Duplicate from OEvortex/HelpingAI-180B-base

Co-authored-by: HelpingAI <Abhaykoul@users.noreply.huggingface.co>

.gitattributes

@@ -0,0 +1,37 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text
+tokenizer[[:space:]](1).json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,26 @@
+---
+license: mit
+language:
+- en
+pipeline_tag: text-generation
+tags:
+- gemma
+---
+
+# HelpingAI-180B-base
+
+## Description
+The HelpingAI-180B-base model is a large-scale artificial intelligence model developed to assist in various natural language processing tasks. Trained on a diverse range of data sources, this model is designed to generate text, facilitate language understanding, and support various downstream tasks.
+
+## Model Information
+- **Model size**: 180 billion parameters
+- **Training data**: Diverse datasets covering a wide range of topics and domains.
+- **Training objective**: Language modeling with an emphasis on understanding and generating human-like text.
+- **Tokenizer**: Gemma tokenizer
+## Intended Use
+The HelpingAI-180B-base model is intended for researchers, developers, and practitioners in the field of natural language processing (NLP). It can be used for a variety of tasks, including but not limited to:
+- Text generation
+- Language understanding
+- Text summarization
+- Dialogue generation
+This model for research

config.json

@@ -0,0 +1,28 @@
+{
+  "architectures": [
+    "HelpingAIForCausalLM"
+  ],
+  "auto_map": {
+    "AutoConfig": "configuration_HelpingAI.HelpingAIConfig",
+    "AutoModelForCausalLM": "modeling_HelpingAI.HelpingAIForCausalLM"
+  },
+  "attention_dropout": 0.0,
+  "attention_softmax_in_fp32": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_dropout": 0.0,
+  "hidden_size": 14336,
+  "initializer_range": 0.02,
+  "layer_norm_epsilon": 1e-05,
+  "masked_softmax_fusion": true,
+  "model_type": "HelpingAI",
+  "n_head": 112,
+  "n_layer": 70,
+  "pad_token_id": 3,
+  "pretraining_tp": 4,
+  "slow_but_exact": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.34.0",
+  "use_cache": true,
+  "vocab_size": 250880
+}

configuration_HelpingAI.py

@@ -0,0 +1,60 @@
+""" HelpingAI model configuration"""
+
+from transformers import PretrainedConfig
+from transformers.utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+class HelpingAIConfig(PretrainedConfig):
+    keys_to_ignore_at_inference = ["past_key_values"]
+    model_type = "HelpingAI"
+    def __init__(
+        self,
+        vocab_size=50304,
+        hidden_size=2560,
+        intermediate_size=6912,
+        num_hidden_layers=32,
+        num_attention_heads=32,
+        num_key_value_heads=32,
+        head_dim=256,
+        hidden_act="silu",
+        max_position_embeddings=4096,
+        initializer_range=0.02,
+        rms_norm_eps=1e-6,
+        use_cache=True,
+        hidden_activation=None,
+        rope_theta=10000,
+        rope_pct=0.25,
+        attention_bias=False,
+        attention_dropout=0.0,
+        num_experts_per_tok=2,
+        num_local_experts=8,
+        router_aux_loss_coef=0.02,
+        output_router_logits=False,
+        norm_eps=1.0e-5,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.max_position_embeddings = max_position_embeddings        
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.head_dim = head_dim
+        self.hidden_act = hidden_act
+        self.hidden_activation = hidden_activation
+        self.num_key_value_heads = num_key_value_heads
+        self.initializer_range = initializer_range
+        self.rms_norm_eps = rms_norm_eps
+        self.use_cache = use_cache
+        self.rope_theta = rope_theta
+        self.attention_bias = attention_bias 
+        self.attention_dropout = attention_dropout
+        self.num_experts_per_tok = num_experts_per_tok
+        self.num_local_experts = num_local_experts
+        self.router_aux_loss_coef = router_aux_loss_coef
+        self.output_router_logits = output_router_logits
+        self.rope_pct = rope_pct
+        self.norm_eps = norm_eps
+        super().__init__(**kwargs)

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 0,
+  "do_sample": true,
+  "transformers_version": "4.40.0.dev0"
+}

