+      "epoch": 2.8560311284046693,
+      "grad_norm": 9.637600750103296,
+      "learning_rate": 1.531778851358414e-05,
+      "loss": 1.1613,
+      "step": 185
+    },
+    {
+      "epoch": 2.8715953307392996,
+      "grad_norm": 6.693770146874348,
+      "learning_rate": 1.5267686145077406e-05,
+      "loss": 1.127,
+      "step": 186
+    },
+    {
+      "epoch": 2.88715953307393,
+      "grad_norm": 18.137993374614158,
+      "learning_rate": 1.5217412567246298e-05,
+      "loss": 1.5315,
+      "step": 187
+    },
+    {
+      "epoch": 2.90272373540856,
+      "grad_norm": 9.571406690209106,
+      "learning_rate": 1.5166969796946087e-05,
+      "loss": 1.1774,
+      "step": 188
+    },
+    {
+      "epoch": 2.9182879377431905,
+      "grad_norm": 19.959548104529457,
+      "learning_rate": 1.5116359857819635e-05,
+      "loss": 1.3707,
+      "step": 189
+    },
+    {
+      "epoch": 2.9338521400778212,
+      "grad_norm": 16.169561781876904,
+      "learning_rate": 1.5065584780216225e-05,
+      "loss": 1.3618,
+      "step": 190
+    },
+    {
+      "epoch": 2.9494163424124515,
+      "grad_norm": 7.308704556972184,
+      "learning_rate": 1.501464660111009e-05,
+      "loss": 1.5718,
+      "step": 191
+    },
+    {
+      "epoch": 2.964980544747082,
+      "grad_norm": 11.175947473356464,
+      "learning_rate": 1.4963547364018711e-05,
+      "loss": 1.1821,
+      "step": 192
+    },
+    {
+      "epoch": 2.980544747081712,
+      "grad_norm": 6.880501518603281,
+      "learning_rate": 1.4912289118920821e-05,
+      "loss": 1.2976,
+      "step": 193
+    },
+    {
+      "epoch": 2.9961089494163424,
+      "grad_norm": 16.45488939160082,
+      "learning_rate": 1.4860873922174188e-05,
+      "loss": 1.5251,
+      "step": 194
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 16.45488939160082,
+      "learning_rate": 1.4809303836433086e-05,
+      "loss": 0.3486,
+      "step": 195
+    },
+    {
+      "epoch": 3.0155642023346303,
+      "grad_norm": 17.26667705122932,
+      "learning_rate": 1.4757580930565569e-05,
+      "loss": 1.2262,
+      "step": 196
+    },
+    {
+      "epoch": 3.0311284046692606,
+      "grad_norm": 7.079930630802389,
+      "learning_rate": 1.4705707279570476e-05,
+      "loss": 1.0964,
+      "step": 197
+    },
+    {
+      "epoch": 3.046692607003891,
+      "grad_norm": 5.9707164634211045,
+      "learning_rate": 1.4653684964494163e-05,
+      "loss": 1.1285,
+      "step": 198
+    },
+    {
+      "epoch": 3.062256809338521,
+      "grad_norm": 7.374420248201497,
+      "learning_rate": 1.460151607234705e-05,
+      "loss": 1.1528,
+      "step": 199
+    },
+    {
+      "epoch": 3.077821011673152,
+      "grad_norm": 7.823768857549868,
+      "learning_rate": 1.4549202696019868e-05,
+      "loss": 1.6089,
+      "step": 200
+    },
+    {
+      "epoch": 3.093385214007782,
+      "grad_norm": 13.095423850316486,
+      "learning_rate": 1.44967469341997e-05,
+      "loss": 1.4053,
+      "step": 201
+    },
+    {
+      "epoch": 3.1089494163424125,
+      "grad_norm": 8.57838469081796,
+      "learning_rate": 1.4444150891285809e-05,
+      "loss": 1.3911,
+      "step": 202
+    },
+    {
+      "epoch": 3.124513618677043,
+      "grad_norm": 10.492752547850309,
+      "learning_rate": 1.4391416677305183e-05,
+      "loss": 1.7466,
+      "step": 203
+    },
+    {
+      "epoch": 3.140077821011673,
+      "grad_norm": 12.04907785845439,
+      "learning_rate": 1.4338546407827912e-05,
+      "loss": 1.3069,
+      "step": 204
+    },
+    {
+      "epoch": 3.1556420233463034,
+      "grad_norm": 9.082131491175343,
+      "learning_rate": 1.4285542203882301e-05,
+      "loss": 1.3909,
+      "step": 205
+    },
+    {
+      "epoch": 3.1712062256809337,
+      "grad_norm": 7.8265612070424035,
