add model

config.json

@@ -0,0 +1,35 @@
+{
+  "activation_function": "quick_gelu",
+  "architectures": [
+    "GPT2LMHeadCustomModel"
+  ],
+  "auto_map": {
+    "AutoModelForCausalLM": "modeling_gpt2_summarize.GPT2LMHeadCustomModel"
+  },
+  "attn_pdrop": 0,
+  "bos_token_id": 50256,
+  "embd_pdrop": 0,
+  "eos_token_id": 50256,
+  "initializer_range": 0.02,
+  "layer_norm_epsilon": 1e-05,
+  "model_type": "gpt2",
+  "n_embd": 2048,
+  "n_head": 16,
+  "n_inner": null,
+  "n_layer": 24,
+  "n_positions": 2048,
+  "reorder_and_upcast_attn": false,
+  "resid_pdrop": 0,
+  "scale_attn_by_inverse_layer_idx": false,
+  "scale_attn_weights": true,
+  "summary_activation": null,
+  "summary_first_dropout": 0.1,
+  "summary_proj_to_labels": true,
+  "summary_type": "cls_index",
+  "summary_use_proj": true,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.25.1",
+  "use_cache": true,
+  "vocab_size": 50304
+}

modeling_gpt2_summarize.py

@@ -0,0 +1,1074 @@
+# coding=utf-8
+# Copyright 2018 The OpenAI Team Authors and HuggingFace Inc. team.
+# Copyright (c) 2018, NVIDIA CORPORATION.  All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+PyTorch OpenAI GPT-2 model in Learning to Summarize with Human Feedback.
+https://openai.com/blog/learning-to-summarize-with-human-feedback/
+https://arxiv.org/abs/2009.01325
+https://github.com/openai/summarize-from-feedback
+"""
+
+import math
+from typing import Optional, Tuple, Union
+
+import torch
+import torch.utils.checkpoint
+from torch import nn
+from torch.cuda.amp import autocast
+from torch.nn import CrossEntropyLoss
+
+from transformers.activations import ACT2FN
+from transformers.modeling_outputs import (
+    BaseModelOutputWithPastAndCrossAttentions,
+    CausalLMOutputWithCrossAttentions,
+)
+from transformers.modeling_utils import PreTrainedModel
+from transformers.utils import (
+    add_code_sample_docstrings,
+    add_start_docstrings,
+    add_start_docstrings_to_model_forward,
+    logging,
+)
+from transformers.utils.model_parallel_utils import assert_device_map, get_device_map
+from transformers import GPT2Config
+
+import numpy as np
+
+logger = logging.get_logger(__name__)
+
+_CHECKPOINT_FOR_DOC = "gpt2"
+_CONFIG_FOR_DOC = "GPT2Config"
+_TOKENIZER_FOR_DOC = "GPT2Tokenizer"
+
+
+class Conv1D(nn.Module):
+    """
+    1D-convolutional layer as defined by Radford et al. for OpenAI GPT (and also used in GPT-2).
+
+    Basically works like a linear layer but the weights are transposed.
+
+    Args:
+        nf (`int`): The number of output features.
+        nx (`int`): The number of input features.
+    """
+
+    def __init__(self, nf, nx, bias=True):
+        super().__init__()
+        self.nf = nf
+        w = torch.empty(nx, nf)
+        nn.init.normal_(w, std=0.02)
+        self.weight = nn.Parameter(w)
+        if bias:
+            self.bias = nn.Parameter(torch.zeros(nf))
+        else:
+            self.bias = None
+
+    def forward(self, x):
+        size_out = x.size()[:-1] + (self.nf,)
+        if self.bias is not None:
+            x = torch.addmm(self.bias, x.view(-1, x.size(-1)), self.weight)
+        else:
+            x = torch.mm(x.view(-1, x.size(-1)), self.weight)
+        x = x.view(size_out)
+        return x
+
+
+def prune_conv1d_layer(layer: Conv1D, index: torch.LongTensor, dim: int = 1, bias: bool = True) -> Conv1D:
+    """
+    Prune a Conv1D layer to keep only entries in index. A Conv1D work as a Linear layer (see e.g. BERT) but the weights
+    are transposed.
+
+    Used to remove heads.
+
+    Args:
+        layer ([`~pytorch_utils.Conv1D`]): The layer to prune.
+        index (`torch.LongTensor`): The indices to keep in the layer.
+        dim (`int`, *optional*, defaults to 1): The dimension on which to keep the indices.
+
+    Returns:
+        [`~pytorch_utils.Conv1D`]: The pruned layer as a new layer with `requires_grad=True`.
+    """
+    index = index.to(layer.weight.device)
+    W = layer.weight.index_select(dim, index).clone().detach()
+    if bias:
+        if dim == 0:
+            b = layer.bias.clone().detach()
+        else:
+            b = layer.bias[index].clone().detach()
+    new_size = list(layer.weight.size())
+    new_size[dim] = len(index)
+    new_layer = Conv1D(new_size[1], new_size[0], bias).to(layer.weight.device)
+    new_layer.weight.requires_grad = False
+    new_layer.weight.copy_(W.contiguous())
+    new_layer.weight.requires_grad = True
+    if bias:
+        new_layer.bias.requires_grad = False
+        new_layer.bias.copy_(b.contiguous())
+        new_layer.bias.requires_grad = True
+    return new_layer
+
+
+class GPT2Attention(nn.Module):
+    def __init__(self, config, is_cross_attention=False, layer_idx=None):
+        super().__init__()
+
+        max_positions = config.max_position_embeddings
+        self.register_buffer(
+            "bias",
+            torch.tril(torch.ones((max_positions, max_positions), dtype=torch.uint8)).view(
+                1, 1, max_positions, max_positions
+            ),
+        )
+        self.register_buffer("masked_bias", torch.tensor(-1e4))
+
+        self.embed_dim = config.hidden_size
+        self.num_heads = config.num_attention_heads
+        self.head_dim = self.embed_dim // self.num_heads
+        self.split_size = self.embed_dim
+        if self.head_dim * self.num_heads != self.embed_dim:
+            raise ValueError(
+                f"`embed_dim` must be divisible by num_heads (got `embed_dim`: {self.embed_dim} and `num_heads`:"
+                f" {self.num_heads})."
+            )
+
+        self.scale_attn_weights = config.scale_attn_weights
+        self.is_cross_attention = is_cross_attention
+
+        # Layer-wise attention scaling, reordering, and upcasting
+        self.scale_attn_by_inverse_layer_idx = config.scale_attn_by_inverse_layer_idx
+        self.layer_idx = layer_idx
+        self.reorder_and_upcast_attn = config.reorder_and_upcast_attn
+
+        if self.is_cross_attention:
+            raise NotImplementedError("should not enter this path.")