modeling_HelpingAI.py

@@ -0,0 +1,670 @@
+""" HelpingAI model . """
+from typing import Optional, Tuple, Union
+import math
+
+import torch
+import torch.utils.checkpoint
+from transformers import AutoModel, AutoModelForCausalLM
+from torch import nn
+from torch.nn import CrossEntropyLoss
+from transformers.modeling_outputs import (
+    BaseModelOutputWithPast,
+    CausalLMOutputWithPast,
+)
+from transformers.modeling_utils import PreTrainedModel
+from transformers.utils import logging
+from .configuration_HelpingAI import HelpingAIConfig
+
+
+logger = logging.get_logger(__name__)
+
+
+# Copied from transformers.models.bart.modeling_bart._make_causal_mask
+def _make_causal_mask(
+    input_ids_shape: torch.Size,
+    dtype: torch.dtype,
+    device: torch.device,
+    past_key_values_length: int = 0,
+):
+    """Make causal mask used for bi-directional self-attention."""
+    batch_size, tgt_len = input_ids_shape
+    mask = torch.full((tgt_len, tgt_len), torch.finfo(torch.float16).min, device=device)
+    mask_cond = torch.arange(mask.size(-1), device=device)
+    mask.masked_fill_(mask_cond < (mask_cond + 1).view(mask.size(-1), 1), 0)
+    mask = mask.to(dtype)
+    if past_key_values_length > 0:
+        mask = torch.cat([torch.zeros(tgt_len, past_key_values_length, dtype=dtype, device=device), mask], dim=-1)
+    return mask[None, None, :, :].expand(batch_size, 1, tgt_len, tgt_len + past_key_values_length)
+
+
+# Copied from transformers.models.bart.modeling_bart._expand_mask
+def _expand_mask(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int] = None):
+    """Expands attention_mask from `[batch_size, seq_len]` to `[batch_size, 1, tgt_seq_len, src_seq_len]`."""
+    batch_size, src_len = mask.size()
+    tgt_len = tgt_len if tgt_len is not None else src_len
+
+    expanded_mask = mask[:, None, None, :].expand(batch_size, 1, tgt_len, src_len).to(dtype)
+    inverted_mask = 1.0 - expanded_mask
+
+    return inverted_mask.masked_fill(
+        inverted_mask.to(torch.bool), torch.finfo(dtype).min
+    )
+
+
+class RotaryEmbedding(nn.Module):
+    def __init__(
+        self,
+        dim: int,
+        max_position_embeddings: int,
+        base: int = 10_000,
+        device: Optional[torch.device] = None,
+    ):
+        super().__init__()
+
+        self.dim = dim
+        self.max_position_embeddings = max_position_embeddings
+        self.base = base
+        inv_freq = 1.0 / (self.base ** (torch.arange(0, self.dim, 2, device=device, dtype=torch.float32) / self.dim))
+        self.register_buffer("inv_freq", inv_freq, persistent=False)
+
+        # Build here to make `torch.jit.trace` work.
+        self._set_cos_sin_cache(
+            seq_len=max_position_embeddings, device=self.inv_freq.device, dtype=torch.get_default_dtype(),
+        )
+
+    def _set_cos_sin_cache(self, seq_len: int, device: torch.device, dtype: torch.dtype):
+        self.max_seq_len_cached = seq_len
+        t = torch.arange(self.max_seq_len_cached, device=device, dtype=torch.float32)
+
+        # Don't do einsum, it converts fp32 to fp16 under AMP
+        # freqs = torch.einsum("i,j->ij", t, self.inv_freq)
+        freqs = torch.outer(t, self.inv_freq)
+        # Different from paper, but it uses a different permutation in order to obtain the same calculation
+        emb = torch.cat((freqs, freqs), dim=-1)
+        self.register_buffer("cos_cached", emb.cos()[None, None, :, :].to(dtype), persistent=False)
+        self.register_buffer("sin_cached", emb.sin()[None, None, :, :].to(dtype), persistent=False)
+
+    def forward(self, x: torch.Tensor, seq_len: Optional[int] = None):
+        # x: [batch_size, num_heads, seq_len, head_size]
+        if seq_len > self.max_seq_len_cached:
+            self._set_cos_sin_cache(seq_len=seq_len, device=x.device, dtype=torch.get_default_dtype())
+        return (
+            self.cos_cached[:, :, :seq_len, ...].to(dtype=x.dtype),
+            self.sin_cached[:, :, :seq_len, ...].to(dtype=x.dtype),
+        )
+
+
+def rotate_half(x: torch.Tensor):
+    """Rotates half the hidden dims of the input."""
+    x1, x2 = torch.chunk(x, 2, dim=-1)
+    return torch.cat((-x2, x1), dim=-1)
+
+
+def apply_rotary_pos_emb(q, k, cos, sin, position_ids):
+    # The first two dimensions of cos and sin are always 1, so we can `squeeze` them.
+    cos = cos.squeeze(1).squeeze(0)  # [seq_len, dim]
+    sin = sin.squeeze(1).squeeze(0)  # [seq_len, dim]
+    cos = cos[position_ids].unsqueeze(1)  # [batch_size, 1, seq_len, dim]
+    sin = sin[position_ids].unsqueeze(1)  # [batch_size, 1, seq_len, dim]
+    q_embed = (q * cos) + (rotate_half(q) * sin)
+    k_embed = (k * cos) + (rotate_half(k) * sin)
+    return q_embed, k_embed
+
+
+class MLP(nn.Module):
+    def __init__(self, config: HelpingAIConfig):
+        super().__init__()
+        self.config = config
+        self.hidden_size = config.hidden_size
+        self.intermediate_size = config.intermediate_size
+        self.gate_proj = nn.Linear(config.hidden_size, config.intermediate_size, bias=False)
+        self.up_proj = nn.Linear(config.hidden_size, config.intermediate_size, bias=False)
+        self.down_proj = nn.Linear(config.intermediate_size, config.hidden_size, bias=False)
+        self.act_fn = nn.SiLU()
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        return self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))
+
+
+def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
+    """
+    This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch,
+    num_key_value_heads, seqlen, head_dim) to (batch, num_attention_heads, seqlen, head_dim)
+    """
+    batch, num_key_value_heads, slen, head_dim = hidden_states.shape
+    if n_rep == 1:
+        return hidden_states
+    hidden_states = hidden_states[:, :, None, :, :].expand(batch, num_key_value_heads, n_rep, slen, head_dim)
+    return hidden_states.reshape(batch, num_key_value_heads * n_rep, slen, head_dim)
+
+
+class Attention(nn.Module):
+    def __init__(self, config: HelpingAIConfig):
+        super().__init__()
+        self.config = config
+        self.hidden_size = config.hidden_size
+        self.num_heads = config.num_attention_heads
+        self.head_dim = self.hidden_size // self.num_heads
+        self.num_key_value_heads = config.num_key_value_heads
+        self.num_key_value_groups = self.num_heads // self.num_key_value_heads
+        self.max_position_embeddings = config.max_position_embeddings
+
+        if (self.head_dim * self.num_heads) != self.hidden_size:
+            raise ValueError(
+                f"hidden_size must be divisible by num_heads (got `hidden_size`: {self.hidden_size}"
+                f" and `num_heads`: {self.num_heads})."
+            )
+        self.q_proj = nn.Linear(self.hidden_size, self.num_heads * self.head_dim, bias=False)
+        self.k_proj = nn.Linear(self.hidden_size, self.num_key_value_heads * self.head_dim, bias=False)
+        self.v_proj = nn.Linear(self.hidden_size, self.num_key_value_heads * self.head_dim, bias=False)
+        self.o_proj = nn.Linear(self.hidden_size, self.hidden_size, bias=False)
+
+        self._init_rope()
+
+    def _init_rope(self):
+        self.rotary_ndims = int(self.head_dim * self.config.rope_pct)
+        self.rotary_emb = RotaryEmbedding(
+            self.rotary_ndims,
+            max_position_embeddings=self.config.max_position_embeddings,
+            base=self.config.rope_theta,
+        )
+
+    def forward(
+        self,
+        hidden_states: torch.FloatTensor,
+        attention_mask: torch.FloatTensor,
+        position_ids: torch.LongTensor,
+        past_key_value: Optional[Tuple[torch.Tensor]] = None,
+        output_attentions: Optional[bool] = False,
+        use_cache: Optional[bool] = False,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states)
+        value_states = self.v_proj(hidden_states)
+
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        query_rot = query_states[..., : self.rotary_ndims]
+        query_pass = query_states[..., self.rotary_ndims :]
+        key_rot = key_states[..., : self.rotary_ndims]
+        key_pass = key_states[..., self.rotary_ndims :]
+
+        kv_seq_len = key_states.shape[-2]
+        if past_key_value is not None:
+            kv_seq_len += past_key_value[0].shape[-2]
+        cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
+        query_states, key_states = apply_rotary_pos_emb(query_rot, key_rot, cos, sin, position_ids)
+
+        # [batch_size, num_heads, seq_len, head_dim]
+        query_states = torch.cat((query_states, query_pass), dim=-1)
+        key_states = torch.cat((key_states, key_pass), dim=-1)
+
+        if past_key_value is not None:
+            # Reuse k, v, self_attention
+            key_states = torch.cat((past_key_value[0], key_states), dim=2)
+            value_states = torch.cat((past_key_value[1], value_states), dim=2)
+
+        past_key_value = (key_states, value_states) if use_cache else None
+
+        # Repeat k/v heads if n_kv_heads < n_heads
+        key_states = repeat_kv(key_states, self.num_key_value_groups)
+        value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+        attn_weights = torch.matmul(query_states, key_states.transpose(2, 3)) / math.sqrt(self.head_dim)
+
+        if attn_weights.size() != (bsz, self.num_heads, q_len, kv_seq_len):
+            raise ValueError(
+                f"Attention weights should be of size {(bsz, self.num_heads, q_len, kv_seq_len)}, but is"
+                f" {attn_weights.size()}"
+            )
+
+        if attention_mask is not None:
+            if attention_mask.size() != (bsz, 1, q_len, kv_seq_len):
+                raise ValueError(
+                    f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, but is {attention_mask.size()}"
+                )
+            attn_weights = attn_weights + attention_mask
+
+        # Upcast attention to fp32
+        attn_weights = nn.functional.softmax(attn_weights, dim=-1, dtype=torch.float32).to(query_states.dtype)
+        attn_output = torch.matmul(attn_weights, value_states)
+
+        if attn_output.size() != (bsz, self.num_heads, q_len, self.head_dim):
+            raise ValueError(
+                f"`attn_output` should be of size {(bsz, self.num_heads, q_len, self.head_dim)}, but is"
+                f" {attn_output.size()}"
+            )
+
+        # Merge heads
+        attn_output = attn_output.transpose(1, 2).contiguous()
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
+
+        # Final linear projection
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+
+
+class DecoderLayer(nn.Module):
+    def __init__(self, config: HelpingAIConfig):
+        super().__init__()
+        self.self_attn = Attention(config)
+        self.mlp = MLP(config)
+        self.input_layernorm = nn.LayerNorm(config.hidden_size, eps=config.norm_eps)
+        self.post_attention_layernorm = nn.LayerNorm(config.hidden_size, eps=config.norm_eps)
+
+    def forward(
+        self,
+        hidden_states: Optional[torch.FloatTensor],
+        attention_mask: Optional[torch.FloatTensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Tuple[torch.Tensor]] = None,
+        output_attentions: Optional[bool] = False,
+        use_cache: Optional[bool] = False,
+    ) -> Union[Tuple[torch.Tensor], Optional[Tuple[torch.Tensor, Tuple[torch.FloatTensor, ...]]]]:
+        residual = hidden_states
+
+        hidden_states = self.input_layernorm(hidden_states)
+
+        # Self Attention
+        hidden_states, self_attn_weights, present_key_value = self.self_attn(
+            hidden_states=hidden_states,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_value=past_key_value,
+            output_attentions=output_attentions,
+            use_cache=use_cache,
+        )
+        hidden_states = residual + hidden_states
+
+        # Fully Connected
+        residual = hidden_states
+        hidden_states = self.post_attention_layernorm(hidden_states)
+        hidden_states = self.mlp(hidden_states)
+        hidden_states = residual + hidden_states
+
+        outputs = (hidden_states,)
+
+        if output_attentions:
+            outputs += (self_attn_weights,)
+
+        if use_cache:
+            outputs += (present_key_value,)
+
+        return outputs
+
+
+class HelpingAIPreTrainedModel(PreTrainedModel):