+      "learning_rate": 1.4232406191869786e-05,
+      "loss": 1.1016,
+      "step": 206
+    },
+    {
+      "epoch": 3.1867704280155644,
+      "grad_norm": 7.017081240597243,
+      "learning_rate": 1.4179140503479622e-05,
+      "loss": 1.3007,
+      "step": 207
+    },
+    {
+      "epoch": 3.2023346303501947,
+      "grad_norm": 11.222861362951033,
+      "learning_rate": 1.4125747275603384e-05,
+      "loss": 1.4584,
+      "step": 208
+    },
+    {
+      "epoch": 3.217898832684825,
+      "grad_norm": 8.228254518059948,
+      "learning_rate": 1.4072228650249205e-05,
+      "loss": 1.1437,
+      "step": 209
+    },
+    {
+      "epoch": 3.2334630350194553,
+      "grad_norm": 5.620105281993818,
+      "learning_rate": 1.4018586774455876e-05,
+      "loss": 1.0801,
+      "step": 210
+    },
+    {
+      "epoch": 3.2490272373540856,
+      "grad_norm": 8.270265614809485,
+      "learning_rate": 1.3964823800206698e-05,
+      "loss": 1.4172,
+      "step": 211
+    },
+    {
+      "epoch": 3.264591439688716,
+      "grad_norm": 10.759581180311525,
+      "learning_rate": 1.3910941884343144e-05,
+      "loss": 1.3431,
+      "step": 212
+    },
+    {
+      "epoch": 3.280155642023346,
+      "grad_norm": 9.256121877658135,
+      "learning_rate": 1.3856943188478353e-05,
+      "loss": 1.1614,
+      "step": 213
+    },
+    {
+      "epoch": 3.2957198443579765,
+      "grad_norm": 8.278774263244427,
+      "learning_rate": 1.3802829878910387e-05,
+      "loss": 1.5056,
+      "step": 214
+    },
+    {
+      "epoch": 3.311284046692607,
+      "grad_norm": 7.988122466692388,
+      "learning_rate": 1.3748604126535335e-05,
+      "loss": 1.1658,
+      "step": 215
+    },
+    {
+      "epoch": 3.3268482490272375,
+      "grad_norm": 25.69960791189383,
+      "learning_rate": 1.3694268106760225e-05,
+      "loss": 1.5151,
+      "step": 216
+    },
+    {
+      "epoch": 3.342412451361868,
+      "grad_norm": 6.716460926595037,
+      "learning_rate": 1.3639823999415744e-05,
+      "loss": 1.2914,
+      "step": 217
+    },
+    {
+      "epoch": 3.357976653696498,
+      "grad_norm": 16.08954441850179,
+      "learning_rate": 1.3585273988668804e-05,
+      "loss": 1.2714,
+      "step": 218
+    },
+    {
+      "epoch": 3.3735408560311284,
+      "grad_norm": 7.277025513470879,
+      "learning_rate": 1.3530620262934892e-05,
+      "loss": 1.3116,
+      "step": 219
+    },
+    {
+      "epoch": 3.3891050583657587,
+      "grad_norm": 10.39394733050087,
+      "learning_rate": 1.3475865014790303e-05,
+      "loss": 1.3044,
+      "step": 220
+    },
+    {
+      "epoch": 3.404669260700389,
+      "grad_norm": 6.793016895617423,
+      "learning_rate": 1.342101044088416e-05,
+      "loss": 1.6471,
+      "step": 221
+    },
+    {
+      "epoch": 3.4202334630350193,
+      "grad_norm": 8.307309122049785,
+      "learning_rate": 1.3366058741850302e-05,
+      "loss": 1.1521,
+      "step": 222
+    },
+    {
+      "epoch": 3.43579766536965,
+      "grad_norm": 11.961031173730973,
+      "learning_rate": 1.3311012122218995e-05,
+      "loss": 1.4236,
+      "step": 223
+    },
+    {
+      "epoch": 3.4513618677042803,
+      "grad_norm": 12.085462142899722,
+      "learning_rate": 1.3255872790328485e-05,
+      "loss": 1.4304,
+      "step": 224
+    },
+    {
+      "epoch": 3.4669260700389106,
+      "grad_norm": 12.043766601436674,
+      "learning_rate": 1.320064295823642e-05,
+      "loss": 1.105,
+      "step": 225
+    },
+    {
+      "epoch": 3.482490272373541,
+      "grad_norm": 6.830210730411899,
+      "learning_rate": 1.3145324841631093e-05,
+      "loss": 1.0992,
+      "step": 226
+    },
+    {
+      "epoch": 3.498054474708171,
+      "grad_norm": 5.183807189599878,
+      "learning_rate": 1.3089920659742561e-05,