+            self.c_attn = Conv1D(2 * self.embed_dim, self.embed_dim)
+            self.q_attn = Conv1D(self.embed_dim, self.embed_dim)
+        else:
+            #self.c_attn = Conv1D(3 * self.embed_dim, self.embed_dim)
+            self.q_proj = Conv1D(self.embed_dim, self.embed_dim)
+            self.k_proj = Conv1D(self.embed_dim, self.embed_dim, bias=False)
+            self.v_proj = Conv1D(self.embed_dim, self.embed_dim)
+        self.c_proj = Conv1D(self.embed_dim, self.embed_dim)
+
+        self.attn_dropout = nn.Dropout(config.attn_pdrop)
+        self.resid_dropout = nn.Dropout(config.resid_pdrop)
+
+        self.pruned_heads = set()
+
+    def prune_heads(self, heads):
+        if len(heads) == 0:
+            return
+        heads, index = find_pruneable_heads_and_indices(heads, self.num_heads, self.head_dim, self.pruned_heads)
+        index_attn = torch.cat([index, index + self.split_size, index + (2 * self.split_size)])
+
+        # Prune conv1d layers
+        #self.c_attn = prune_conv1d_layer(self.c_attn, index_attn, dim=1)
+        self.q_proj = prune_conv1d_layer(self.q_proj, index_attn, dim=1)
+        self.k_proj = prune_conv1d_layer(self.q_proj, index_attn, dim=1, bias=False)
+        self.v_proj = prune_conv1d_layer(self.q_proj, index_attn, dim=1)
+        self.c_proj = prune_conv1d_layer(self.c_proj, index, dim=0)
+
+        # Update hyper params
+        self.split_size = (self.split_size // self.num_heads) * (self.num_heads - len(heads))
+        self.num_heads = self.num_heads - len(heads)
+        self.pruned_heads = self.pruned_heads.union(heads)
+
+    def _attn(self, query, key, value, attention_mask=None, head_mask=None):
+        if self.scale_attn_weights:
+            # Pre-divide by fp16_stability_scale to prevent fp16 overflow
+            softmax_scale = 1.0 / np.sqrt(np.sqrt(query.size(-1)))
+            query = query * softmax_scale
+            key = key * softmax_scale
+
+        attn_weights = torch.matmul(query, key)
+
+        if not self.is_cross_attention:
+            # if only "normal" attention layer implements causal mask
+            query_length, key_length = query.size(-2), key.size(-1)
+            causal_mask = self.bias[:, :, key_length - query_length : key_length, :key_length].to(torch.bool)
+            mask_value = torch.finfo(attn_weights.dtype).min
+            # Need to be a tensor, otherwise we get error: `RuntimeError: expected scalar type float but found double`.
+            # Need to be on the same device, otherwise `RuntimeError: ..., x and y to be on the same device`
+            mask_value = torch.full([], mask_value, dtype=attn_weights.dtype).to(attn_weights.device)
+            attn_weights = torch.where(causal_mask, attn_weights, mask_value)
+
+        if attention_mask is not None:
+            # Apply the attention mask
+            attn_weights = attn_weights + attention_mask
+
+        attn_weights = nn.functional.softmax(attn_weights, dim=-1)
+
+        # Downcast (if necessary) back to V's dtype (if in mixed-precision) -- No-Op otherwise
+        attn_weights = attn_weights.type(value.dtype)
+        attn_weights = self.attn_dropout(attn_weights)
+
+        # Mask heads if we want to
+        if head_mask is not None:
+            attn_weights = attn_weights * head_mask
+
+        attn_output = torch.matmul(attn_weights, value)
+
+        return attn_output, attn_weights
+
+    def _upcast_and_reordered_attn(self, query, key, value, attention_mask=None, head_mask=None):
+        # Use `torch.baddbmm` (a bit more efficient w/ alpha param for scaling -- from Megatron-LM)
+        bsz, num_heads, q_seq_len, dk = query.size()
+        _, _, k_seq_len, _ = key.size()
+
+        # Preallocate attn_weights for `baddbmm`
+        attn_weights = torch.empty(bsz * num_heads, q_seq_len, k_seq_len, dtype=torch.float32, device=query.device)
+
+        # Compute Scale Factor
+        scale_factor = 1.0
+        if self.scale_attn_weights:
+            scale_factor /= float(value.size(-1)) ** 0.5
+
+        if self.scale_attn_by_inverse_layer_idx:
+            scale_factor /= float(self.layer_idx + 1)
+
+        # Upcast (turn off autocast) and reorder (Scale K by 1 / root(dk))
+        with autocast(enabled=False):
+            q, k = query.reshape(-1, q_seq_len, dk), key.transpose(-1, -2).reshape(-1, dk, k_seq_len)
+            attn_weights = torch.baddbmm(attn_weights, q.float(), k.float(), beta=0, alpha=scale_factor)
+            attn_weights = attn_weights.reshape(bsz, num_heads, q_seq_len, k_seq_len)
+
+        if not self.is_cross_attention:
+            # if only "normal" attention layer implements causal mask
+            query_length, key_length = query.size(-2), key.size(-2)
+            causal_mask = self.bias[:, :, key_length - query_length : key_length, :key_length].bool()
+            mask_value = torch.finfo(attn_weights.dtype).min
+            # Need to be a tensor, otherwise we get error: `RuntimeError: expected scalar type float but found double`.