+    """An abstract class to handle weights initialization and a simple interface
+    for downloading and loading pretrained models.
+    """
+
+    config_class = HelpingAIConfig
+    base_model_prefix = "transformer"
+    supports_gradient_checkpointing = True
+    _no_split_modules = ["DecoderLayer"]
+    _skip_keys_device_placement = "past_key_values"
+
+    def _init_weights(self, module: nn.Module):
+        """Initialize the weights"""
+        if isinstance(module, nn.Linear):
+            module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
+            if module.bias is not None:
+                module.bias.data.zero_()
+        elif isinstance(module, nn.Embedding):
+            module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
+            if module.padding_idx is not None:
+                module.weight.data[module.padding_idx].zero_()
+        elif isinstance(module, nn.LayerNorm):
+            module.bias.data.zero_()
+            module.weight.data.fill_(1.0)
+
+    def _set_gradient_checkpointing(self, module: nn.Module, value=False):
+        if isinstance(module, HelpingAIModel):
+            module.gradient_checkpointing = value
+
+
+class HelpingAIModel(HelpingAIPreTrainedModel):
+    def __init__(self, config: HelpingAIConfig):
+        super().__init__(config)
+        self.embed_tokens = nn.Embedding(config.vocab_size, config.hidden_size, config.pad_token_id)
+        self.layers = nn.ModuleList([DecoderLayer(config) for _ in range(config.num_hidden_layers)])
+        self.norm = nn.LayerNorm(config.hidden_size, eps=config.norm_eps)
+
+        self.gradient_checkpointing = False
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.embed_tokens
+
+    def set_input_embeddings(self, value: nn.Module):
+        self.embed_tokens = value
+
+    # Copied from transformers.models.bart.modeling_bart.BartDecoder._prepare_decoder_attention_mask
+    def _prepare_decoder_attention_mask(
+        self,
+        attention_mask: torch.Tensor,
+        input_shape: torch.Size,
+        inputs_embeds: torch.Tensor,
+        past_key_values_length: int,
+    ):
+        # Create causal mask
+        # [batch_size, seq_len] -> [batch_size, 1, tgt_seq_len, src_seq_len]
+        combined_attention_mask = None
+        if input_shape[-1] > 1:
+            combined_attention_mask = _make_causal_mask(
+                input_shape,
+                inputs_embeds.dtype,
+                device=inputs_embeds.device,
+                past_key_values_length=past_key_values_length,
+            )
+
+        if attention_mask is not None:
+            # [batch_size, seq_len] -> [batch_size, 1, tgt_seq_len, src_seq_len]
+            expanded_attn_mask = _expand_mask(
+                attention_mask, inputs_embeds.dtype, tgt_len=input_shape[-1]
+            ).to(inputs_embeds.device)
+            combined_attention_mask = expanded_attn_mask if combined_attention_mask is None else expanded_attn_mask + combined_attention_mask
+
+        return combined_attention_mask
+
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Tuple[Tuple[torch.FloatTensor]]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, BaseModelOutputWithPast]:
+        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_hidden_states = output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        # Retrieve input_ids and inputs_embeds
+        if input_ids is not None and inputs_embeds is not None:
+            raise ValueError(
+                "You cannot specify both decoder_input_ids and decoder_inputs_embeds at the same time"
+            )
+        elif input_ids is not None:
+            batch_size, seq_length = input_ids.shape
+        elif inputs_embeds is not None:
+            batch_size, seq_length, _ = inputs_embeds.shape
+        else:
+            raise ValueError(
+                "You have to specify either decoder_input_ids or decoder_inputs_embeds"
+            )
+
+        seq_length_with_past = seq_length
+        past_key_values_length = 0
+
+        if past_key_values is not None:
+            past_key_values_length = past_key_values[0][0].shape[2]
+            seq_length_with_past = seq_length_with_past + past_key_values_length
+
+        if position_ids is None:
+            device = input_ids.device if input_ids is not None else inputs_embeds.device
+            position_ids = torch.arange(
+                past_key_values_length,
+                seq_length + past_key_values_length,
+                dtype=torch.long,
+                device=device,
+            )
+            position_ids = position_ids.unsqueeze(0).view(-1, seq_length)
+        else:
+            position_ids = position_ids.view(-1, seq_length).long()
+
+        if inputs_embeds is None:
+            inputs_embeds = self.embed_tokens(input_ids)
+        # Embed positions
+        if attention_mask is None:
+            attention_mask = torch.ones(
+                (batch_size, seq_length_with_past),
+                dtype=torch.bool,
+                device=inputs_embeds.device,
+            )
+        attention_mask = self._prepare_decoder_attention_mask(
+            attention_mask,
+            (batch_size, seq_length),
+            inputs_embeds,
+            past_key_values_length,
+        )
+
+        hidden_states = inputs_embeds
+
+        if self.gradient_checkpointing and self.training:
+            if use_cache:
+                logger.warning(
+                    "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
+                )
+                use_cache = False
+
+        # Decoder layers
+        all_hidden_states = () if output_hidden_states else None
+        all_self_attns = () if output_attentions else None
+        next_decoder_cache = () if use_cache else None
+
+        for idx, decoder_layer in enumerate(self.layers):
+            if output_hidden_states:
+                all_hidden_states += (hidden_states,)
+
+            past_key_value = (
+                past_key_values[idx] if past_key_values is not None else None
+            )
+
+            if self.gradient_checkpointing and self.training:
+
+                def create_custom_forward(module):
+                    def custom_forward(*inputs):
+                        # None for past_key_value
+                        return module(*inputs, past_key_value, output_attentions)
+
+                    return custom_forward
+
+                layer_outputs = torch.utils.checkpoint.checkpoint(
+                    create_custom_forward(decoder_layer),
+                    hidden_states,
+                    attention_mask,
+                    position_ids,
+                )
+            else:
+                layer_outputs = decoder_layer(
+                    hidden_states,
+                    attention_mask=attention_mask,
+                    position_ids=position_ids,
+                    past_key_value=past_key_value,
+                    output_attentions=output_attentions,
+                    use_cache=use_cache,
+                )
+
+            hidden_states = layer_outputs[0]
+
+            if use_cache:
+                next_decoder_cache += (layer_outputs[2 if output_attentions else 1],)
+
+            if output_attentions:
+                all_self_attns += (layer_outputs[1],)
+
+        hidden_states = self.norm(hidden_states)
+
+        # Add hidden states from the last decoder layer
+        if output_hidden_states:
+            all_hidden_states += (hidden_states,)
+
+        next_cache = next_decoder_cache if use_cache else None
+        if not return_dict:
+            return tuple(
+                v
+                for v in [hidden_states, next_cache, all_hidden_states, all_self_attns]
+                if v is not None
+            )
+        return BaseModelOutputWithPast(
+            last_hidden_state=hidden_states,
+            past_key_values=next_cache,
+            hidden_states=all_hidden_states,
+            attentions=all_self_attns,
+        )
+
+
+class HelpingAIForCausalLM(HelpingAIPreTrainedModel):
+    _tied_weights_keys = ["lm_head.weight"]
+
+    def __init__(self, config: HelpingAIConfig):
+        super().__init__(config)
+
+        self.model = HelpingAIModel(config)
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.model.embed_tokens
+
+    def set_input_embeddings(self, value):
+        self.model.embed_tokens = value
+
+    def get_output_embeddings(self):
+        return self.lm_head
+
+    def set_output_embeddings(self, new_embeddings: nn.Module):
+        self.lm_head = new_embeddings
+
+    def get_decoder(self):
+        return self.transformer
+
+    def set_decoder(self, decoder):
+        self.transformer = decoder
+
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Tuple[Tuple[torch.FloatTensor]]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, CausalLMOutputWithPast]:
+        output_attentions = (
+            output_attentions
+            if output_attentions is not None
+            else self.config.output_attentions
+        )
+        output_hidden_states = (
+            output_hidden_states
+            if output_hidden_states is not None
+            else self.config.output_hidden_states
+        )
+        return_dict = (
+            return_dict if return_dict is not None else self.config.use_return_dict
+        )
+
+        # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
+        outputs = self.model(
+            input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = outputs[0]
+        logits = self.lm_head(hidden_states).float()
+
+        loss = None
+        if labels is not None:
+            # Shift so that tokens < n predict n
+            shift_logits = logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss()
+            shift_logits = shift_logits.view(-1, self.config.vocab_size)
+            shift_labels = shift_labels.view(-1)
+            # Enable model parallelism
+            shift_labels = shift_labels.to(shift_logits.device)
+            loss = loss_fct(shift_logits, shift_labels)
+
+        if not return_dict:
+            output = (logits,) + outputs[1:]
+            return (loss,) + output if loss is not None else output
+
+        return CausalLMOutputWithPast(
+            loss=loss,
+            logits=logits,
+            past_key_values=outputs.past_key_values,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+    def prepare_inputs_for_generation(
+        self,
+        input_ids,
+        past_key_values: Optional[torch.Tensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        inputs_embeds: Optional[torch.Tensor] = None,
+        **kwargs,
+    ):
+        # Trim decoder_input_ids if past is used
+        if past_key_values and past_key_values[0] is not None:
+            input_ids = input_ids[:, -1:]
+
+        position_ids = kwargs.get("position_ids", None)
+        if attention_mask is not None and position_ids is None:
+            # Create position_ids on the fly for batch generation
+            position_ids = attention_mask.long().cumsum(-1) - 1
+            position_ids.masked_fill_(attention_mask == 0, 1)
+            if past_key_values:
+                position_ids = position_ids[:, -1].unsqueeze(-1)
+
+        # If `inputs_embeds` are passed, we only want to use them in the 1st generation step
+        if inputs_embeds is not None and past_key_values is None:
+            model_inputs = {"inputs_embeds": inputs_embeds}
+        else:
+            model_inputs = {"input_ids": input_ids}
+
+        model_inputs.update(
+            {
+                "attention_mask": attention_mask,
+                "past_key_values": past_key_values,
+                "use_cache": kwargs.get("use_cache"),
+                "position_ids": position_ids,
+            }
+        )
+        return model_inputs
+
+    @staticmethod
+    def _reorder_cache(past_key_values, beam_idx):
+        reordered_past = ()
+        for layer_past in past_key_values:
+            reordered_past += (
+                tuple(
+                    past_state.index_select(0, beam_idx.to(past_state.device))
+                    for past_state in layer_past
+                ),
+            )
+        return reordered_past
+
+
+HelpingAIConfig.register_for_auto_class()
+HelpingAIForCausalLM.register_for_auto_class("AutoModelForCausalLM")

