Upload 9 files

config.json

@@ -0,0 +1,28 @@
+{
+  "_name_or_path": "/home/ea/work/my_optimum_intel/optimum-intel/xverse-7b",
+  "architectures": [
+    "XverseForCausalLM"
+  ],
+  "auto_map": {
+    "AutoConfig": "configuration_xverse.XverseConfig",
+    "AutoModelForCausalLM": "modeling_xverse.XverseForCausalLM"
+  },
+  "bos_token_id": 2,
+  "eos_token_id": 3,
+  "hidden_act": "silu",
+  "hidden_size": 32,
+  "initializer_range": 0.02,
+  "intermediate_size": 86,
+  "max_position_embeddings": 64,
+  "max_tokenizer_truncation": 6144,
+  "model_type": "xverse",
+  "num_attention_heads": 2,
+  "num_hidden_layers": 2,
+  "pad_token_id": 1,
+  "rms_norm_eps": 1e-06,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.40.1",
+  "use_cache": true,
+  "vocab_size": 100534
+}

configuration_xverse.py

@@ -0,0 +1,123 @@
+# coding=utf-8
+# Copyright 2022 EleutherAI and the HuggingFace Inc. team. All rights reserved.
+#
+# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
+# and OPT implementations in this library. It has been modified from its
+# original forms to accommodate minor architectural differences compared
+# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
+#
+# 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.
+""" XVERSE model configuration"""
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+XVERSE_PRETRAINED_CONFIG_ARCHIVE_MAP = {}
+
+
+class XverseConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`XverseModel`]. It is used to instantiate an Xverse
+    model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
+    defaults will yield a similar configuration to that of the XVERSE-13B.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        vocab_size (`int`, *optional*, defaults to 100278):
+            Vocabulary size of the XVERSE model. Defines the number of different tokens that can be represented by the
+            `inputs_ids` passed when calling [`XverseModel`]
+        hidden_size (`int`, *optional*, defaults to 5120):
+            Dimension of the hidden representations.
+        intermediate_size (`int`, *optional*, defaults to 13824):
+            Dimension of the MLP representations.
+        num_hidden_layers (`int`, *optional*, defaults to 40):
+            Number of hidden layers in the Transformer encoder.
+        num_attention_heads (`int`, *optional*, defaults to 40):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
+            The non-linear activation function (function or string) in the decoder.
+        max_position_embeddings (`int`, *optional*, defaults to 8192):
+            The maximum sequence length that this model might ever be used with. Typically set this to something large
+            just in case (e.g., 512 or 1024 or 2048).
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        rms_norm_eps (`float`, *optional*, defaults to 1e-6):
+            The epsilon used by the rms normalization layers.
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should return the last key/values attentions (not used by all models). Only
+            relevant if `config.is_decoder=True`.
+        tie_word_embeddings(`bool`, *optional*, defaults to `False`):
+            Whether to tie weight embeddings
+
+        Example:
+
+    ```python
+    >>> from transformers import XverseModel, XverseConfig
+
+    >>> # Initializing a Xverse XVERSE-13B style configuration
+    >>> configuration = XverseConfig()
+
+    >>> # Initializing a model from the XVERSE-13B style configuration
+    >>> model = XverseModel(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+    model_type = "xverse"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=100278,
+        hidden_size=5120,
+        intermediate_size=13824,
+        num_hidden_layers=40,
+        num_attention_heads=40,
+        hidden_act="silu",
+        max_position_embeddings=8192,
+        max_tokenizer_truncation=8192,
+        initializer_range=0.02,
+        rms_norm_eps=1e-6,
+        use_cache=True,
+        pad_token_id=None,
+        bos_token_id=1,
+        eos_token_id=2,
+        tie_word_embeddings=False,
+        **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.hidden_act = hidden_act
+        self.initializer_range = initializer_range
+        self.rms_norm_eps = rms_norm_eps
+        self.use_cache = use_cache
+        self.max_tokenizer_truncation = max_tokenizer_truncation
+
+        super().__init__(
+            pad_token_id=pad_token_id,
+            bos_token_id=bos_token_id,
+            eos_token_id=eos_token_id,
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": 3,
+  "pad_token_id": 1,
+  "transformers_version": "4.40.1"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94e9267dcbb3f13cf163388721167ba2e989d8a0baf230cf8625af7d8044f85d
+size 25838672

modeling_xverse.py

@@ -0,0 +1,870 @@
+# coding=utf-8
+# Copyright 2022 EleutherAI and the HuggingFace Inc. team. All rights reserved.