+            # Need to be on the same device, otherwise `RuntimeError: ..., x and y to be on the same device`
+            mask_value = torch.tensor(mask_value, dtype=attn_weights.dtype).to(attn_weights.device)
+            attn_weights = torch.where(causal_mask, attn_weights, mask_value)
+
+        if attention_mask is not None:
+            # Apply the attention mask
+            attn_weights = attn_weights + attention_mask
+
+        attn_weights = nn.functional.softmax(attn_weights, dim=-1)
+
+        # Downcast (if necessary) back to V's dtype (if in mixed-precision) -- No-Op if otherwise
+        if attn_weights.dtype != torch.float32:
+            raise RuntimeError("Error with upcasting, attn_weights does not have dtype torch.float32")
+        attn_weights = attn_weights.type(value.dtype)
+        attn_weights = self.attn_dropout(attn_weights)
+
+        # Mask heads if we want to
+        if head_mask is not None:
+            attn_weights = attn_weights * head_mask
+
+        attn_output = torch.matmul(attn_weights, value)
+
+        return attn_output, attn_weights
+
+    def _split_heads(self, tensor, num_heads, attn_head_size, k=False):
+        """
+        Splits hidden_size dim into attn_head_size and num_heads
+        """
+        new_shape = tensor.size()[:-1] + (num_heads, attn_head_size)
+        tensor = tensor.view(new_shape)
+        if k:
+            return tensor.permute(0, 2, 3, 1)  # (batch, head, head_features, seq_length)
+        else:
+            return tensor.permute(0, 2, 1, 3)  # (batch, head, seq_length, head_features)
+
+    def _merge_heads(self, tensor, num_heads, attn_head_size):
+        """
+        Merges attn_head_size dim and num_attn_heads dim into hidden_size
+        """
+        tensor = tensor.permute(0, 2, 1, 3).contiguous()
+        new_shape = tensor.size()[:-2] + (num_heads * attn_head_size,)
+        return tensor.view(new_shape)
+
+    def forward(
+        self,
+        hidden_states: Optional[Tuple[torch.FloatTensor]],
+        layer_past: Optional[Tuple[torch.Tensor]] = None,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+        encoder_hidden_states: Optional[torch.Tensor] = None,
+        encoder_attention_mask: Optional[torch.FloatTensor] = None,
+        use_cache: Optional[bool] = False,
+        output_attentions: Optional[bool] = False,
+    ) -> Tuple[Union[torch.Tensor, Tuple[torch.Tensor]], ...]:
+        if encoder_hidden_states is not None:
+            raise NotImplementedError("should not enter this path.")
+            if not hasattr(self, "q_attn"):
+                raise ValueError(
+                    "If class is used as cross attention, the weights `q_attn` have to be defined. "
+                    "Please make sure to instantiate class with `GPT2Attention(..., is_cross_attention=True)`."
+                )
+
+            query = self.q_attn(hidden_states)
+            key, value = self.c_attn(encoder_hidden_states).split(self.split_size, dim=2)
+            attention_mask = encoder_attention_mask
+        else:
+            #query, key, value = self.c_attn(hidden_states).split(self.split_size, dim=2)
+            query = self.q_proj(hidden_states)
+            key = self.k_proj(hidden_states)
+            value = self.v_proj(hidden_states)
+
+        query = self._split_heads(query, self.num_heads, self.head_dim)
+        key = self._split_heads(key, self.num_heads, self.head_dim, k=True)
+        value = self._split_heads(value, self.num_heads, self.head_dim)
+
+        if layer_past is not None:
+            past_key, past_value = layer_past
+            key = torch.cat((past_key, key), dim=-2)
+            value = torch.cat((past_value, value), dim=-2)
+
+        if use_cache is True:
+            present = (key, value)
+        else:
+            present = None
+
+        if self.reorder_and_upcast_attn:
+            raise NotImplementedError("should not enter this path.")
+            attn_output, attn_weights = self._upcast_and_reordered_attn(query, key, value, attention_mask, head_mask)
+        else:
+            attn_output, attn_weights = self._attn(query, key, value, attention_mask, head_mask)
+
+        attn_output = self._merge_heads(attn_output, self.num_heads, self.head_dim)
+        attn_output = self.c_proj(attn_output)
+        attn_output = self.resid_dropout(attn_output)
+
+        outputs = (attn_output, present)
+        if output_attentions:
+            outputs += (attn_weights,)
+
+        return outputs  # a, present, (attentions)
+
+
+class GPT2MLP(nn.Module):
+    def __init__(self, intermediate_size, config):
+        super().__init__()
+        embed_dim = config.hidden_size
+        self.c_fc = Conv1D(intermediate_size, embed_dim)
+        self.c_proj = Conv1D(embed_dim, intermediate_size)
+        self.act = ACT2FN[config.activation_function]
+        self.dropout = nn.Dropout(config.resid_pdrop)
+
+    def forward(self, hidden_states: Optional[Tuple[torch.FloatTensor]]) -> torch.FloatTensor:
+        hidden_states = self.c_fc(hidden_states)
+        hidden_states = self.act(hidden_states)
+        hidden_states = self.c_proj(hidden_states)
+        hidden_states = self.dropout(hidden_states)
+        return hidden_states
+
+
+class GPT2Block(nn.Module):
+    def __init__(self, config, layer_idx=None):
+        super().__init__()
+        hidden_size = config.hidden_size
+        inner_dim = config.n_inner if config.n_inner is not None else 4 * hidden_size
+
+        self.ln_1 = nn.LayerNorm(hidden_size, eps=config.layer_norm_epsilon)
+        self.attn = GPT2Attention(config, layer_idx=layer_idx)
+        self.ln_2 = nn.LayerNorm(hidden_size, eps=config.layer_norm_epsilon)
+
+        if config.add_cross_attention:
+            self.crossattention = GPT2Attention(config, is_cross_attention=True, layer_idx=layer_idx)
+            self.ln_cross_attn = nn.LayerNorm(hidden_size, eps=config.layer_norm_epsilon)
+
+        self.mlp = GPT2MLP(inner_dim, config)
+
+    def forward(
+        self,
+        hidden_states: Optional[Tuple[torch.FloatTensor]],
+        layer_past: Optional[Tuple[torch.Tensor]] = None,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+        encoder_hidden_states: Optional[torch.Tensor] = None,
+        encoder_attention_mask: Optional[torch.FloatTensor] = None,
+        use_cache: Optional[bool] = False,
+        output_attentions: Optional[bool] = False,
+    ) -> Union[Tuple[torch.Tensor], Optional[Tuple[torch.Tensor, Tuple[torch.FloatTensor, ...]]]]:
+        inputs = hidden_states
+        residual = hidden_states
+        hidden_states = self.ln_1(hidden_states)
+        attn_outputs = self.attn(
+            hidden_states,
+            layer_past=layer_past,
+            attention_mask=attention_mask,
+            head_mask=head_mask,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+        )
+        attn_output = attn_outputs[0]  # output_attn: a, present, (attentions)
+        outputs = attn_outputs[1:]
+        # residual connection
+        hidden_states = attn_output + residual
+
+        if encoder_hidden_states is not None:
+            # add one self-attention block for cross-attention
+            if not hasattr(self, "crossattention"):
+                raise ValueError(
+                    f"If `encoder_hidden_states` are passed, {self} has to be instantiated with "
+                    "cross-attention layers by setting `config.add_cross_attention=True`"
+                )
+            residual = hidden_states
+            hidden_states = self.ln_cross_attn(hidden_states)
+            cross_attn_outputs = self.crossattention(
+                hidden_states,
+                attention_mask=attention_mask,
+                head_mask=head_mask,
+                encoder_hidden_states=encoder_hidden_states,
+                encoder_attention_mask=encoder_attention_mask,
+                output_attentions=output_attentions,
+            )
+            attn_output = cross_attn_outputs[0]
+            # residual connection
+            hidden_states = residual + attn_output
+            outputs = outputs + cross_attn_outputs[2:]  # add cross attentions if we output attention weights
+
+        residual = hidden_states
+        hidden_states = self.ln_2(hidden_states)
+        feed_forward_hidden_states = self.mlp(hidden_states)
+        # we use unnormalized inputs to all functions for residuals
+        # 1.0 here is `res_scale``
+        hidden_states = inputs + 1.0 * (attn_output + feed_forward_hidden_states)
+
+        if use_cache:
+            outputs = (hidden_states,) + outputs
+        else:
+            outputs = (hidden_states,) + outputs[1:]
+
+        return outputs  # hidden_states, present, (attentions, cross_attentions)
+
+
+class GPT2PreTrainedModel(PreTrainedModel):
+    """
+    An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
+    models.