pytorch_model-00001-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca3187da2ba70772c77cab754eafd535b0e4a530c5c43ec3449257461ecd30d3
+size 8837690076

pytorch_model-00002-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ebc9777c81fbd62d2895775cf659f00f0cf6a66d3f65c5e00859ec28e64fa4f5
+size 9865757318

pytorch_model-00003-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b474d55ae804d828ede1dfd59b90b494985d260f7aafcab711165d0060863b84
+size 9865757318

pytorch_model-00004-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a472b3eb2e3a5c6a8d01594f3839ca802a641726b9761853aef72f943a0fdd54
+size 9865757318

pytorch_model-00005-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7fade6a15060368fbd037d9638945d71c853534f1ed87e48f6b0cce6663ea348
+size 9865757318

pytorch_model-00006-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3861b788114ac983fec33c8a8f789ce0aac41f8d0c5d69c110abd462c85ea4c
+size 9865757382

pytorch_model-00007-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56e4cd89630a2725e630b1e13e1db882203474bfcb92ace7620989c1d201104e
+size 9865757382

pytorch_model-00008-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba7375b56a1535f88c5f1f488e37212b708369e8b98fbac039497e1b7286e656
+size 9865757382

pytorch_model-00009-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7de2e69a9db40b2e2f2441da2cda4e3f90776d9e23850b7425625c491a255aad
+size 9865757382

pytorch_model-00010-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70dee8587d56760414f1f6b5e28de93db450d46e31ef75efede83c3ea438bab4
+size 9865757382

pytorch_model-00011-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:136b56084635f1c422cddb5185db34fb80866a84c7170a43cb40db631a230ac6
+size 9865757382

pytorch_model-00012-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38fe8957ccaee533e54f97aa95b33eee1c7c2de10424c76142d48ad800de9103
+size 9865757382

pytorch_model-00013-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e583bd86212872a58aba123e0cfd7f96d01186df1823e6e093ea4367c2d2cbd3
+size 9865757382

pytorch_model-00014-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7f6c50bb9710c879a791d59974fbc5ce2fc4f5eae1a96f12ca15973bcab7640
+size 9865757382

pytorch_model-00015-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6bc99c13110b14e03e38e7af312c7aa3d5655dad89afdc6db872aa144fb07c39
+size 9865757382

pytorch_model-00016-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d59dc96927dd17d1945eeaea31d97d5feaa6e3c3dd1b60c638d25a7dd0796bf
+size 9865757382

pytorch_model-00017-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:162d15ec31369b932d37a6cd04c85bc546bd350f3235eb096c445587a913e370
+size 9865757382

pytorch_model-00018-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:189949529ef7a5533fef059f86de583343e14e9ad8799d0ea67a51890b452cd7
+size 9865757382

pytorch_model-00019-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:351623dc9b9131e771f7e995a8b02261547a2a4249f8d8ced7de291f370e6d84
+size 9865757382

pytorch_model-00020-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:217f9ea1fc5ff5a7910ea9222ec70fc24a06b2ccfbfc69924c94555d40141986
+size 9865757382

pytorch_model-00021-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d46df3ba9538a5790d2c35264af6516ee9357ba9760216b2d8ec057000b1316
+size 9865757382

pytorch_model-00022-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a99aeb8cb6bfce134df868b787c4680439a21d9185add42ff45d067c624841ac
+size 9865757382

pytorch_model-00023-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a46f4b50de561c80d3a882c7f8ed498231b89c36919da417d0675fceaf303e62
+size 9865757382

pytorch_model-00024-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f65841392f00a806c3b4b0081befd1260f854cea1d60ab2d87939950b2b2e88
+size 9865757382

pytorch_model-00025-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fac5c1abeb05511cf496f47984a0efcce929678bc01352eae240e988cfe15e52
+size 9865757382

pytorch_model-00026-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8a952fb468ce32332307880ad7a54deec7119608ce2d02e3556f85626a1d111
+size 9865757382

pytorch_model-00027-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c331a3536db08191b2704dc78c01aeeb62207f2aaf4db0188d3fd4d4553ac96
+size 9865757382

pytorch_model-00028-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c189d575771c8c9e96030f91e3c532ca006c04fc573dd4ac26399f1305e515c0
+size 9865757382

pytorch_model-00029-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d9e5e6c448f8f6e4d8b76c08ccd43f00a28c819111b86b438578bd4793b809c
+size 9865757382

pytorch_model-00030-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:666234aee85def980de53cbe63d86922e4371d14e13ba380bd665926cf16260e
+size 9865757382

pytorch_model-00031-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d32da8fbe15f3a8dda294f63d4e1d210796e3a952fc0488c8ff4f2a09666235
+size 9865757382

pytorch_model-00032-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53371c3c0a1e97ed02db8e5a83ab76a038a386724c10522bd05c1ceff9b37908
+size 9865757382

pytorch_model-00033-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:647c442ff7fd4896fc2915f4e6f4bd85b12aa4663a5d5d6e1f56bf0b585c3434
+size 9865757382

pytorch_model-00034-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf885237c29d7c83babd2684c9ada2a6227a025e56c3ab6b6d4e5583c47927ee
+size 9865757382

pytorch_model-00035-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddb922a126dc39bfe52277d189b8dcb21e56ff50d557e7320b07897972fdd599
+size 9865757382

pytorch_model-00036-of-00036.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b96d0d49263134b6ac99130ecef9500ba2eef11c89663362a0f59f39e3285838
+size 8221416146