+      "loss": 1.1104,
+      "step": 227
+    },
+    {
+      "epoch": 3.5136186770428015,
+      "grad_norm": 8.173248197932283,
+      "learning_rate": 1.3034432635253615e-05,
+      "loss": 1.3416,
+      "step": 228
+    },
+    {
+      "epoch": 3.529182879377432,
+      "grad_norm": 6.840510399122009,
+      "learning_rate": 1.2978862994210609e-05,
+      "loss": 1.1301,
+      "step": 229
+    },
+    {
+      "epoch": 3.544747081712062,
+      "grad_norm": 6.795335624835114,
+      "learning_rate": 1.2923213965934158e-05,
+      "loss": 1.1824,
+      "step": 230
+    },
+    {
+      "epoch": 3.5603112840466924,
+      "grad_norm": 12.487628895610484,
+      "learning_rate": 1.2867487782929702e-05,
+      "loss": 1.4,
+      "step": 231
+    },
+    {
+      "epoch": 3.5758754863813227,
+      "grad_norm": 7.498953182696051,
+      "learning_rate": 1.2811686680797942e-05,
+      "loss": 1.0117,
+      "step": 232
+    },
+    {
+      "epoch": 3.5914396887159534,
+      "grad_norm": 11.599970960534753,
+      "learning_rate": 1.2755812898145157e-05,
+      "loss": 1.457,
+      "step": 233
+    },
+    {
+      "epoch": 3.6070038910505837,
+      "grad_norm": 10.728915546115905,
+      "learning_rate": 1.269986867649339e-05,
+      "loss": 1.0275,
+      "step": 234
+    },
+    {
+      "epoch": 3.622568093385214,
+      "grad_norm": 7.212471068564826,
+      "learning_rate": 1.2643856260190533e-05,
+      "loss": 1.2426,
+      "step": 235
+    },
+    {
+      "epoch": 3.6381322957198443,
+      "grad_norm": 7.257310852456952,
+      "learning_rate": 1.2587777896320279e-05,
+      "loss": 1.1293,
+      "step": 236
+    },
+    {
+      "epoch": 3.6536964980544746,
+      "grad_norm": 5.41922031825349,
+      "learning_rate": 1.2531635834611981e-05,
+      "loss": 1.2113,
+      "step": 237
+    },
+    {
+      "epoch": 3.669260700389105,
+      "grad_norm": 6.480868678608291,
+      "learning_rate": 1.2475432327350396e-05,
+      "loss": 1.2706,
+      "step": 238
+    },
+    {
+      "epoch": 3.6848249027237356,
+      "grad_norm": 6.80783495792981,
+      "learning_rate": 1.2419169629285335e-05,
+      "loss": 1.0698,
+      "step": 239
+    },
+    {
+      "epoch": 3.700389105058366,
+      "grad_norm": 7.234690686703731,
+      "learning_rate": 1.236284999754119e-05,
+      "loss": 1.4999,
+      "step": 240
+    },
+    {
+      "epoch": 3.7159533073929962,
+      "grad_norm": 9.331788321366725,
+      "learning_rate": 1.2306475691526407e-05,
+      "loss": 1.0748,
+      "step": 241
+    },
+    {
+      "epoch": 3.7315175097276265,
+      "grad_norm": 7.024835954207243,
+      "learning_rate": 1.2250048972842823e-05,
+      "loss": 1.241,
+      "step": 242
+    },
+    {
+      "epoch": 3.747081712062257,
+      "grad_norm": 8.381718396842675,
+      "learning_rate": 1.2193572105194953e-05,
+      "loss": 1.4368,
+      "step": 243
+    },
+    {
+      "epoch": 3.762645914396887,
+      "grad_norm": 8.467657085467351,
+      "learning_rate": 1.2137047354299165e-05,
+      "loss": 1.1316,
+      "step": 244
+    },
+    {
+      "epoch": 3.7782101167315174,
+      "grad_norm": 6.56413902354136,
+      "learning_rate": 1.2080476987792786e-05,
+      "loss": 1.2924,
+      "step": 245
+    },
+    {
+      "epoch": 3.7937743190661477,
+      "grad_norm": 11.457150627414464,
+      "learning_rate": 1.2023863275143138e-05,
+      "loss": 1.1807,
+      "step": 246
+    },
+    {
+      "epoch": 3.809338521400778,
+      "grad_norm": 11.784032118298212,
+      "learning_rate": 1.1967208487556477e-05,
+      "loss": 1.3601,
+      "step": 247
+    },
+    {
+      "epoch": 3.8249027237354083,
+      "grad_norm": 7.618595171387596,
+      "learning_rate": 1.1910514897886892e-05,
+      "loss": 1.0757,