+#
+# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
+# and OPT implementations in this library. It has been modified from its
+# original forms to accommodate minor architectural differences compared
+# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
+#
+# 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 XVERSE model."""
+import math
+from typing import List, Optional, Tuple, Union
+
+import torch
+import torch.nn.functional as F
+import torch.utils.checkpoint
+from torch import nn
+from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss
+
+from transformers.activations import ACT2FN
+from transformers.modeling_outputs import BaseModelOutputWithPast, CausalLMOutputWithPast
+from transformers.modeling_utils import PreTrainedModel
+from transformers.utils import add_start_docstrings, add_start_docstrings_to_model_forward, logging, replace_return_docstrings
+from transformers.generation.utils import GenerationConfig
+from .configuration_xverse import XverseConfig
+
+
+logger = logging.get_logger(__name__)
+
+_CONFIG_FOR_DOC = "XverseConfig"
+
+
+# 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.
+    """
+    bsz, tgt_len = input_ids_shape
+    mask = torch.full((tgt_len, tgt_len), torch.tensor(torch.finfo(dtype).min, device=device), 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(bsz, 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 `[bsz, seq_len]` to `[bsz, 1, tgt_seq_len, src_seq_len]`.
+    """
+    bsz, src_len = mask.size()
+    tgt_len = tgt_len if tgt_len is not None else src_len
+
+    expanded_mask = mask[:, None, None, :].expand(bsz, 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 XverseRMSNorm(nn.Module):
+    def __init__(self, hidden_size, eps=1e-6):
+        """
+        XverseRMSNorm is equivalent to T5LayerNorm
+        """
+        super().__init__()
+        self.weight = nn.Parameter(torch.ones(hidden_size))
+        self.variance_epsilon = eps
+
+    def forward(self, hidden_states):
+        input_dtype = hidden_states.dtype
+        variance = hidden_states.to(torch.float32).pow(2).mean(-1, keepdim=True)
+        hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon)
+
+        return (self.weight * hidden_states).to(input_dtype)
+
+
+class XverseRotaryEmbedding(torch.nn.Module):
+    def __init__(self, dim, max_position_embeddings=2048, base=10000, device=None):
+        super().__init__()
+        inv_freq = 1.0 / (base ** (torch.arange(0, dim, 2).float().to(device) / dim))
+        self.register_buffer("inv_freq", inv_freq)
+
+        # Build here to make `torch.jit.trace` work.
+        self.max_seq_len_cached = max_position_embeddings
+        t = torch.arange(self.max_seq_len_cached, device=self.inv_freq.device, dtype=self.inv_freq.dtype)
+        freqs = torch.einsum("i,j->ij", 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, :, :], persistent=False)
+        self.register_buffer("sin_cached", emb.sin()[None, None, :, :], persistent=False)
+
+    def forward(self, x, seq_len=None):
+        # x: [bs, num_attention_heads, seq_len, head_size]
+        # This `if` block is unlikely to be run after we build sin/cos in `__init__`. Keep the logic here just in case.
+        if seq_len > self.max_seq_len_cached:
+            self.max_seq_len_cached = seq_len
+            t = torch.arange(self.max_seq_len_cached, device=x.device, dtype=self.inv_freq.dtype)
+            freqs = torch.einsum("i,j->ij", 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).to(x.device)
+            self.register_buffer("cos_cached", emb.cos()[None, None, :, :], persistent=False)
+            self.register_buffer("sin_cached", emb.sin()[None, None, :, :], persistent=False)
+        return (
+            self.cos_cached[:, :, :seq_len, ...].to(dtype=x.dtype),
+            self.sin_cached[:, :, :seq_len, ...].to(dtype=x.dtype),
+        )
+
+
+def rotate_half(x):
+    """Rotates half the hidden dims of the input."""