pytorch_model.bin.index.json

@@ -0,0 +1,853 @@
+{
+  "metadata": {
+    "total_size": 352494542848
+  },
+  "weight_map": {
+    "lm_head.weight": "pytorch_model-00001-of-00036.bin",
+    "transformer.h.0.input_layernorm.bias": "pytorch_model-00001-of-00036.bin",
+    "transformer.h.0.input_layernorm.weight": "pytorch_model-00001-of-00036.bin",
+    "transformer.h.0.mlp.dense_4h_to_h.bias": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.0.mlp.dense_4h_to_h.weight": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.0.mlp.dense_h_to_4h.bias": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.0.mlp.dense_h_to_4h.weight": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.0.post_attention_layernorm.bias": "pytorch_model-00001-of-00036.bin",
+    "transformer.h.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00036.bin",
+    "transformer.h.0.self_attention.dense.bias": "pytorch_model-00001-of-00036.bin",
+    "transformer.h.0.self_attention.dense.weight": "pytorch_model-00001-of-00036.bin",
+    "transformer.h.0.self_attention.query_key_value.bias": "pytorch_model-00001-of-00036.bin",
+    "transformer.h.0.self_attention.query_key_value.weight": "pytorch_model-00001-of-00036.bin",
+    "transformer.h.1.input_layernorm.bias": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.1.input_layernorm.weight": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.1.mlp.dense_4h_to_h.bias": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.1.mlp.dense_4h_to_h.weight": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.1.mlp.dense_h_to_4h.bias": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.1.mlp.dense_h_to_4h.weight": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.1.post_attention_layernorm.bias": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.1.post_attention_layernorm.weight": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.1.self_attention.dense.bias": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.1.self_attention.dense.weight": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.1.self_attention.query_key_value.bias": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.1.self_attention.query_key_value.weight": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.10.input_layernorm.bias": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.10.input_layernorm.weight": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.10.mlp.dense_4h_to_h.bias": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.10.mlp.dense_4h_to_h.weight": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.10.mlp.dense_h_to_4h.bias": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.10.mlp.dense_h_to_4h.weight": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.10.post_attention_layernorm.bias": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.10.post_attention_layernorm.weight": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.10.self_attention.dense.bias": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.10.self_attention.dense.weight": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.10.self_attention.query_key_value.bias": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.10.self_attention.query_key_value.weight": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.11.input_layernorm.bias": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.11.input_layernorm.weight": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.11.mlp.dense_4h_to_h.bias": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.11.mlp.dense_4h_to_h.weight": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.11.mlp.dense_h_to_4h.bias": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.11.mlp.dense_h_to_4h.weight": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.11.post_attention_layernorm.bias": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.11.post_attention_layernorm.weight": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.11.self_attention.dense.bias": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.11.self_attention.dense.weight": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.11.self_attention.query_key_value.bias": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.11.self_attention.query_key_value.weight": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.12.input_layernorm.bias": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.12.input_layernorm.weight": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.12.mlp.dense_4h_to_h.bias": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.12.mlp.dense_4h_to_h.weight": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.12.mlp.dense_h_to_4h.bias": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.12.mlp.dense_h_to_4h.weight": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.12.post_attention_layernorm.bias": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.12.post_attention_layernorm.weight": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.12.self_attention.dense.bias": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.12.self_attention.dense.weight": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.12.self_attention.query_key_value.bias": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.12.self_attention.query_key_value.weight": "pytorch_model-00007-of-00036.bin",
+    "transformer.h.13.input_layernorm.bias": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.13.input_layernorm.weight": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.13.mlp.dense_4h_to_h.bias": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.13.mlp.dense_4h_to_h.weight": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.13.mlp.dense_h_to_4h.bias": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.13.mlp.dense_h_to_4h.weight": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.13.post_attention_layernorm.bias": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.13.post_attention_layernorm.weight": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.13.self_attention.dense.bias": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.13.self_attention.dense.weight": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.13.self_attention.query_key_value.bias": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.13.self_attention.query_key_value.weight": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.14.input_layernorm.bias": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.14.input_layernorm.weight": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.14.mlp.dense_4h_to_h.bias": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.14.mlp.dense_4h_to_h.weight": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.14.mlp.dense_h_to_4h.bias": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.14.mlp.dense_h_to_4h.weight": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.14.post_attention_layernorm.bias": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.14.post_attention_layernorm.weight": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.14.self_attention.dense.bias": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.14.self_attention.dense.weight": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.14.self_attention.query_key_value.bias": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.14.self_attention.query_key_value.weight": "pytorch_model-00008-of-00036.bin",
+    "transformer.h.15.input_layernorm.bias": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.15.input_layernorm.weight": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.15.mlp.dense_4h_to_h.bias": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.15.mlp.dense_4h_to_h.weight": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.15.mlp.dense_h_to_4h.bias": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.15.mlp.dense_h_to_4h.weight": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.15.post_attention_layernorm.bias": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.15.post_attention_layernorm.weight": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.15.self_attention.dense.bias": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.15.self_attention.dense.weight": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.15.self_attention.query_key_value.bias": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.15.self_attention.query_key_value.weight": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.16.input_layernorm.bias": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.16.input_layernorm.weight": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.16.mlp.dense_4h_to_h.bias": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.16.mlp.dense_4h_to_h.weight": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.16.mlp.dense_h_to_4h.bias": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.16.mlp.dense_h_to_4h.weight": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.16.post_attention_layernorm.bias": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.16.post_attention_layernorm.weight": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.16.self_attention.dense.bias": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.16.self_attention.dense.weight": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.16.self_attention.query_key_value.bias": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.16.self_attention.query_key_value.weight": "pytorch_model-00009-of-00036.bin",
+    "transformer.h.17.input_layernorm.bias": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.17.input_layernorm.weight": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.17.mlp.dense_4h_to_h.bias": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.17.mlp.dense_4h_to_h.weight": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.17.mlp.dense_h_to_4h.bias": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.17.mlp.dense_h_to_4h.weight": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.17.post_attention_layernorm.bias": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.17.post_attention_layernorm.weight": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.17.self_attention.dense.bias": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.17.self_attention.dense.weight": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.17.self_attention.query_key_value.bias": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.17.self_attention.query_key_value.weight": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.18.input_layernorm.bias": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.18.input_layernorm.weight": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.18.mlp.dense_4h_to_h.bias": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.18.mlp.dense_4h_to_h.weight": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.18.mlp.dense_h_to_4h.bias": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.18.mlp.dense_h_to_4h.weight": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.18.post_attention_layernorm.bias": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.18.post_attention_layernorm.weight": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.18.self_attention.dense.bias": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.18.self_attention.dense.weight": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.18.self_attention.query_key_value.bias": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.18.self_attention.query_key_value.weight": "pytorch_model-00010-of-00036.bin",
+    "transformer.h.19.input_layernorm.bias": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.19.input_layernorm.weight": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.19.mlp.dense_4h_to_h.bias": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.19.mlp.dense_4h_to_h.weight": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.19.mlp.dense_h_to_4h.bias": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.19.mlp.dense_h_to_4h.weight": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.19.post_attention_layernorm.bias": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.19.post_attention_layernorm.weight": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.19.self_attention.dense.bias": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.19.self_attention.dense.weight": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.19.self_attention.query_key_value.bias": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.19.self_attention.query_key_value.weight": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.2.input_layernorm.bias": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.2.input_layernorm.weight": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.2.mlp.dense_4h_to_h.bias": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.2.mlp.dense_4h_to_h.weight": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.2.mlp.dense_h_to_4h.bias": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.2.mlp.dense_h_to_4h.weight": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.2.post_attention_layernorm.bias": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.2.post_attention_layernorm.weight": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.2.self_attention.dense.bias": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.2.self_attention.dense.weight": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.2.self_attention.query_key_value.bias": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.2.self_attention.query_key_value.weight": "pytorch_model-00002-of-00036.bin",
+    "transformer.h.20.input_layernorm.bias": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.20.input_layernorm.weight": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.20.mlp.dense_4h_to_h.bias": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.20.mlp.dense_4h_to_h.weight": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.20.mlp.dense_h_to_4h.bias": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.20.mlp.dense_h_to_4h.weight": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.20.post_attention_layernorm.bias": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.20.post_attention_layernorm.weight": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.20.self_attention.dense.bias": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.20.self_attention.dense.weight": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.20.self_attention.query_key_value.bias": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.20.self_attention.query_key_value.weight": "pytorch_model-00011-of-00036.bin",
+    "transformer.h.21.input_layernorm.bias": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.21.input_layernorm.weight": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.21.mlp.dense_4h_to_h.bias": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.21.mlp.dense_4h_to_h.weight": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.21.mlp.dense_h_to_4h.bias": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.21.mlp.dense_h_to_4h.weight": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.21.post_attention_layernorm.bias": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.21.post_attention_layernorm.weight": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.21.self_attention.dense.bias": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.21.self_attention.dense.weight": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.21.self_attention.query_key_value.bias": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.21.self_attention.query_key_value.weight": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.22.input_layernorm.bias": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.22.input_layernorm.weight": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.22.mlp.dense_4h_to_h.bias": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.22.mlp.dense_4h_to_h.weight": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.22.mlp.dense_h_to_4h.bias": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.22.mlp.dense_h_to_4h.weight": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.22.post_attention_layernorm.bias": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.22.post_attention_layernorm.weight": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.22.self_attention.dense.bias": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.22.self_attention.dense.weight": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.22.self_attention.query_key_value.bias": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.22.self_attention.query_key_value.weight": "pytorch_model-00012-of-00036.bin",
+    "transformer.h.23.input_layernorm.bias": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.23.input_layernorm.weight": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.23.mlp.dense_4h_to_h.bias": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.23.mlp.dense_4h_to_h.weight": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.23.mlp.dense_h_to_4h.bias": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.23.mlp.dense_h_to_4h.weight": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.23.post_attention_layernorm.bias": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.23.post_attention_layernorm.weight": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.23.self_attention.dense.bias": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.23.self_attention.dense.weight": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.23.self_attention.query_key_value.bias": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.23.self_attention.query_key_value.weight": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.24.input_layernorm.bias": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.24.input_layernorm.weight": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.24.mlp.dense_4h_to_h.bias": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.24.mlp.dense_4h_to_h.weight": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.24.mlp.dense_h_to_4h.bias": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.24.mlp.dense_h_to_4h.weight": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.24.post_attention_layernorm.bias": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.24.post_attention_layernorm.weight": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.24.self_attention.dense.bias": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.24.self_attention.dense.weight": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.24.self_attention.query_key_value.bias": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.24.self_attention.query_key_value.weight": "pytorch_model-00013-of-00036.bin",