+      "step": 248
+    },
+    {
+      "epoch": 3.840466926070039,
+      "grad_norm": 15.913563034938784,
+      "learning_rate": 1.1853784780545123e-05,
+      "loss": 1.381,
+      "step": 249
+    },
+    {
+      "epoch": 3.8560311284046693,
+      "grad_norm": 9.448026664890813,
+      "learning_rate": 1.1797020411407303e-05,
+      "loss": 1.1996,
+      "step": 250
+    },
+    {
+      "epoch": 3.8715953307392996,
+      "grad_norm": 9.36740858593225,
+      "learning_rate": 1.1740224067723676e-05,
+      "loss": 1.3333,
+      "step": 251
+    },
+    {
+      "epoch": 3.88715953307393,
+      "grad_norm": 6.202904532995713,
+      "learning_rate": 1.1683398028027218e-05,
+      "loss": 1.0989,
+      "step": 252
+    },
+    {
+      "epoch": 3.90272373540856,
+      "grad_norm": 6.807434814034836,
+      "learning_rate": 1.162654457204224e-05,
+      "loss": 1.0997,
+      "step": 253
+    },
+    {
+      "epoch": 3.9182879377431905,
+      "grad_norm": 7.474757873337292,
+      "learning_rate": 1.1569665980592934e-05,
+      "loss": 1.3777,
+      "step": 254
+    },
+    {
+      "epoch": 3.9338521400778212,
+      "grad_norm": 10.348844472235802,
+      "learning_rate": 1.1512764535511862e-05,
+      "loss": 1.4729,
+      "step": 255
+    },
+    {
+      "epoch": 3.9494163424124515,
+      "grad_norm": 9.595077808457091,
+      "learning_rate": 1.1455842519548417e-05,
+      "loss": 1.1649,
+      "step": 256
+    },
+    {
+      "epoch": 3.964980544747082,
+      "grad_norm": 12.298828537750817,
+      "learning_rate": 1.139890221627725e-05,
+      "loss": 1.1849,
+      "step": 257
+    },
+    {
+      "epoch": 3.980544747081712,
+      "grad_norm": 9.042727283207734,
+      "learning_rate": 1.1341945910006656e-05,
+      "loss": 1.3065,
+      "step": 258
+    },
+    {
+      "epoch": 3.9961089494163424,
+      "grad_norm": 7.530830029119465,
+      "learning_rate": 1.1284975885686926e-05,
+      "loss": 1.2184,
+      "step": 259
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 7.889222687866491,
+      "learning_rate": 1.1227994428818692e-05,
+      "loss": 0.4148,
+      "step": 260
+    },
+    {
+      "epoch": 4.01556420233463,
+      "grad_norm": 7.442531502305794,
+      "learning_rate": 1.1171003825361233e-05,
+      "loss": 1.2908,
+      "step": 261
+    },
+    {
+      "epoch": 4.031128404669261,
+      "grad_norm": 5.5861970223083945,
+      "learning_rate": 1.1114006361640763e-05,
+      "loss": 1.1309,
+      "step": 262
+    },
+    {
+      "epoch": 4.046692607003891,
+      "grad_norm": 7.19625969261882,
+      "learning_rate": 1.105700432425871e-05,
+      "loss": 1.149,
+      "step": 263
+    },
+    {
+      "epoch": 4.062256809338521,
+      "grad_norm": 7.319827903412528,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.2781,
+      "step": 264
+    },
+    {
+      "epoch": 4.0778210116731515,
+      "grad_norm": 8.579128780187071,
+      "learning_rate": 1.094299567574129e-05,
+      "loss": 1.3237,
+      "step": 265
+    },
+    {
+      "epoch": 4.093385214007782,
+      "grad_norm": 6.407050357928513,
+      "learning_rate": 1.0885993638359242e-05,
+      "loss": 1.4427,
+      "step": 266
+    },
+    {
+      "epoch": 4.108949416342412,
+      "grad_norm": 8.132256321978414,
+      "learning_rate": 1.0828996174638768e-05,
+      "loss": 1.394,
+      "step": 267
+    },
+    {
+      "epoch": 4.124513618677042,
+      "grad_norm": 6.3812196369738645,
+      "learning_rate": 1.0772005571181313e-05,
+      "loss": 1.1583,
+      "step": 268
+    },
+    {
+      "epoch": 4.1400778210116735,
+      "grad_norm": 13.85707637253458,
+      "learning_rate": 1.0715024114313077e-05,
+      "loss": 1.4429,
+      "step": 269
+    },
+    {
+      "epoch": 4.155642023346304,