+    x1 = x[..., : x.shape[-1] // 2]
+    x2 = x[..., x.shape[-1] // 2 :]
+    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)  # [bs, 1, seq_len, dim]
+    sin = sin[position_ids].unsqueeze(1)  # [bs, 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 XverseMLP(nn.Module):
+    def __init__(
+        self,
+        hidden_size: int,
+        intermediate_size: int,
+        hidden_act: str,
+    ):
+        super().__init__()
+        self.gate_proj = nn.Linear(hidden_size, intermediate_size, bias=False)
+        self.down_proj = nn.Linear(intermediate_size, hidden_size, bias=False)
+        self.up_proj = nn.Linear(hidden_size, intermediate_size, bias=False)
+        self.act_fn = ACT2FN[hidden_act]
+
+    def forward(self, x):
+        return self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))
+
+
+class XverseAttention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(self, config: XverseConfig):
+        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.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_heads * self.head_dim, bias=False)
+        self.v_proj = nn.Linear(self.hidden_size, self.num_heads * self.head_dim, bias=False)
+        self.o_proj = nn.Linear(self.num_heads * self.head_dim, self.hidden_size, bias=False)
+        self.rotary_emb = XverseRotaryEmbedding(self.head_dim, max_position_embeddings=self.max_position_embeddings)
+
+    def _shape(self, tensor: torch.Tensor, seq_len: int, bsz: int):
+        return tensor.view(bsz, seq_len, self.num_heads, self.head_dim).transpose(1, 2).contiguous()
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Tuple[torch.Tensor]] = None,
+        output_attentions: bool = False,
+        use_cache: 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).view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        key_states = self.k_proj(hidden_states).view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        value_states = self.v_proj(hidden_states).view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+
+        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_states, key_states, cos, sin, position_ids)
+        # [bsz, nh, t, hd]
+
+        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
+
+        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
+            attn_weights = torch.max(
+                attn_weights, torch.tensor(torch.finfo(attn_weights.dtype).min, device=attn_weights.device)
+            )
+
+        # 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()}"
+            )
+
+        attn_output = attn_output.transpose(1, 2)
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
+
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+
+
+class XverseDecoderLayer(nn.Module):
+    def __init__(self, config: XverseConfig):
+        super().__init__()
+        self.hidden_size = config.hidden_size
+        self.self_attn = XverseAttention(config=config)
+        self.mlp = XverseMLP(
+            hidden_size=self.hidden_size,
+            intermediate_size=config.intermediate_size,
+            hidden_act=config.hidden_act,
+        )
+        self.input_layernorm = XverseRMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+        self.post_attention_layernorm = XverseRMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = 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,
+    ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
+        """
+        Args:
+            hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
+            attention_mask (`torch.FloatTensor`, *optional*): attention mask of size
+                `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative 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.
+            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`).
+            past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
+        """
+
+        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
+
+
+XVERSE_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 ([`XverseConfig`]):
+            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.
+"""
+
+
+@add_start_docstrings(
+    "The bare Xverse Model outputting raw hidden-states without any specific head on top.",
+    XVERSE_START_DOCSTRING,
+)
+class XversePreTrainedModel(PreTrainedModel):
+    config_class = XverseConfig
+    base_model_prefix = "model"
+    supports_gradient_checkpointing = True
+    _no_split_modules = ["XverseDecoderLayer"]
+    _skip_keys_device_placement = "past_key_values"
+    _keys_to_ignore_on_load_unexpected = [r"decoder\.version"]
+
+    def _init_weights(self, module):
+        std = self.config.initializer_range
+        if isinstance(module, nn.Linear):
+            module.weight.data.normal_(mean=0.0, std=std)
+            if module.bias is not None:
+                module.bias.data.zero_()
+        elif isinstance(module, nn.Embedding):
+            module.weight.data.normal_(mean=0.0, std=std)
+            if module.padding_idx is not None:
+                module.weight.data[module.padding_idx].zero_()
+
+    def _set_gradient_checkpointing(self, module, value=False):
+        if isinstance(module, XverseModel):
+            module.gradient_checkpointing = value
+
+
+XVERSE_INPUTS_DOCSTRING = r"""
+    Args:
+        input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
+            Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide
+            it.
+
+            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
+            [`PreTrainedTokenizer.__call__`] for details.
+
+            [What are input IDs?](../glossary#input-ids)
+        attention_mask (`torch.Tensor` 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**.
+
+            [What are attention masks?](../glossary#attention-mask)
+
+            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
+            [`PreTrainedTokenizer.__call__`] for details.
+
+            If `past_key_values` is used, optionally only the last `decoder_input_ids` have to be input (see
+            `past_key_values`).
+
+            If you want to change padding behavior, you should read [`modeling_opt._prepare_decoder_attention_mask`]
+            and modify to your needs. See diagram 1 in [the paper](https://arxiv.org/abs/1910.13461) for more
+            information on the default strategy.
+
+            - 1 indicates the head is **not masked**,
+            - 0 indicates the head is **masked**.