+    "transformer.h.25.input_layernorm.bias": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.25.input_layernorm.weight": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.25.mlp.dense_4h_to_h.bias": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.25.mlp.dense_4h_to_h.weight": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.25.mlp.dense_h_to_4h.bias": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.25.mlp.dense_h_to_4h.weight": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.25.post_attention_layernorm.bias": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.25.post_attention_layernorm.weight": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.25.self_attention.dense.bias": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.25.self_attention.dense.weight": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.25.self_attention.query_key_value.bias": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.25.self_attention.query_key_value.weight": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.26.input_layernorm.bias": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.26.input_layernorm.weight": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.26.mlp.dense_4h_to_h.bias": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.26.mlp.dense_4h_to_h.weight": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.26.mlp.dense_h_to_4h.bias": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.26.mlp.dense_h_to_4h.weight": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.26.post_attention_layernorm.bias": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.26.post_attention_layernorm.weight": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.26.self_attention.dense.bias": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.26.self_attention.dense.weight": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.26.self_attention.query_key_value.bias": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.26.self_attention.query_key_value.weight": "pytorch_model-00014-of-00036.bin",
+    "transformer.h.27.input_layernorm.bias": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.27.input_layernorm.weight": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.27.mlp.dense_4h_to_h.bias": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.27.mlp.dense_4h_to_h.weight": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.27.mlp.dense_h_to_4h.bias": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.27.mlp.dense_h_to_4h.weight": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.27.post_attention_layernorm.bias": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.27.post_attention_layernorm.weight": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.27.self_attention.dense.bias": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.27.self_attention.dense.weight": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.27.self_attention.query_key_value.bias": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.27.self_attention.query_key_value.weight": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.28.input_layernorm.bias": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.28.input_layernorm.weight": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.28.mlp.dense_4h_to_h.bias": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.28.mlp.dense_4h_to_h.weight": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.28.mlp.dense_h_to_4h.bias": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.28.mlp.dense_h_to_4h.weight": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.28.post_attention_layernorm.bias": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.28.post_attention_layernorm.weight": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.28.self_attention.dense.bias": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.28.self_attention.dense.weight": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.28.self_attention.query_key_value.bias": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.28.self_attention.query_key_value.weight": "pytorch_model-00015-of-00036.bin",
+    "transformer.h.29.input_layernorm.bias": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.29.input_layernorm.weight": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.29.mlp.dense_4h_to_h.bias": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.29.mlp.dense_4h_to_h.weight": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.29.mlp.dense_h_to_4h.bias": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.29.mlp.dense_h_to_4h.weight": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.29.post_attention_layernorm.bias": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.29.post_attention_layernorm.weight": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.29.self_attention.dense.bias": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.29.self_attention.dense.weight": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.29.self_attention.query_key_value.bias": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.29.self_attention.query_key_value.weight": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.3.input_layernorm.bias": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.3.input_layernorm.weight": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.3.mlp.dense_4h_to_h.bias": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.3.mlp.dense_4h_to_h.weight": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.3.mlp.dense_h_to_4h.bias": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.3.mlp.dense_h_to_4h.weight": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.3.post_attention_layernorm.bias": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.3.post_attention_layernorm.weight": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.3.self_attention.dense.bias": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.3.self_attention.dense.weight": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.3.self_attention.query_key_value.bias": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.3.self_attention.query_key_value.weight": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.30.input_layernorm.bias": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.30.input_layernorm.weight": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.30.mlp.dense_4h_to_h.bias": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.30.mlp.dense_4h_to_h.weight": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.30.mlp.dense_h_to_4h.bias": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.30.mlp.dense_h_to_4h.weight": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.30.post_attention_layernorm.bias": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.30.post_attention_layernorm.weight": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.30.self_attention.dense.bias": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.30.self_attention.dense.weight": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.30.self_attention.query_key_value.bias": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.30.self_attention.query_key_value.weight": "pytorch_model-00016-of-00036.bin",
+    "transformer.h.31.input_layernorm.bias": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.31.input_layernorm.weight": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.31.mlp.dense_4h_to_h.bias": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.31.mlp.dense_4h_to_h.weight": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.31.mlp.dense_h_to_4h.bias": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.31.mlp.dense_h_to_4h.weight": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.31.post_attention_layernorm.bias": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.31.post_attention_layernorm.weight": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.31.self_attention.dense.bias": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.31.self_attention.dense.weight": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.31.self_attention.query_key_value.bias": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.31.self_attention.query_key_value.weight": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.32.input_layernorm.bias": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.32.input_layernorm.weight": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.32.mlp.dense_4h_to_h.bias": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.32.mlp.dense_4h_to_h.weight": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.32.mlp.dense_h_to_4h.bias": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.32.mlp.dense_h_to_4h.weight": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.32.post_attention_layernorm.bias": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.32.post_attention_layernorm.weight": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.32.self_attention.dense.bias": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.32.self_attention.dense.weight": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.32.self_attention.query_key_value.bias": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.32.self_attention.query_key_value.weight": "pytorch_model-00017-of-00036.bin",
+    "transformer.h.33.input_layernorm.bias": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.33.input_layernorm.weight": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.33.mlp.dense_4h_to_h.bias": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.33.mlp.dense_4h_to_h.weight": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.33.mlp.dense_h_to_4h.bias": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.33.mlp.dense_h_to_4h.weight": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.33.post_attention_layernorm.bias": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.33.post_attention_layernorm.weight": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.33.self_attention.dense.bias": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.33.self_attention.dense.weight": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.33.self_attention.query_key_value.bias": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.33.self_attention.query_key_value.weight": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.34.input_layernorm.bias": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.34.input_layernorm.weight": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.34.mlp.dense_4h_to_h.bias": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.34.mlp.dense_4h_to_h.weight": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.34.mlp.dense_h_to_4h.bias": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.34.mlp.dense_h_to_4h.weight": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.34.post_attention_layernorm.bias": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.34.post_attention_layernorm.weight": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.34.self_attention.dense.bias": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.34.self_attention.dense.weight": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.34.self_attention.query_key_value.bias": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.34.self_attention.query_key_value.weight": "pytorch_model-00018-of-00036.bin",
+    "transformer.h.35.input_layernorm.bias": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.35.input_layernorm.weight": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.35.mlp.dense_4h_to_h.bias": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.35.mlp.dense_4h_to_h.weight": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.35.mlp.dense_h_to_4h.bias": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.35.mlp.dense_h_to_4h.weight": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.35.post_attention_layernorm.bias": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.35.post_attention_layernorm.weight": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.35.self_attention.dense.bias": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.35.self_attention.dense.weight": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.35.self_attention.query_key_value.bias": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.35.self_attention.query_key_value.weight": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.36.input_layernorm.bias": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.36.input_layernorm.weight": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.36.mlp.dense_4h_to_h.bias": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.36.mlp.dense_4h_to_h.weight": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.36.mlp.dense_h_to_4h.bias": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.36.mlp.dense_h_to_4h.weight": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.36.post_attention_layernorm.bias": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.36.post_attention_layernorm.weight": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.36.self_attention.dense.bias": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.36.self_attention.dense.weight": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.36.self_attention.query_key_value.bias": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.36.self_attention.query_key_value.weight": "pytorch_model-00019-of-00036.bin",
+    "transformer.h.37.input_layernorm.bias": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.37.input_layernorm.weight": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.37.mlp.dense_4h_to_h.bias": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.37.mlp.dense_4h_to_h.weight": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.37.mlp.dense_h_to_4h.bias": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.37.mlp.dense_h_to_4h.weight": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.37.post_attention_layernorm.bias": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.37.post_attention_layernorm.weight": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.37.self_attention.dense.bias": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.37.self_attention.dense.weight": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.37.self_attention.query_key_value.bias": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.37.self_attention.query_key_value.weight": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.38.input_layernorm.bias": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.38.input_layernorm.weight": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.38.mlp.dense_4h_to_h.bias": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.38.mlp.dense_4h_to_h.weight": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.38.mlp.dense_h_to_4h.bias": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.38.mlp.dense_h_to_4h.weight": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.38.post_attention_layernorm.bias": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.38.post_attention_layernorm.weight": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.38.self_attention.dense.bias": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.38.self_attention.dense.weight": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.38.self_attention.query_key_value.bias": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.38.self_attention.query_key_value.weight": "pytorch_model-00020-of-00036.bin",
+    "transformer.h.39.input_layernorm.bias": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.39.input_layernorm.weight": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.39.mlp.dense_4h_to_h.bias": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.39.mlp.dense_4h_to_h.weight": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.39.mlp.dense_h_to_4h.bias": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.39.mlp.dense_h_to_4h.weight": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.39.post_attention_layernorm.bias": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.39.post_attention_layernorm.weight": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.39.self_attention.dense.bias": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.39.self_attention.dense.weight": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.39.self_attention.query_key_value.bias": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.39.self_attention.query_key_value.weight": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.4.input_layernorm.bias": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.4.input_layernorm.weight": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.4.mlp.dense_4h_to_h.bias": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.4.mlp.dense_4h_to_h.weight": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.4.mlp.dense_h_to_4h.bias": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.4.mlp.dense_h_to_4h.weight": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.4.post_attention_layernorm.bias": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.4.post_attention_layernorm.weight": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.4.self_attention.dense.bias": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.4.self_attention.dense.weight": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.4.self_attention.query_key_value.bias": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.4.self_attention.query_key_value.weight": "pytorch_model-00003-of-00036.bin",