+      "grad_norm": 21.528973223922698,
+      "learning_rate": 1.0658054089993349e-05,
+      "loss": 1.4796,
+      "step": 270
+    },
+    {
+      "epoch": 4.171206225680934,
+      "grad_norm": 6.331202799716651,
+      "learning_rate": 1.0601097783722751e-05,
+      "loss": 1.2535,
+      "step": 271
+    },
+    {
+      "epoch": 4.186770428015564,
+      "grad_norm": 6.192886059438914,
+      "learning_rate": 1.0544157480451586e-05,
+      "loss": 1.365,
+      "step": 272
+    },
+    {
+      "epoch": 4.202334630350195,
+      "grad_norm": 9.092962942972557,
+      "learning_rate": 1.048723546448814e-05,
+      "loss": 1.2418,
+      "step": 273
+    },
+    {
+      "epoch": 4.217898832684825,
+      "grad_norm": 9.088713897338243,
+      "learning_rate": 1.0430334019407069e-05,
+      "loss": 1.5284,
+      "step": 274
+    },
+    {
+      "epoch": 4.233463035019455,
+      "grad_norm": 7.935649176333802,
+      "learning_rate": 1.0373455427957762e-05,
+      "loss": 1.1285,
+      "step": 275
+    },
+    {
+      "epoch": 4.249027237354086,
+      "grad_norm": 6.747077137812801,
+      "learning_rate": 1.0316601971972787e-05,
+      "loss": 1.1133,
+      "step": 276
+    },
+    {
+      "epoch": 4.264591439688716,
+      "grad_norm": 6.190409945190735,
+      "learning_rate": 1.0259775932276325e-05,
+      "loss": 1.309,
+      "step": 277
+    },
+    {
+      "epoch": 4.280155642023346,
+      "grad_norm": 7.659504834977579,
+      "learning_rate": 1.0202979588592702e-05,
+      "loss": 1.134,
+      "step": 278
+    },
+    {
+      "epoch": 4.2957198443579765,
+      "grad_norm": 6.035623878393655,
+      "learning_rate": 1.0146215219454882e-05,
+      "loss": 1.1742,
+      "step": 279
+    },
+    {
+      "epoch": 4.311284046692607,
+      "grad_norm": 14.743602784497947,
+      "learning_rate": 1.0089485102113113e-05,
+      "loss": 1.2349,
+      "step": 280
+    },
+    {
+      "epoch": 4.326848249027237,
+      "grad_norm": 34.32059780452938,
+      "learning_rate": 1.0032791512443527e-05,
+      "loss": 1.1312,
+      "step": 281
+    },
+    {
+      "epoch": 4.342412451361867,
+      "grad_norm": 6.537432858619071,
+      "learning_rate": 9.976136724856869e-06,
+      "loss": 1.2261,
+      "step": 282
+    },
+    {
+      "epoch": 4.357976653696498,
+      "grad_norm": 6.805760395487357,
+      "learning_rate": 9.919523012207217e-06,
+      "loss": 1.2109,
+      "step": 283
+    },
+    {
+      "epoch": 4.373540856031129,
+      "grad_norm": 8.577713820915738,
+      "learning_rate": 9.862952645700841e-06,
+      "loss": 1.4719,
+      "step": 284
+    },
+    {
+      "epoch": 4.389105058365759,
+      "grad_norm": 5.490284251492559,
+      "learning_rate": 9.806427894805048e-06,
+      "loss": 1.312,
+      "step": 285
+    },
+    {
+      "epoch": 4.404669260700389,
+      "grad_norm": 8.327345481649647,
+      "learning_rate": 9.74995102715718e-06,
+      "loss": 1.2461,
+      "step": 286
+    },
+    {
+      "epoch": 4.42023346303502,
+      "grad_norm": 6.111747370932479,
+      "learning_rate": 9.693524308473596e-06,
+      "loss": 1.3926,
+      "step": 287
+    },
+    {
+      "epoch": 4.43579766536965,
+      "grad_norm": 8.141202180806642,
+      "learning_rate": 9.637150002458813e-06,
+      "loss": 1.2008,
+      "step": 288
+    },
+    {
+      "epoch": 4.45136186770428,
+      "grad_norm": 9.12454930317004,
+      "learning_rate": 9.58083037071467e-06,
+      "loss": 1.095,
+      "step": 289
+    },
+    {
+      "epoch": 4.466926070038911,
+      "grad_norm": 8.109583837297146,
+      "learning_rate": 9.524567672649606e-06,
+      "loss": 1.1697,
+      "step": 290
+    },
+    {
+      "epoch": 4.482490272373541,
+      "grad_norm": 6.940093378941827,
+      "learning_rate": 9.468364165388022e-06,