+    """
+
+    config_class = GPT2Config
+    base_model_prefix = "transformer"
+    is_parallelizable = True
+    supports_gradient_checkpointing = True
+    _no_split_modules = ["GPT2Block"]
+
+    def __init__(self, *inputs, **kwargs):
+        super().__init__(*inputs, **kwargs)
+
+    def _init_weights(self, module):
+        """Initialize the weights."""
+        if isinstance(module, (nn.Linear, Conv1D)):
+            # Slightly different from the TF version which uses truncated_normal for initialization
+            # cf https://github.com/pytorch/pytorch/pull/5617
+            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)
+
+        # Reinitialize selected weights subject to the OpenAI GPT-2 Paper Scheme:
+        #   > A modified initialization which accounts for the accumulation on the residual path with model depth. Scale
+        #   > the weights of residual layers at initialization by a factor of 1/√N where N is the # of residual layers.
+        #   >   -- GPT-2 :: https://openai.com/blog/better-language-models/
+        #
+        # Reference (Megatron-LM): https://github.com/NVIDIA/Megatron-LM/blob/main/megatron/model/gpt_model.py
+        for name, p in module.named_parameters():
+            if name == "c_proj.weight":
+                # Special Scaled Initialization --> There are 2 Layer Norms per Transformer Block
+                p.data.normal_(mean=0.0, std=(self.config.initializer_range / math.sqrt(2 * self.config.n_layer)))
+
+    def _set_gradient_checkpointing(self, module, value=False):
+        if isinstance(module, GPT2Model):
+            module.gradient_checkpointing = value
+
+
+GPT2_START_DOCSTRING = r"""
+
+    This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic methods the
+    library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
+    etc.)
+
+    This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass.
+    Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage
+    and behavior.
+
+    Parameters:
+        config ([`GPT2Config`]): Model configuration class with all the parameters of the model.
+            Initializing with a config file does not load the weights associated with the model, only the
+            configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model weights.
+"""
+
+GPT2_INPUTS_DOCSTRING = r"""
+    Args:
+        input_ids (`torch.LongTensor` of shape `(batch_size, input_ids_length)`):
+            `input_ids_length` = `sequence_length` if `past_key_values` is `None` else
+            `past_key_values[0][0].shape[-2]` (`sequence_length` of input past key value states). Indices of input
+            sequence tokens in the vocabulary.
+
+            If `past_key_values` is used, only `input_ids` that do not have their past calculated should be passed as
+            `input_ids`.
+
+            Indices can be obtained using [`GPT2Tokenizer`]. See [`PreTrainedTokenizer.encode`] and
+            [`PreTrainedTokenizer.__call__`] for details.
+
+            [What are input IDs?](../glossary#input-ids)
+        past_key_values (`Tuple[Tuple[torch.Tensor]]` of length `config.n_layers`):
+            Contains precomputed hidden-states (key and values in the attention blocks) as computed by the model (see
+            `past_key_values` output below). Can be used to speed up sequential decoding. The `input_ids` which have
+            their past given to this model should not be passed as `input_ids` as they have already been computed.
+        attention_mask (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
+
+            - 1 for tokens that are **not masked**,
+            - 0 for tokens that are **masked**.
+
+            If `past_key_values` is used, `attention_mask` needs to contain the masking strategy that was used for
+            `past_key_values`. In other words, the `attention_mask` always has to have the length:
+            `len(past_key_values) + len(input_ids)`
+
+            [What are attention masks?](../glossary#attention-mask)
+        token_type_ids (`torch.LongTensor` of shape `(batch_size, input_ids_length)`, *optional*):
+            Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0,
+            1]`:
+
+            - 0 corresponds to a *sentence A* token,
+            - 1 corresponds to a *sentence B* token.
+
+            [What are token type IDs?](../glossary#token-type-ids)
+        position_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0,
+            config.max_position_embeddings - 1]`.
+
+            [What are position IDs?](../glossary#position-ids)
+        head_mask (`torch.FloatTensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*):
+            Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
+
+            - 1 indicates the head is **not masked**,
+            - 0 indicates the head is **masked**.
+
+        inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
+            Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This
+            is useful if you want more control over how to convert `input_ids` indices into associated vectors than the
+            model's internal embedding lookup matrix.
+
+            If `past_key_values` is used, optionally only the last `inputs_embeds` have to be input (see
+            `past_key_values`).
+        use_cache (`bool`, *optional*):
+            If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see
+            `past_key_values`).
+        output_attentions (`bool`, *optional*):
+            Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
+            tensors for more detail.
+        output_hidden_states (`bool`, *optional*):
+            Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
+            more detail.
+        return_dict (`bool`, *optional*):
+            Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
+"""
+PARALLELIZE_DOCSTRING = r"""
+    This is an experimental feature and is a subject to change at a moment's notice.
+
+    Uses a device map to distribute attention modules of the model across several devices. If no device map is given,
+    it will evenly distribute blocks across all devices.
+
+    Args:
+        device_map (`Dict[int, list]`, optional, defaults to None):
+            A dictionary that maps attention modules to devices. Note that the embedding module and LMHead are always
+            automatically mapped to the first device (for esoteric reasons). That means that the first device should
+            have fewer attention modules mapped to it than other devices. For reference, the gpt2 models have the
+            following number of attention modules:
+
+                - gpt2: 12
+                - gpt2-medium: 24
+                - gpt2-large: 36
+                - gpt2-xl: 48
+
+    Example:
+
+    ```python
+    # Here is an example of a device map on a machine with 4 GPUs using gpt2-xl, which has a total of 48 attention modules:
+    model = GPT2LMHeadModel.from_pretrained("gpt2-xl")
+    device_map = {
+        0: [0, 1, 2, 3, 4, 5, 6, 7, 8],
+        1: [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21],
+        2: [22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34],
+        3: [35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47],
+    }
+    model.parallelize(device_map)
+    ```
+"""
+DEPARALLELIZE_DOCSTRING = r"""
+    Moves the model to cpu from a model parallel state.