+    "transformer.h.40.input_layernorm.bias": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.40.input_layernorm.weight": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.40.mlp.dense_4h_to_h.bias": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.40.mlp.dense_4h_to_h.weight": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.40.mlp.dense_h_to_4h.bias": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.40.mlp.dense_h_to_4h.weight": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.40.post_attention_layernorm.bias": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.40.post_attention_layernorm.weight": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.40.self_attention.dense.bias": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.40.self_attention.dense.weight": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.40.self_attention.query_key_value.bias": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.40.self_attention.query_key_value.weight": "pytorch_model-00021-of-00036.bin",
+    "transformer.h.41.input_layernorm.bias": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.41.input_layernorm.weight": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.41.mlp.dense_4h_to_h.bias": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.41.mlp.dense_4h_to_h.weight": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.41.mlp.dense_h_to_4h.bias": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.41.mlp.dense_h_to_4h.weight": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.41.post_attention_layernorm.bias": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.41.post_attention_layernorm.weight": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.41.self_attention.dense.bias": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.41.self_attention.dense.weight": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.41.self_attention.query_key_value.bias": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.41.self_attention.query_key_value.weight": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.42.input_layernorm.bias": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.42.input_layernorm.weight": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.42.mlp.dense_4h_to_h.bias": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.42.mlp.dense_4h_to_h.weight": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.42.mlp.dense_h_to_4h.bias": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.42.mlp.dense_h_to_4h.weight": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.42.post_attention_layernorm.bias": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.42.post_attention_layernorm.weight": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.42.self_attention.dense.bias": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.42.self_attention.dense.weight": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.42.self_attention.query_key_value.bias": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.42.self_attention.query_key_value.weight": "pytorch_model-00022-of-00036.bin",
+    "transformer.h.43.input_layernorm.bias": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.43.input_layernorm.weight": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.43.mlp.dense_4h_to_h.bias": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.43.mlp.dense_4h_to_h.weight": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.43.mlp.dense_h_to_4h.bias": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.43.mlp.dense_h_to_4h.weight": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.43.post_attention_layernorm.bias": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.43.post_attention_layernorm.weight": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.43.self_attention.dense.bias": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.43.self_attention.dense.weight": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.43.self_attention.query_key_value.bias": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.43.self_attention.query_key_value.weight": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.44.input_layernorm.bias": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.44.input_layernorm.weight": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.44.mlp.dense_4h_to_h.bias": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.44.mlp.dense_4h_to_h.weight": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.44.mlp.dense_h_to_4h.bias": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.44.mlp.dense_h_to_4h.weight": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.44.post_attention_layernorm.bias": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.44.post_attention_layernorm.weight": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.44.self_attention.dense.bias": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.44.self_attention.dense.weight": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.44.self_attention.query_key_value.bias": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.44.self_attention.query_key_value.weight": "pytorch_model-00023-of-00036.bin",
+    "transformer.h.45.input_layernorm.bias": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.45.input_layernorm.weight": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.45.mlp.dense_4h_to_h.bias": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.45.mlp.dense_4h_to_h.weight": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.45.mlp.dense_h_to_4h.bias": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.45.mlp.dense_h_to_4h.weight": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.45.post_attention_layernorm.bias": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.45.post_attention_layernorm.weight": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.45.self_attention.dense.bias": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.45.self_attention.dense.weight": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.45.self_attention.query_key_value.bias": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.45.self_attention.query_key_value.weight": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.46.input_layernorm.bias": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.46.input_layernorm.weight": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.46.mlp.dense_4h_to_h.bias": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.46.mlp.dense_4h_to_h.weight": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.46.mlp.dense_h_to_4h.bias": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.46.mlp.dense_h_to_4h.weight": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.46.post_attention_layernorm.bias": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.46.post_attention_layernorm.weight": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.46.self_attention.dense.bias": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.46.self_attention.dense.weight": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.46.self_attention.query_key_value.bias": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.46.self_attention.query_key_value.weight": "pytorch_model-00024-of-00036.bin",
+    "transformer.h.47.input_layernorm.bias": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.47.input_layernorm.weight": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.47.mlp.dense_4h_to_h.bias": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.47.mlp.dense_4h_to_h.weight": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.47.mlp.dense_h_to_4h.bias": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.47.mlp.dense_h_to_4h.weight": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.47.post_attention_layernorm.bias": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.47.post_attention_layernorm.weight": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.47.self_attention.dense.bias": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.47.self_attention.dense.weight": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.47.self_attention.query_key_value.bias": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.47.self_attention.query_key_value.weight": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.48.input_layernorm.bias": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.48.input_layernorm.weight": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.48.mlp.dense_4h_to_h.bias": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.48.mlp.dense_4h_to_h.weight": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.48.mlp.dense_h_to_4h.bias": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.48.mlp.dense_h_to_4h.weight": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.48.post_attention_layernorm.bias": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.48.post_attention_layernorm.weight": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.48.self_attention.dense.bias": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.48.self_attention.dense.weight": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.48.self_attention.query_key_value.bias": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.48.self_attention.query_key_value.weight": "pytorch_model-00025-of-00036.bin",
+    "transformer.h.49.input_layernorm.bias": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.49.input_layernorm.weight": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.49.mlp.dense_4h_to_h.bias": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.49.mlp.dense_4h_to_h.weight": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.49.mlp.dense_h_to_4h.bias": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.49.mlp.dense_h_to_4h.weight": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.49.post_attention_layernorm.bias": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.49.post_attention_layernorm.weight": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.49.self_attention.dense.bias": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.49.self_attention.dense.weight": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.49.self_attention.query_key_value.bias": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.49.self_attention.query_key_value.weight": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.5.input_layernorm.bias": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.5.input_layernorm.weight": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.5.mlp.dense_4h_to_h.bias": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.5.mlp.dense_4h_to_h.weight": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.5.mlp.dense_h_to_4h.bias": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.5.mlp.dense_h_to_4h.weight": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.5.post_attention_layernorm.bias": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.5.post_attention_layernorm.weight": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.5.self_attention.dense.bias": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.5.self_attention.dense.weight": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.5.self_attention.query_key_value.bias": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.5.self_attention.query_key_value.weight": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.50.input_layernorm.bias": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.50.input_layernorm.weight": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.50.mlp.dense_4h_to_h.bias": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.50.mlp.dense_4h_to_h.weight": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.50.mlp.dense_h_to_4h.bias": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.50.mlp.dense_h_to_4h.weight": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.50.post_attention_layernorm.bias": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.50.post_attention_layernorm.weight": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.50.self_attention.dense.bias": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.50.self_attention.dense.weight": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.50.self_attention.query_key_value.bias": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.50.self_attention.query_key_value.weight": "pytorch_model-00026-of-00036.bin",
+    "transformer.h.51.input_layernorm.bias": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.51.input_layernorm.weight": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.51.mlp.dense_4h_to_h.bias": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.51.mlp.dense_4h_to_h.weight": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.51.mlp.dense_h_to_4h.bias": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.51.mlp.dense_h_to_4h.weight": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.51.post_attention_layernorm.bias": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.51.post_attention_layernorm.weight": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.51.self_attention.dense.bias": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.51.self_attention.dense.weight": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.51.self_attention.query_key_value.bias": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.51.self_attention.query_key_value.weight": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.52.input_layernorm.bias": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.52.input_layernorm.weight": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.52.mlp.dense_4h_to_h.bias": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.52.mlp.dense_4h_to_h.weight": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.52.mlp.dense_h_to_4h.bias": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.52.mlp.dense_h_to_4h.weight": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.52.post_attention_layernorm.bias": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.52.post_attention_layernorm.weight": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.52.self_attention.dense.bias": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.52.self_attention.dense.weight": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.52.self_attention.query_key_value.bias": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.52.self_attention.query_key_value.weight": "pytorch_model-00027-of-00036.bin",
+    "transformer.h.53.input_layernorm.bias": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.53.input_layernorm.weight": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.53.mlp.dense_4h_to_h.bias": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.53.mlp.dense_4h_to_h.weight": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.53.mlp.dense_h_to_4h.bias": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.53.mlp.dense_h_to_4h.weight": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.53.post_attention_layernorm.bias": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.53.post_attention_layernorm.weight": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.53.self_attention.dense.bias": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.53.self_attention.dense.weight": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.53.self_attention.query_key_value.bias": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.53.self_attention.query_key_value.weight": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.54.input_layernorm.bias": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.54.input_layernorm.weight": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.54.mlp.dense_4h_to_h.bias": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.54.mlp.dense_4h_to_h.weight": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.54.mlp.dense_h_to_4h.bias": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.54.mlp.dense_h_to_4h.weight": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.54.post_attention_layernorm.bias": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.54.post_attention_layernorm.weight": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.54.self_attention.dense.bias": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.54.self_attention.dense.weight": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.54.self_attention.query_key_value.bias": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.54.self_attention.query_key_value.weight": "pytorch_model-00028-of-00036.bin",
+    "transformer.h.55.input_layernorm.bias": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.55.input_layernorm.weight": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.55.mlp.dense_4h_to_h.bias": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.55.mlp.dense_4h_to_h.weight": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.55.mlp.dense_h_to_4h.bias": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.55.mlp.dense_h_to_4h.weight": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.55.post_attention_layernorm.bias": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.55.post_attention_layernorm.weight": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.55.self_attention.dense.bias": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.55.self_attention.dense.weight": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.55.self_attention.query_key_value.bias": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.55.self_attention.query_key_value.weight": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.56.input_layernorm.bias": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.56.input_layernorm.weight": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.56.mlp.dense_4h_to_h.bias": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.56.mlp.dense_4h_to_h.weight": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.56.mlp.dense_h_to_4h.bias": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.56.mlp.dense_h_to_4h.weight": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.56.post_attention_layernorm.bias": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.56.post_attention_layernorm.weight": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.56.self_attention.dense.bias": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.56.self_attention.dense.weight": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.56.self_attention.query_key_value.bias": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.56.self_attention.query_key_value.weight": "pytorch_model-00029-of-00036.bin",