+      "loss": 1.5673,
+      "step": 291
+    },
+    {
+      "epoch": 4.498054474708171,
+      "grad_norm": 8.029108149503637,
+      "learning_rate": 9.412222103679728e-06,
+      "loss": 1.4509,
+      "step": 292
+    },
+    {
+      "epoch": 4.5136186770428015,
+      "grad_norm": 7.3797751032553345,
+      "learning_rate": 9.356143739809472e-06,
+      "loss": 1.1467,
+      "step": 293
+    },
+    {
+      "epoch": 4.529182879377432,
+      "grad_norm": 11.435244944138505,
+      "learning_rate": 9.300131323506617e-06,
+      "loss": 1.1765,
+      "step": 294
+    },
+    {
+      "epoch": 4.544747081712062,
+      "grad_norm": 9.332357406857541,
+      "learning_rate": 9.244187101854846e-06,
+      "loss": 1.2847,
+      "step": 295
+    },
+    {
+      "epoch": 4.560311284046692,
+      "grad_norm": 7.009210489706155,
+      "learning_rate": 9.188313319202057e-06,
+      "loss": 1.3632,
+      "step": 296
+    },
+    {
+      "epoch": 4.575875486381323,
+      "grad_norm": 8.623706053802827,
+      "learning_rate": 9.132512217070301e-06,
+      "loss": 1.3002,
+      "step": 297
+    },
+    {
+      "epoch": 4.591439688715953,
+      "grad_norm": 7.10477639818194,
+      "learning_rate": 9.076786034065843e-06,
+      "loss": 1.2559,
+      "step": 298
+    },
+    {
+      "epoch": 4.607003891050583,
+      "grad_norm": 16.24344498357353,
+      "learning_rate": 9.021137005789393e-06,
+      "loss": 1.3538,
+      "step": 299
+    },
+    {
+      "epoch": 4.622568093385214,
+      "grad_norm": 5.84525864201727,
+      "learning_rate": 8.965567364746388e-06,
+      "loss": 1.2,
+      "step": 300
+    },
+    {
+      "epoch": 4.638132295719844,
+      "grad_norm": 14.06325953577117,
+      "learning_rate": 8.910079340257442e-06,
+      "loss": 1.1461,
+      "step": 301
+    },
+    {
+      "epoch": 4.653696498054475,
+      "grad_norm": 7.743168891211479,
+      "learning_rate": 8.854675158368908e-06,
+      "loss": 1.3149,
+      "step": 302
+    },
+    {
+      "epoch": 4.669260700389105,
+      "grad_norm": 5.646503833678955,
+      "learning_rate": 8.799357041763583e-06,
+      "loss": 1.1332,
+      "step": 303
+    },
+    {
+      "epoch": 4.684824902723736,
+      "grad_norm": 7.846526869237445,
+      "learning_rate": 8.744127209671516e-06,
+      "loss": 1.2928,
+      "step": 304
+    },
+    {
+      "epoch": 4.700389105058366,
+      "grad_norm": 24.312874392544217,
+      "learning_rate": 8.688987877781008e-06,
+      "loss": 1.4022,
+      "step": 305
+    },
+    {
+      "epoch": 4.715953307392996,
+      "grad_norm": 11.006982390019688,
+      "learning_rate": 8.633941258149699e-06,
+      "loss": 1.6169,
+      "step": 306
+    },
+    {
+      "epoch": 4.7315175097276265,
+      "grad_norm": 8.161038346958115,
+      "learning_rate": 8.578989559115842e-06,
+      "loss": 1.1205,
+      "step": 307
+    },
+    {
+      "epoch": 4.747081712062257,
+      "grad_norm": 10.360221596629799,
+      "learning_rate": 8.524134985209698e-06,
+      "loss": 1.1426,
+      "step": 308
+    },
+    {
+      "epoch": 4.762645914396887,
+      "grad_norm": 8.086983191969743,
+      "learning_rate": 8.46937973706511e-06,
+      "loss": 1.1742,
+      "step": 309
+    },
+    {
+      "epoch": 4.778210116731517,
+      "grad_norm": 7.93821224397039,
+      "learning_rate": 8.414726011331197e-06,
+      "loss": 1.3003,
+      "step": 310
+    },
+    {
+      "epoch": 4.793774319066148,
+      "grad_norm": 6.8964113141193195,
+      "learning_rate": 8.360176000584257e-06,
+      "loss": 1.3921,
+      "step": 311
+    },
+    {
+      "epoch": 4.809338521400778,
+      "grad_norm": 9.541918636824354,
+      "learning_rate": 8.30573189323978e-06,
+      "loss": 1.4406,
+      "step": 312
+    },
+    {