+
+    Example:
+
+    ```python
+    # On a 4 GPU machine with gpt2-large:
+    model = GPT2LMHeadModel.from_pretrained("gpt2-large")
+    device_map = {
+        0: [0, 1, 2, 3, 4, 5, 6, 7],
+        1: [8, 9, 10, 11, 12, 13, 14, 15],
+        2: [16, 17, 18, 19, 20, 21, 22, 23],
+        3: [24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35],
+    }
+    model.parallelize(device_map)  # Splits the model across several devices
+    model.deparallelize()  # Put the model back on cpu and cleans memory by calling torch.cuda.empty_cache()
+    ```
+"""
+
+
+@add_start_docstrings(
+    "The bare GPT2 Model transformer outputting raw hidden-states without any specific head on top.",
+    GPT2_START_DOCSTRING,
+)
+class GPT2Model(GPT2PreTrainedModel):
+    _keys_to_ignore_on_load_missing = ["attn.masked_bias"]
+
+    def __init__(self, config):
+        super().__init__(config)
+
+        self.embed_dim = config.hidden_size
+
+        self.wte = nn.Embedding(config.vocab_size, self.embed_dim)
+        self.wpe = nn.Embedding(config.max_position_embeddings, self.embed_dim)
+
+        self.drop = nn.Dropout(config.embd_pdrop)
+        self.h = nn.ModuleList([GPT2Block(config, layer_idx=i) for i in range(config.num_hidden_layers)])
+        self.ln_f = nn.LayerNorm(self.embed_dim, eps=config.layer_norm_epsilon)
+
+        # Model parallel
+        self.model_parallel = False
+        self.device_map = None
+        self.gradient_checkpointing = False
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    @add_start_docstrings(PARALLELIZE_DOCSTRING)
+    def parallelize(self, device_map=None):
+        # Check validity of device_map
+        self.device_map = (
+            get_device_map(len(self.h), range(torch.cuda.device_count())) if device_map is None else device_map
+        )
+        assert_device_map(self.device_map, len(self.h))
+        self.model_parallel = True
+        self.first_device = "cpu" if "cpu" in self.device_map.keys() else "cuda:" + str(min(self.device_map.keys()))
+        self.last_device = "cuda:" + str(max(self.device_map.keys()))
+        self.wte = self.wte.to(self.first_device)
+        self.wpe = self.wpe.to(self.first_device)
+        # Load onto devices
+        for k, v in self.device_map.items():
+            for block in v:
+                cuda_device = "cuda:" + str(k)
+                self.h[block] = self.h[block].to(cuda_device)
+        # ln_f to last
+        self.ln_f = self.ln_f.to(self.last_device)
+
+    @add_start_docstrings(DEPARALLELIZE_DOCSTRING)
+    def deparallelize(self):
+        self.model_parallel = False
+        self.device_map = None
+        self.first_device = "cpu"
+        self.last_device = "cpu"
+        self.wte = self.wte.to("cpu")
+        self.wpe = self.wpe.to("cpu")
+        for index in range(len(self.h)):
+            self.h[index] = self.h[index].to("cpu")
+        self.ln_f = self.ln_f.to("cpu")
+        torch.cuda.empty_cache()
+
+    def get_input_embeddings(self):
+        return self.wte
+
+    def set_input_embeddings(self, new_embeddings):
+        self.wte = new_embeddings
+
+    def _prune_heads(self, heads_to_prune):
+        """
+        Prunes heads of the model. heads_to_prune: dict of {layer_num: list of heads to prune in this layer}
+        """
+        for layer, heads in heads_to_prune.items():
+            self.h[layer].attn.prune_heads(heads)
+
+    @add_start_docstrings_to_model_forward(GPT2_INPUTS_DOCSTRING)
+    @add_code_sample_docstrings(
+        processor_class=_TOKENIZER_FOR_DOC,
+        checkpoint=_CHECKPOINT_FOR_DOC,
+        output_type=BaseModelOutputWithPastAndCrossAttentions,
+        config_class=_CONFIG_FOR_DOC,
+    )
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        token_type_ids: Optional[torch.LongTensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        encoder_hidden_states: Optional[torch.Tensor] = None,
+        encoder_attention_mask: 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, BaseModelOutputWithPastAndCrossAttentions]:
+        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        if input_ids is not None and inputs_embeds is not None:
+            raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
+        elif input_ids is not None:
+            input_shape = input_ids.size()
+            input_ids = input_ids.view(-1, input_shape[-1])
+            batch_size = input_ids.shape[0]
+        elif inputs_embeds is not None:
+            input_shape = inputs_embeds.size()[:-1]
+            batch_size = inputs_embeds.shape[0]
+        else:
+            raise ValueError("You have to specify either input_ids or inputs_embeds")
+
+        device = input_ids.device if input_ids is not None else inputs_embeds.device
+
+        if token_type_ids is not None:
+            token_type_ids = token_type_ids.view(-1, input_shape[-1])
+        if position_ids is not None:
+            position_ids = position_ids.view(-1, input_shape[-1])
+
+        if past_key_values is None:
+            past_length = 0
+            past_key_values = tuple([None] * len(self.h))
+        else:
+            past_length = past_key_values[0][0].size(-2)
+        if position_ids is None:
+            position_ids = torch.arange(past_length, input_shape[-1] + past_length, dtype=torch.long, device=device)
+            position_ids = position_ids.unsqueeze(0).view(-1, input_shape[-1])
+
+        # GPT2Attention mask.
+        if attention_mask is not None:
+            if batch_size <= 0:
+                raise ValueError("batch_size has to be defined and > 0")
+            attention_mask = attention_mask.view(batch_size, -1)
+            # We create a 3D attention mask from a 2D tensor mask.
+            # Sizes are [batch_size, 1, 1, to_seq_length]
+            # So we can broadcast to [batch_size, num_heads, from_seq_length, to_seq_length]
+            # this attention mask is more simple than the triangular masking of causal attention
+            # used in OpenAI GPT, we just need to prepare the broadcast dimension here.
+            attention_mask = attention_mask[:, None, None, :]
+
+            # Since attention_mask is 1.0 for positions we want to attend and 0.0 for
+            # masked positions, this operation will create a tensor which is 0.0 for
+            # positions we want to attend and the dtype's smallest value for masked positions.
+            # Since we are adding it to the raw scores before the softmax, this is
+            # effectively the same as removing these entirely.