+    "transformer.h.57.input_layernorm.bias": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.57.input_layernorm.weight": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.57.mlp.dense_4h_to_h.bias": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.57.mlp.dense_4h_to_h.weight": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.57.mlp.dense_h_to_4h.bias": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.57.mlp.dense_h_to_4h.weight": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.57.post_attention_layernorm.bias": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.57.post_attention_layernorm.weight": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.57.self_attention.dense.bias": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.57.self_attention.dense.weight": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.57.self_attention.query_key_value.bias": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.57.self_attention.query_key_value.weight": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.58.input_layernorm.bias": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.58.input_layernorm.weight": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.58.mlp.dense_4h_to_h.bias": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.58.mlp.dense_4h_to_h.weight": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.58.mlp.dense_h_to_4h.bias": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.58.mlp.dense_h_to_4h.weight": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.58.post_attention_layernorm.bias": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.58.post_attention_layernorm.weight": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.58.self_attention.dense.bias": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.58.self_attention.dense.weight": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.58.self_attention.query_key_value.bias": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.58.self_attention.query_key_value.weight": "pytorch_model-00030-of-00036.bin",
+    "transformer.h.59.input_layernorm.bias": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.59.input_layernorm.weight": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.59.mlp.dense_4h_to_h.bias": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.59.mlp.dense_4h_to_h.weight": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.59.mlp.dense_h_to_4h.bias": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.59.mlp.dense_h_to_4h.weight": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.59.post_attention_layernorm.bias": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.59.post_attention_layernorm.weight": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.59.self_attention.dense.bias": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.59.self_attention.dense.weight": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.59.self_attention.query_key_value.bias": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.59.self_attention.query_key_value.weight": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.6.input_layernorm.bias": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.6.input_layernorm.weight": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.6.mlp.dense_4h_to_h.bias": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.6.mlp.dense_4h_to_h.weight": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.6.mlp.dense_h_to_4h.bias": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.6.mlp.dense_h_to_4h.weight": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.6.post_attention_layernorm.bias": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.6.post_attention_layernorm.weight": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.6.self_attention.dense.bias": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.6.self_attention.dense.weight": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.6.self_attention.query_key_value.bias": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.6.self_attention.query_key_value.weight": "pytorch_model-00004-of-00036.bin",
+    "transformer.h.60.input_layernorm.bias": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.60.input_layernorm.weight": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.60.mlp.dense_4h_to_h.bias": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.60.mlp.dense_4h_to_h.weight": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.60.mlp.dense_h_to_4h.bias": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.60.mlp.dense_h_to_4h.weight": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.60.post_attention_layernorm.bias": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.60.post_attention_layernorm.weight": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.60.self_attention.dense.bias": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.60.self_attention.dense.weight": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.60.self_attention.query_key_value.bias": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.60.self_attention.query_key_value.weight": "pytorch_model-00031-of-00036.bin",
+    "transformer.h.61.input_layernorm.bias": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.61.input_layernorm.weight": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.61.mlp.dense_4h_to_h.bias": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.61.mlp.dense_4h_to_h.weight": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.61.mlp.dense_h_to_4h.bias": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.61.mlp.dense_h_to_4h.weight": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.61.post_attention_layernorm.bias": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.61.post_attention_layernorm.weight": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.61.self_attention.dense.bias": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.61.self_attention.dense.weight": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.61.self_attention.query_key_value.bias": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.61.self_attention.query_key_value.weight": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.62.input_layernorm.bias": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.62.input_layernorm.weight": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.62.mlp.dense_4h_to_h.bias": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.62.mlp.dense_4h_to_h.weight": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.62.mlp.dense_h_to_4h.bias": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.62.mlp.dense_h_to_4h.weight": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.62.post_attention_layernorm.bias": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.62.post_attention_layernorm.weight": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.62.self_attention.dense.bias": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.62.self_attention.dense.weight": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.62.self_attention.query_key_value.bias": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.62.self_attention.query_key_value.weight": "pytorch_model-00032-of-00036.bin",
+    "transformer.h.63.input_layernorm.bias": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.63.input_layernorm.weight": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.63.mlp.dense_4h_to_h.bias": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.63.mlp.dense_4h_to_h.weight": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.63.mlp.dense_h_to_4h.bias": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.63.mlp.dense_h_to_4h.weight": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.63.post_attention_layernorm.bias": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.63.post_attention_layernorm.weight": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.63.self_attention.dense.bias": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.63.self_attention.dense.weight": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.63.self_attention.query_key_value.bias": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.63.self_attention.query_key_value.weight": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.64.input_layernorm.bias": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.64.input_layernorm.weight": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.64.mlp.dense_4h_to_h.bias": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.64.mlp.dense_4h_to_h.weight": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.64.mlp.dense_h_to_4h.bias": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.64.mlp.dense_h_to_4h.weight": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.64.post_attention_layernorm.bias": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.64.post_attention_layernorm.weight": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.64.self_attention.dense.bias": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.64.self_attention.dense.weight": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.64.self_attention.query_key_value.bias": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.64.self_attention.query_key_value.weight": "pytorch_model-00033-of-00036.bin",
+    "transformer.h.65.input_layernorm.bias": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.65.input_layernorm.weight": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.65.mlp.dense_4h_to_h.bias": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.65.mlp.dense_4h_to_h.weight": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.65.mlp.dense_h_to_4h.bias": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.65.mlp.dense_h_to_4h.weight": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.65.post_attention_layernorm.bias": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.65.post_attention_layernorm.weight": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.65.self_attention.dense.bias": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.65.self_attention.dense.weight": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.65.self_attention.query_key_value.bias": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.65.self_attention.query_key_value.weight": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.66.input_layernorm.bias": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.66.input_layernorm.weight": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.66.mlp.dense_4h_to_h.bias": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.66.mlp.dense_4h_to_h.weight": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.66.mlp.dense_h_to_4h.bias": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.66.mlp.dense_h_to_4h.weight": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.66.post_attention_layernorm.bias": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.66.post_attention_layernorm.weight": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.66.self_attention.dense.bias": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.66.self_attention.dense.weight": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.66.self_attention.query_key_value.bias": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.66.self_attention.query_key_value.weight": "pytorch_model-00034-of-00036.bin",
+    "transformer.h.67.input_layernorm.bias": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.67.input_layernorm.weight": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.67.mlp.dense_4h_to_h.bias": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.67.mlp.dense_4h_to_h.weight": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.67.mlp.dense_h_to_4h.bias": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.67.mlp.dense_h_to_4h.weight": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.67.post_attention_layernorm.bias": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.67.post_attention_layernorm.weight": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.67.self_attention.dense.bias": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.67.self_attention.dense.weight": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.67.self_attention.query_key_value.bias": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.67.self_attention.query_key_value.weight": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.68.input_layernorm.bias": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.68.input_layernorm.weight": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.68.mlp.dense_4h_to_h.bias": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.68.mlp.dense_4h_to_h.weight": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.68.mlp.dense_h_to_4h.bias": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.68.mlp.dense_h_to_4h.weight": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.68.post_attention_layernorm.bias": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.68.post_attention_layernorm.weight": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.68.self_attention.dense.bias": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.68.self_attention.dense.weight": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.68.self_attention.query_key_value.bias": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.68.self_attention.query_key_value.weight": "pytorch_model-00035-of-00036.bin",
+    "transformer.h.69.input_layernorm.bias": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.69.input_layernorm.weight": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.69.mlp.dense_4h_to_h.bias": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.69.mlp.dense_4h_to_h.weight": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.69.mlp.dense_h_to_4h.bias": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.69.mlp.dense_h_to_4h.weight": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.69.post_attention_layernorm.bias": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.69.post_attention_layernorm.weight": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.69.self_attention.dense.bias": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.69.self_attention.dense.weight": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.69.self_attention.query_key_value.bias": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.69.self_attention.query_key_value.weight": "pytorch_model-00036-of-00036.bin",
+    "transformer.h.7.input_layernorm.bias": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.7.input_layernorm.weight": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.7.mlp.dense_4h_to_h.bias": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.7.mlp.dense_4h_to_h.weight": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.7.mlp.dense_h_to_4h.bias": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.7.mlp.dense_h_to_4h.weight": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.7.post_attention_layernorm.bias": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.7.post_attention_layernorm.weight": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.7.self_attention.dense.bias": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.7.self_attention.dense.weight": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.7.self_attention.query_key_value.bias": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.7.self_attention.query_key_value.weight": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.8.input_layernorm.bias": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.8.input_layernorm.weight": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.8.mlp.dense_4h_to_h.bias": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.8.mlp.dense_4h_to_h.weight": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.8.mlp.dense_h_to_4h.bias": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.8.mlp.dense_h_to_4h.weight": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.8.post_attention_layernorm.bias": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.8.post_attention_layernorm.weight": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.8.self_attention.dense.bias": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.8.self_attention.dense.weight": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.8.self_attention.query_key_value.bias": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.8.self_attention.query_key_value.weight": "pytorch_model-00005-of-00036.bin",
+    "transformer.h.9.input_layernorm.bias": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.9.input_layernorm.weight": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.9.mlp.dense_4h_to_h.bias": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.9.mlp.dense_4h_to_h.weight": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.9.mlp.dense_h_to_4h.bias": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.9.mlp.dense_h_to_4h.weight": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.9.post_attention_layernorm.bias": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.9.post_attention_layernorm.weight": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.9.self_attention.dense.bias": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.9.self_attention.dense.weight": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.9.self_attention.query_key_value.bias": "pytorch_model-00006-of-00036.bin",
+    "transformer.h.9.self_attention.query_key_value.weight": "pytorch_model-00006-of-00036.bin",
+    "transformer.ln_f.bias": "pytorch_model-00036-of-00036.bin",
+    "transformer.ln_f.weight": "pytorch_model-00036-of-00036.bin",
+    "transformer.word_embeddings.weight": "pytorch_model-00001-of-00036.bin",
+    "transformer.word_embeddings_layernorm.bias": "pytorch_model-00001-of-00036.bin",
+    "transformer.word_embeddings_layernorm.weight": "pytorch_model-00001-of-00036.bin"
+  }
+}

special_tokens_map.json

@@ -0,0 +1,30 @@
+{
+  "bos_token": {
+    "content": "<|endoftext|>",
+    "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": "[PAD]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0d908b4f9326e0998815690e325b6abbd378978553e10627924dd825db7e243
+size 17477553

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003

tokenizer_config.json

@@ -0,0 +1,49 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}