+      "epoch": 4.824902723735408,
+      "grad_norm": 9.88288044965242,
+      "learning_rate": 8.251395873464671e-06,
+      "loss": 1.3104,
+      "step": 313
+    },
+    {
+      "epoch": 4.840466926070039,
+      "grad_norm": 7.034997485501755,
+      "learning_rate": 8.197170121089617e-06,
+      "loss": 1.4781,
+      "step": 314
+    },
+    {
+      "epoch": 4.856031128404669,
+      "grad_norm": 7.024182095026262,
+      "learning_rate": 8.143056811521653e-06,
+      "loss": 1.0645,
+      "step": 315
+    },
+    {
+      "epoch": 4.8715953307393,
+      "grad_norm": 6.763562775528538,
+      "learning_rate": 8.089058115656859e-06,
+      "loss": 1.1622,
+      "step": 316
+    },
+    {
+      "epoch": 4.88715953307393,
+      "grad_norm": 22.915236408148893,
+      "learning_rate": 8.035176199793309e-06,
+      "loss": 1.2201,
+      "step": 317
+    },
+    {
+      "epoch": 4.902723735408561,
+      "grad_norm": 22.9486570021447,
+      "learning_rate": 7.981413225544128e-06,
+      "loss": 1.3326,
+      "step": 318
+    },
+    {
+      "epoch": 4.918287937743191,
+      "grad_norm": 7.766492426098674,
+      "learning_rate": 7.9277713497508e-06,
+      "loss": 1.3444,
+      "step": 319
+    },
+    {
+      "epoch": 4.933852140077821,
+      "grad_norm": 8.922409990504782,
+      "learning_rate": 7.87425272439662e-06,
+      "loss": 1.2533,
+      "step": 320
+    },
+    {
+      "epoch": 4.9494163424124515,
+      "grad_norm": 10.120435136565488,
+      "learning_rate": 7.82085949652038e-06,
+      "loss": 1.4161,
+      "step": 321
+    },
+    {
+      "epoch": 4.964980544747082,
+      "grad_norm": 11.585619162092177,
+      "learning_rate": 7.767593808130216e-06,
+      "loss": 1.408,
+      "step": 322
+    },
+    {
+      "epoch": 4.980544747081712,
+      "grad_norm": 8.068469524272025,
+      "learning_rate": 7.714457796117705e-06,
+      "loss": 1.175,
+      "step": 323
+    },
+    {
+      "epoch": 4.996108949416342,
+      "grad_norm": 6.472014606610803,
+      "learning_rate": 7.661453592172093e-06,
+      "loss": 1.356,
+      "step": 324
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 6.472014606610803,
+      "learning_rate": 7.60858332269482e-06,
+      "loss": 0.4053,
+      "step": 325
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 512,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 8,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 176455044300800.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55a4adaab8848e3bd314a2a5b4b72c4727f222798a39a203c1916539b3a3cc70
+size 8120

zero_to_fp32.py

@@ -0,0 +1,760 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example:
+#   python zero_to_fp32.py . output_dir/
+#   or
+#   python zero_to_fp32.py . output_dir/ --safe_serialization
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+import gc
+import json
+import numpy as np
+from tqdm import tqdm
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device, weights_only=False)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+    total_files = len(files)
+    state_dicts = []
+    for f in tqdm(files, desc='Loading checkpoint shards'):
+        state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+class GatheredTensor:
+    """
+    A pseudo tensor that collects partitioned weights.
+    It is more memory efficient when there are multiple groups.
+    """
+
+    def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
+        self.flat_groups = flat_groups
+        self.flat_groups_offset = flat_groups_offset
+        self.offset = offset
+        self.partitioned_numel = partitioned_numel
+        self.shape = shape
+        self.dtype = self.flat_groups[0][0].dtype
+
+    def contiguous(self):
+        """
+        Merge partitioned weights from flat_groups into a single tensor.