+            attention_mask = attention_mask.to(dtype=self.dtype)  # fp16 compatibility
+            attention_mask = (1.0 - attention_mask) * torch.finfo(self.dtype).min
+
+        # If a 2D or 3D attention mask is provided for the cross-attention
+        # we need to make broadcastable to [batch_size, num_heads, seq_length, seq_length]
+        if self.config.add_cross_attention and encoder_hidden_states is not None:
+            encoder_batch_size, encoder_sequence_length, _ = encoder_hidden_states.size()
+            encoder_hidden_shape = (encoder_batch_size, encoder_sequence_length)
+            if encoder_attention_mask is None:
+                encoder_attention_mask = torch.ones(encoder_hidden_shape, device=device)
+            encoder_attention_mask = self.invert_attention_mask(encoder_attention_mask)
+        else:
+            encoder_attention_mask = None
+
+        # Prepare head mask if needed
+        # 1.0 in head_mask indicate we keep the head
+        # attention_probs has shape bsz x n_heads x N x N
+        # head_mask has shape n_layer x batch x n_heads x N x N
+        head_mask = self.get_head_mask(head_mask, self.config.n_layer)
+
+        if inputs_embeds is None:
+            inputs_embeds = self.wte(input_ids)
+        position_embeds = self.wpe(position_ids)
+        hidden_states = inputs_embeds + position_embeds
+
+        if token_type_ids is not None:
+            token_type_embeds = self.wte(token_type_ids)
+            hidden_states = hidden_states + token_type_embeds
+
+        hidden_states = self.drop(hidden_states)
+
+        output_shape = input_shape + (hidden_states.size(-1),)
+
+        presents = () if use_cache else None
+        all_self_attentions = () if output_attentions else None
+        all_cross_attentions = () if output_attentions and self.config.add_cross_attention else None
+        all_hidden_states = () if output_hidden_states else None
+        for i, (block, layer_past) in enumerate(zip(self.h, past_key_values)):
+
+            # Model parallel
+            if self.model_parallel:
+                torch.cuda.set_device(hidden_states.device)
+                # Ensure layer_past is on same device as hidden_states (might not be correct)
+                if layer_past is not None:
+                    layer_past = tuple(past_state.to(hidden_states.device) for past_state in layer_past)
+                # Ensure that attention_mask is always on the same device as hidden_states
+                if attention_mask is not None:
+                    attention_mask = attention_mask.to(hidden_states.device)
+                if isinstance(head_mask, torch.Tensor):
+                    head_mask = head_mask.to(hidden_states.device)
+            if output_hidden_states:
+                all_hidden_states = all_hidden_states + (hidden_states,)
+
+            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
+
+                def create_custom_forward(module):
+                    def custom_forward(*inputs):
+                        # None for past_key_value
+                        return module(*inputs, use_cache, output_attentions)
+
+                    return custom_forward
+
+                outputs = torch.utils.checkpoint.checkpoint(
+                    create_custom_forward(block),
+                    hidden_states,
+                    None,
+                    attention_mask,
+                    head_mask[i],
+                    encoder_hidden_states,
+                    encoder_attention_mask,
+                )
+            else:
+                outputs = block(
+                    hidden_states,
+                    layer_past=layer_past,
+                    attention_mask=attention_mask,
+                    head_mask=head_mask[i],
+                    encoder_hidden_states=encoder_hidden_states,
+                    encoder_attention_mask=encoder_attention_mask,
+                    use_cache=use_cache,
+                    output_attentions=output_attentions,
+                )
+
+            hidden_states = outputs[0]
+            if use_cache is True:
+                presents = presents + (outputs[1],)
+
+            if output_attentions:
+                all_self_attentions = all_self_attentions + (outputs[2 if use_cache else 1],)
+                if self.config.add_cross_attention:
+                    all_cross_attentions = all_cross_attentions + (outputs[3 if use_cache else 2],)
+
+            # Model Parallel: If it's the last layer for that device, put things on the next device
+            if self.model_parallel:
+                for k, v in self.device_map.items():
+                    if i == v[-1] and "cuda:" + str(k) != self.last_device:
+                        hidden_states = hidden_states.to("cuda:" + str(k + 1))
+
+        hidden_states = self.ln_f(hidden_states)
+
+        hidden_states = hidden_states.view(output_shape)
+        # Add last hidden state
+        if output_hidden_states:
+            all_hidden_states = all_hidden_states + (hidden_states,)
+
+        if not return_dict:
+            return tuple(
+                v
+                for v in [hidden_states, presents, all_hidden_states, all_self_attentions, all_cross_attentions]
+                if v is not None
+            )
+
+        return BaseModelOutputWithPastAndCrossAttentions(
+            last_hidden_state=hidden_states,
+            past_key_values=presents,
+            hidden_states=all_hidden_states,
+            attentions=all_self_attentions,
+            cross_attentions=all_cross_attentions,
+        )
+
+
+@add_start_docstrings(
+    """
+    The GPT2 Model transformer with a language modeling head on top (linear layer with weights tied to the input
+    embeddings).