+        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.n_positions - 1]`.
+
+            [What are position IDs?](../glossary#position-ids)
+        past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
+            Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of shape
+            `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of shape
+            `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`.
+
+            Contains pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention
+            blocks) that can be used (see `past_key_values` input) to speed up sequential decoding.
+
+            If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those that
+            don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of all
+            `decoder_input_ids` of shape `(batch_size, sequence_length)`.
+        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.
+        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.
+"""
+
+@add_start_docstrings(
+    "The bare Xverse Model outputting raw hidden-states without any specific head on top.",
+    XVERSE_START_DOCSTRING,
+)
+class XverseModel(XversePreTrainedModel):
+    """
+    Transformer decoder consisting of *config.num_hidden_layers* layers. Each layer is a [`XverseDecoderLayer`]
+
+    Args:
+        config: XverseConfig
+    """
+
+    def __init__(self, config: XverseConfig):
+        super().__init__(config)
+        self.padding_idx = config.pad_token_id
+        self.vocab_size = config.vocab_size
+
+        self.embed_tokens = nn.Embedding(config.vocab_size, config.hidden_size, self.padding_idx)
+        self.layers = nn.ModuleList([XverseDecoderLayer(config) for _ in range(config.num_hidden_layers)])
+        self.norm = XverseRMSNorm(config.hidden_size, eps=config.rms_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):
+        self.embed_tokens = value
+
+    # Copied from transformers.models.bart.modeling_bart.BartDecoder._prepare_decoder_attention_mask
+    def _prepare_decoder_attention_mask(self, attention_mask, input_shape, inputs_embeds, past_key_values_length):
+        # create causal mask
+        # [bsz, seq_len] -> [bsz, 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:
+            # [bsz, seq_len] -> [bsz, 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
+
+    @add_start_docstrings_to_model_forward(XVERSE_INPUTS_DOCSTRING)
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[List[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_once(
+                    "`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, output_attentions, None)
+
+                    return custom_forward
+
+                layer_outputs = torch.utils.checkpoint.checkpoint(
+                    create_custom_forward(decoder_layer),
+                    hidden_states,
+                    attention_mask,
+                    position_ids,
+                    None,
+                )
+            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 XverseForCausalLM(XversePreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+        self.model = XverseModel(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):
+        self.lm_head = new_embeddings
+
+    def set_decoder(self, decoder):
+        self.model = decoder
+
+    def get_decoder(self):
+        return self.model
+
+    @add_start_docstrings_to_model_forward(XVERSE_INPUTS_DOCSTRING)
+    @replace_return_docstrings(output_type=CausalLMOutputWithPast, config_class=_CONFIG_FOR_DOC)
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[List[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]:
+        r"""
+        Args:
+            labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+                Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
+                config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
+                (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
+
+        Returns:
+
+        Example:
+
+        ```python
+        >>> from transformers import AutoTokenizer, AutoModelForCausalLM
+
+        >>> model = AutoModelForCausalLM.from_pretrained(PATH_TO_CONVERTED_WEIGHTS, trust_remote_code=True)
+        >>> tokenizer = AutoTokenizer.from_pretrained(PATH_TO_CONVERTED_TOKENIZER)
+
+        >>> prompt = "Hey, are you conscious? Can you talk to me?"