+        """
+        end_idx = self.offset + self.partitioned_numel
+        world_size = len(self.flat_groups)
+        pad_flat_param_chunks = []
+
+        for rank_i in range(world_size):
+            # for each rank, we need to collect weights from related group/groups
+            flat_groups_at_rank_i = self.flat_groups[rank_i]
+            start_group_id = None
+            end_group_id = None
+            for group_id in range(len(self.flat_groups_offset)):
+                if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
+                    start_group_id = group_id
+                if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
+                    end_group_id = group_id
+                    break
+            # collect weights from related group/groups
+            for group_id in range(start_group_id, end_group_id + 1):
+                flat_tensor = flat_groups_at_rank_i[group_id]
+                start_offset = self.offset - self.flat_groups_offset[group_id]
+                end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
+                pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
+
+        # collect weights from all ranks
+        pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
+        param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
+        return param
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size
+
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
+    for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # memory efficient tensor
+        tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
+        state_dict[name] = tensor
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def to_torch_tensor(state_dict, return_empty_tensor=False):
+    """
+    Convert state_dict of GatheredTensor to torch tensor
+    """
+    torch_state_dict = {}
+    converted_tensors = {}
+    for name, tensor in state_dict.items():
+        tensor_id = id(tensor)
+        if tensor_id in converted_tensors:  # shared tensors
+            shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
+            torch_state_dict[name] = shared_tensor
+        else:
+            converted_tensors[tensor_id] = name
+            if return_empty_tensor:
+                torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
+            else:
+                torch_state_dict[name] = tensor.contiguous()
+    return torch_state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                             tag=None,
+                                             exclude_frozen_parameters=False,
+                                             lazy_mode=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+        - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
+          Convert the pesduo tensor to torch tensor by ``.contiguous()``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
+    You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint. Or you can load state_dict in lazy mode ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
+        for name, lazy_tensor in state_dict.item():
+            tensor = lazy_tensor.contiguous()  # to cpu
+            print(name, tensor)
+            # del tensor to release memory if it no longer in use
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+    if lazy_mode:
+        return state_dict
+    else:
+        return to_torch_tensor(state_dict)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
+                                               output_dir,
+                                               max_shard_size="5GB",
+                                               safe_serialization=False,
+                                               tag=None,
+                                               exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_dir``: directory to the pytorch fp32 state_dict output files
+        - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
+        - ``safe_serialization``:  whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+
+    # Dependency pre-check
+    if safe_serialization:
+        try:
+            from safetensors.torch import save_file
+        except ImportError:
+            print('If you want to use `safe_serialization`, please `pip install safetensors`')
+            raise
+    if max_shard_size is not None:
+        try:
+            from huggingface_hub import split_torch_state_dict_into_shards
+        except ImportError:
+            print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
+            raise
+
+    # Convert zero checkpoint to state_dict
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                                          tag,
+                                                          exclude_frozen_parameters,
+                                                          lazy_mode=True)
+
+    # Shard the model if it is too big.
+    weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
+    if max_shard_size is not None:
+        filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
+        # an memory-efficient approach for sharding
+        empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
+        state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
+                                                              filename_pattern=filename_pattern,
+                                                              max_shard_size=max_shard_size)
+    else:
+        from collections import namedtuple
+        StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
+        state_dict_split = StateDictSplit(is_sharded=False,
+                                          filename_to_tensors={weights_name: list(state_dict.keys())})
+
+    # Save the model by shard
+    os.makedirs(output_dir, exist_ok=True)
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+    for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
+        shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
+        shard_state_dict = to_torch_tensor(shard_state_dict)
+        output_path = os.path.join(output_dir, shard_file)
+        if safe_serialization:
+            save_file(shard_state_dict, output_path, metadata={"format": "pt"})
+        else:
+            torch.save(shard_state_dict, output_path)
+        # release the memory of current shard
+        for tensor_name in list(shard_state_dict.keys()):
+            del state_dict[tensor_name]
+            del shard_state_dict[tensor_name]
+        del shard_state_dict
+        gc.collect()
+
+    # Save index if sharded
+    if state_dict_split.is_sharded:
+        index = {
+            "metadata": state_dict_split.metadata,
+            "weight_map": state_dict_split.tensor_to_filename,
+        }
+        save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
+        save_index_file = os.path.join(output_dir, save_index_file)
+        with open(save_index_file, "w", encoding="utf-8") as f:
+            content = json.dumps(index, indent=2, sort_keys=True) + "\n"
+            f.write(content)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument("output_dir",
+                        type=str,
+                        help="directory to the pytorch fp32 state_dict output files"
+                        "(e.g. path/checkpoint-12-output/)")
+    parser.add_argument(
+        "--max_shard_size",
+        type=str,
+        default="5GB",
+        help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
+        "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
+        "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
+        "without CPU OOM issues.")
+    parser.add_argument(
+        "--safe_serialization",
+        default=False,
+        action='store_true',
+        help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+                                               args.output_dir,
+                                               max_shard_size=args.max_shard_size,
+                                               safe_serialization=args.safe_serialization,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)