+    """,
+    GPT2_START_DOCSTRING,
+)
+class GPT2LMHeadCustomModel(GPT2PreTrainedModel):
+    _keys_to_ignore_on_load_missing = [r"attn.masked_bias", r"attn.bias", r"lm_head.weight"]
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.transformer = GPT2Model(config)
+        self.lm_head = nn.Linear(config.n_embd, config.vocab_size, bias=False)
+
+        # Model parallel
+        self.model_parallel = False
+        self.device_map = None
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    @add_start_docstrings(PARALLELIZE_DOCSTRING)
+    def parallelize(self, device_map=None):
+        self.device_map = (
+            get_device_map(len(self.transformer.h), range(torch.cuda.device_count()))
+            if device_map is None
+            else device_map
+        )
+        assert_device_map(self.device_map, len(self.transformer.h))
+        self.transformer.parallelize(self.device_map)
+        self.lm_head = self.lm_head.to(self.transformer.first_device)
+        self.model_parallel = True
+
+    @add_start_docstrings(DEPARALLELIZE_DOCSTRING)
+    def deparallelize(self):
+        self.transformer.deparallelize()
+        self.transformer = self.transformer.to("cpu")
+        self.lm_head = self.lm_head.to("cpu")
+        self.model_parallel = False
+        torch.cuda.empty_cache()
+
+    def get_output_embeddings(self):
+        return self.lm_head
+
+    def set_output_embeddings(self, new_embeddings):
+        self.lm_head = new_embeddings
+
+    def prepare_inputs_for_generation(self, input_ids, past=None, **kwargs):
+        token_type_ids = kwargs.get("token_type_ids", None)
+        # only last token for inputs_ids if past is defined in kwargs
+        if past:
+            input_ids = input_ids[:, -1].unsqueeze(-1)
+            if token_type_ids is not None:
+                token_type_ids = token_type_ids[:, -1].unsqueeze(-1)
+
+        attention_mask = kwargs.get("attention_mask", None)
+        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:
+                position_ids = position_ids[:, -1].unsqueeze(-1)
+        else:
+            position_ids = None
+        return {
+            "input_ids": input_ids,
+            "past_key_values": past,
+            "use_cache": kwargs.get("use_cache"),
+            "position_ids": position_ids,
+            "attention_mask": attention_mask,
+            "token_type_ids": token_type_ids,
+        }
+
+    @add_start_docstrings_to_model_forward(GPT2_INPUTS_DOCSTRING)
+    @add_code_sample_docstrings(
+        processor_class=_TOKENIZER_FOR_DOC,
+        checkpoint=_CHECKPOINT_FOR_DOC,
+        output_type=CausalLMOutputWithCrossAttentions,
+        config_class=_CONFIG_FOR_DOC,
+    )
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        token_type_ids: Optional[torch.LongTensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        encoder_hidden_states: Optional[torch.Tensor] = None,
+        encoder_attention_mask: 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, CausalLMOutputWithCrossAttentions]:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Labels for language modeling. Note that the labels **are shifted** inside the model, i.e. you can set
+            `labels = input_ids` Indices are selected in `[-100, 0, ..., config.vocab_size]` All labels set to `-100`
+            are ignored (masked), the loss is only computed for labels in `[0, ..., config.vocab_size]`
+        """
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        transformer_outputs = self.transformer(
+            input_ids,
+            past_key_values=past_key_values,
+            attention_mask=attention_mask,
+            token_type_ids=token_type_ids,
+            position_ids=position_ids,
+            head_mask=head_mask,
+            inputs_embeds=inputs_embeds,
+            encoder_hidden_states=encoder_hidden_states,
+            encoder_attention_mask=encoder_attention_mask,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+        hidden_states = transformer_outputs[0]
+
+        # Set device for model parallelism
+        if self.model_parallel:
+            torch.cuda.set_device(self.transformer.first_device)
+            hidden_states = hidden_states.to(self.lm_head.weight.device)
+
+        lm_logits = self.lm_head(hidden_states)
+
+        loss = None
+        if labels is not None:
+            # Shift so that tokens < n predict n
+            shift_logits = lm_logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss()
+            loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1))
+
+        if not return_dict:
+            output = (lm_logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return CausalLMOutputWithCrossAttentions(
+            loss=loss,
+            logits=lm_logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+            cross_attentions=transformer_outputs.cross_attentions,
+        )
+
+    @staticmethod
+    def _reorder_cache(past: Tuple[Tuple[torch.Tensor]], beam_idx: torch.Tensor) -> Tuple[Tuple[torch.Tensor]]:
+        """
+        This function is used to re-order the `past_key_values` cache if [`~PreTrainedModel.beam_search`] or
+        [`~PreTrainedModel.beam_sample`] is called. This is required to match `past_key_values` with the correct
+        beam_idx at every generation step.
+        """
+        return tuple(
+            tuple(past_state.index_select(0, beam_idx.to(past_state.device)) for past_state in layer_past)
+            for layer_past in past
+        )
+

pytorch_model-00001-of-00006.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d491a6bded42bb3ade1fcef6167d7c38d93709ab72d7b2519d48cb083680f63a
+size 978625381

pytorch_model-00002-of-00006.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6f2981feff090a7da4e5a724459c3a7d24c4e0d57dad2fd3144533e6c476879
+size 960982501

pytorch_model-00003-of-00006.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe90003bed77d15cae1eeb5c3dd976e46774d0b7623eb6de0cb00cd8508b75d3
+size 994535645

pytorch_model-00004-of-00006.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1fea5e1bf6bd1a6aa79fbe2cf00011b41c396f30dc03e601bac1da6a1db45fdf
+size 994561235

pytorch_model-00005-of-00006.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0531a9cd7c51e05fb4d191a97362051b1f4f3e460ff3dfaaaad1df10f48abb3
+size 956795165

pytorch_model-00006-of-00006.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3043425123753b7f43c288367f2c2c01ab485163e542ca6d355c7fed1abd9d87
+size 890475405

pytorch_model.bin.index.json

@@ -0,0 +1,420 @@
+{
+  "metadata": {
+    "total_size": 5775835232
+  },
+  "weight_map": {
+    "lm_head.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.0.attn.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.attn.c_proj.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.attn.c_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.attn.k_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.attn.masked_bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.attn.q_proj.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.attn.q_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.attn.v_proj.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.attn.v_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.ln_1.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.ln_1.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.ln_2.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.ln_2.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.mlp.c_fc.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.mlp.c_fc.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.mlp.c_proj.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.0.mlp.c_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.attn.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.attn.c_proj.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.attn.c_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.attn.k_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.attn.masked_bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.attn.q_proj.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.attn.q_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.attn.v_proj.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.attn.v_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.ln_1.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.ln_1.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.ln_2.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.ln_2.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.mlp.c_fc.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.mlp.c_fc.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.mlp.c_proj.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.1.mlp.c_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.10.attn.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.attn.c_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.attn.c_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.attn.k_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.attn.masked_bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.attn.q_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.attn.q_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.attn.v_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.attn.v_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.ln_1.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.ln_1.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.ln_2.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.ln_2.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.mlp.c_fc.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.mlp.c_fc.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.mlp.c_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.10.mlp.c_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.attn.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.attn.c_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.attn.c_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.attn.k_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.attn.masked_bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.attn.q_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.attn.q_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.attn.v_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.attn.v_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.ln_1.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.ln_1.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.ln_2.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.ln_2.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.mlp.c_fc.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.mlp.c_fc.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.mlp.c_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.11.mlp.c_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.12.attn.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.12.attn.c_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.12.attn.c_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.12.attn.k_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.12.attn.masked_bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.12.attn.q_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.12.attn.q_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.12.attn.v_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.12.attn.v_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.12.ln_1.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.12.ln_1.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.12.ln_2.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.12.ln_2.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.12.mlp.c_fc.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.12.mlp.c_fc.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.12.mlp.c_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.12.mlp.c_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.attn.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.attn.c_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.attn.c_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.attn.k_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.attn.masked_bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.attn.q_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.attn.q_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.attn.v_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.attn.v_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.ln_1.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.ln_1.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.ln_2.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.ln_2.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.mlp.c_fc.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.mlp.c_fc.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.mlp.c_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.13.mlp.c_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.attn.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.attn.c_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.attn.c_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.attn.k_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.attn.masked_bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.attn.q_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.attn.q_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.attn.v_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.attn.v_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.ln_1.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.ln_1.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.ln_2.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.ln_2.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.mlp.c_fc.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.mlp.c_fc.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.mlp.c_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.14.mlp.c_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.attn.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.attn.c_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.attn.c_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.attn.k_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.attn.masked_bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.attn.q_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.attn.q_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.attn.v_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.attn.v_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.ln_1.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.ln_1.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.ln_2.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.ln_2.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.mlp.c_fc.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.mlp.c_fc.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.mlp.c_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.15.mlp.c_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.attn.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.attn.c_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.attn.c_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.attn.k_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.attn.masked_bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.attn.q_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.attn.q_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.attn.v_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.attn.v_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.ln_1.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.ln_1.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.ln_2.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.ln_2.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.mlp.c_fc.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.mlp.c_fc.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.mlp.c_proj.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.16.mlp.c_proj.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.17.attn.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.17.attn.c_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.17.attn.c_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.17.attn.k_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.17.attn.masked_bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.17.attn.q_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.17.attn.q_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.17.attn.v_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.17.attn.v_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.17.ln_1.bias": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.17.ln_1.weight": "pytorch_model-00004-of-00006.bin",
+    "transformer.h.17.ln_2.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.17.ln_2.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.17.mlp.c_fc.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.17.mlp.c_fc.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.17.mlp.c_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.17.mlp.c_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.attn.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.attn.c_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.attn.c_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.attn.k_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.attn.masked_bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.attn.q_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.attn.q_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.attn.v_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.attn.v_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.ln_1.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.ln_1.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.ln_2.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.ln_2.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.mlp.c_fc.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.mlp.c_fc.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.mlp.c_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.18.mlp.c_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.attn.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.attn.c_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.attn.c_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.attn.k_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.attn.masked_bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.attn.q_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.attn.q_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.attn.v_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.attn.v_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.ln_1.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.ln_1.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.ln_2.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.ln_2.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.mlp.c_fc.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.mlp.c_fc.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.mlp.c_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.19.mlp.c_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.2.attn.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.attn.c_proj.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.attn.c_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.attn.k_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.attn.masked_bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.attn.q_proj.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.attn.q_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.attn.v_proj.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.attn.v_proj.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.ln_1.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.ln_1.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.ln_2.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.ln_2.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.mlp.c_fc.bias": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.mlp.c_fc.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.h.2.mlp.c_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.2.mlp.c_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.20.attn.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.attn.c_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.attn.c_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.attn.k_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.attn.masked_bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.attn.q_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.attn.q_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.attn.v_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.attn.v_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.ln_1.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.ln_1.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.ln_2.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.ln_2.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.mlp.c_fc.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.mlp.c_fc.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.mlp.c_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.20.mlp.c_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.attn.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.attn.c_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.attn.c_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.attn.k_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.attn.masked_bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.attn.q_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.attn.q_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.attn.v_proj.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.attn.v_proj.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.ln_1.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.ln_1.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.ln_2.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.ln_2.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.mlp.c_fc.bias": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.mlp.c_fc.weight": "pytorch_model-00005-of-00006.bin",
+    "transformer.h.21.mlp.c_proj.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.21.mlp.c_proj.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.attn.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.attn.c_proj.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.attn.c_proj.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.attn.k_proj.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.attn.masked_bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.attn.q_proj.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.attn.q_proj.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.attn.v_proj.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.attn.v_proj.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.ln_1.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.ln_1.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.ln_2.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.ln_2.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.mlp.c_fc.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.mlp.c_fc.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.mlp.c_proj.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.22.mlp.c_proj.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.attn.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.attn.c_proj.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.attn.c_proj.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.attn.k_proj.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.attn.masked_bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.attn.q_proj.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.attn.q_proj.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.attn.v_proj.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.attn.v_proj.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.ln_1.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.ln_1.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.ln_2.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.ln_2.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.mlp.c_fc.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.mlp.c_fc.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.mlp.c_proj.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.23.mlp.c_proj.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.h.3.attn.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.attn.c_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.attn.c_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.attn.k_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.attn.masked_bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.attn.q_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.attn.q_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.attn.v_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.attn.v_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.ln_1.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.ln_1.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.ln_2.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.ln_2.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.mlp.c_fc.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.mlp.c_fc.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.mlp.c_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.3.mlp.c_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.attn.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.attn.c_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.attn.c_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.attn.k_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.attn.masked_bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.attn.q_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.attn.q_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.attn.v_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.attn.v_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.ln_1.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.ln_1.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.ln_2.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.ln_2.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.mlp.c_fc.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.mlp.c_fc.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.mlp.c_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.4.mlp.c_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.attn.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.attn.c_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.attn.c_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.attn.k_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.attn.masked_bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.attn.q_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.attn.q_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.attn.v_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.attn.v_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.ln_1.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.ln_1.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.ln_2.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.ln_2.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.mlp.c_fc.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.mlp.c_fc.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.mlp.c_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.5.mlp.c_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.attn.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.attn.c_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.attn.c_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.attn.k_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.attn.masked_bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.attn.q_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.attn.q_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.attn.v_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.attn.v_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.ln_1.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.ln_1.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.ln_2.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.ln_2.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.mlp.c_fc.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.mlp.c_fc.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.mlp.c_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.6.mlp.c_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.attn.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.attn.c_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.attn.c_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.attn.k_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.attn.masked_bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.attn.q_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.attn.q_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.attn.v_proj.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.attn.v_proj.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.ln_1.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.ln_1.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.ln_2.bias": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.ln_2.weight": "pytorch_model-00002-of-00006.bin",
+    "transformer.h.7.mlp.c_fc.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.7.mlp.c_fc.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.7.mlp.c_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.7.mlp.c_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.attn.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.attn.c_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.attn.c_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.attn.k_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.attn.masked_bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.attn.q_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.attn.q_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.attn.v_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.attn.v_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.ln_1.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.ln_1.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.ln_2.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.ln_2.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.mlp.c_fc.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.mlp.c_fc.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.mlp.c_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.8.mlp.c_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.attn.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.attn.c_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.attn.c_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.attn.k_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.attn.masked_bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.attn.q_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.attn.q_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.attn.v_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.attn.v_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.ln_1.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.ln_1.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.ln_2.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.ln_2.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.mlp.c_fc.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.mlp.c_fc.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.mlp.c_proj.bias": "pytorch_model-00003-of-00006.bin",
+    "transformer.h.9.mlp.c_proj.weight": "pytorch_model-00003-of-00006.bin",
+    "transformer.ln_f.bias": "pytorch_model-00006-of-00006.bin",
+    "transformer.ln_f.weight": "pytorch_model-00006-of-00006.bin",
+    "transformer.wpe.weight": "pytorch_model-00001-of-00006.bin",
+    "transformer.wte.weight": "pytorch_model-00001-of-00006.bin"
+  }
+}