+        >>> inputs = tokenizer(prompt, return_tensors="pt")
+
+        >>> # Generate
+        >>> generate_ids = model.generate(inputs.input_ids, max_length=30)
+        >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
+        "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
+        ```"""
+
+        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=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)
+
+        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 _build_chat_input(self, tokenizer, messages: List[dict], max_new_tokens: int=2048):
+        max_new_tokens = max_new_tokens or self.generation_config.max_new_tokens
+        max_input_tokens = self.config.max_position_embeddings - max_new_tokens
+        max_input_tokens = max(self.config.max_position_embeddings // 2, max_input_tokens)
+        max_input_tokens = min(self.config.max_tokenizer_truncation, max_input_tokens)
+
+        total_input, round_input = [], []
+        user_prompt_tokens = tokenizer.encode("Human: ", return_token_type_ids=False)
+        assist_prompt_tokens = tokenizer.encode("Assistant: ", return_token_type_ids=False)
+        assist_prompt_len = len(assist_prompt_tokens)
+
+        for i, message in enumerate(messages[::-1]):
+            if message['role'] == 'user':
+                user_content = f"{message['content']}\n\n"
+                content_tokens = user_prompt_tokens + tokenizer.encode(user_content, return_token_type_ids=False)
+                if i == 0:
+                    content_tokens = content_tokens[:max_input_tokens-assist_prompt_len]
+                    content_tokens += assist_prompt_tokens
+                round_input = content_tokens + round_input
+
+                if i != 0:
+                    if len(total_input) + len(round_input) > max_input_tokens:
+                        break
+                    else:
+                        total_input = round_input + total_input
+                else:
+                    total_input = round_input + total_input
+                    if len(total_input) >= max_input_tokens:
+                        break
+                round_input = []
+            elif message['role'] == 'assistant':
+                assist_content = f"{message['content']}"
+                content_tokens = assist_prompt_tokens + tokenizer.encode(assist_content, return_token_type_ids=False)
+                round_input = content_tokens + [self.generation_config.eos_token_id] + round_input
+            else:
+                raise ValueError(f"message role not supported yet: {message['role']}")
+        total_input = torch.LongTensor([total_input]).to(self.device)
+        return total_input
+
+    @torch.no_grad()
+    def chat(self, tokenizer, messages: List[dict], stream=False,
+             generation_config: Optional[GenerationConfig]=None):
+        generation_config = generation_config or self.generation_config
+        input_ids = self._build_chat_input(tokenizer, messages, generation_config.max_new_tokens)
+        if stream:
+            from transformers import TextIteratorStreamer
+            from threading import Thread
+            streamer = TextIteratorStreamer(tokenizer, skip_prompt=True)
+            self.__class__.generate = PreTrainedModel.generate
+
+            def stream_generator():
+                generation_kwargs = dict(inputs=input_ids, generation_config=generation_config, streamer=streamer)
+                thread = Thread(target=self.generate, kwargs=generation_kwargs)
+                thread.start()
+                for next_text in streamer:
+                    yield next_text.replace(tokenizer.eos_token, "")
+
+            return stream_generator()
+        else:
+            self.__class__.generate = PreTrainedModel.generate  # disable stream
+            outputs = self.generate(input_ids, generation_config=generation_config)
+            response = tokenizer.decode(outputs[0][len(input_ids[0]):], skip_special_tokens=True)
+            return response
+
+    def prepare_inputs_for_generation(
+        self, input_ids, past_key_values=None, attention_mask=None, inputs_embeds=None, **kwargs
+    ):
+        if past_key_values:
+            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(
+            {
+                "position_ids": position_ids,
+                "past_key_values": past_key_values,
+                "use_cache": kwargs.get("use_cache"),
+                "attention_mask": attention_mask,
+            }
+        )
+        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) for past_state in layer_past),)
+        return reordered_past
+
+    def quantize(self, bit_length: int):
+        from .quantization import QuantizationLinear
+
+        for layer in self.model.layers:
+            layer.self_attn.q_proj = QuantizationLinear(
+                bit_length=bit_length,
+                weight=layer.self_attn.q_proj.weight.to(torch.cuda.current_device()),
+                device=layer.self_attn.q_proj.weight.device,
+            )
+            layer.self_attn.k_proj = QuantizationLinear(
+                bit_length=bit_length,
+                weight=layer.self_attn.k_proj.weight.to(torch.cuda.current_device()),
+                device=layer.self_attn.k_proj.weight.device
+            )
+            layer.self_attn.v_proj = QuantizationLinear(
+                bit_length=bit_length,
+                weight=layer.self_attn.v_proj.weight.to(torch.cuda.current_device()),
+                device=layer.self_attn.v_proj.weight.device
+            )
+            layer.self_attn.o_proj = QuantizationLinear(
+                bit_length=bit_length,
+                weight=layer.self_attn.o_proj.weight.to(torch.cuda.current_device()),
+                device=layer.self_attn.o_proj.weight.device
+            )
+            layer.mlp.gate_proj = QuantizationLinear(
+                bit_length=bit_length,
+                weight=layer.mlp.gate_proj.weight.to(torch.cuda.current_device()),
+                device=layer.mlp.gate_proj.weight.device
+            )
+            layer.mlp.down_proj = QuantizationLinear(
+                bit_length=bit_length,
+                weight=layer.mlp.down_proj.weight.to(torch.cuda.current_device()),
+                device=layer.mlp.down_proj.weight.device
+            )
+            layer.mlp.up_proj = QuantizationLinear(
+                bit_length=bit_length,
+                weight=layer.mlp.up_proj.weight.to(torch.cuda.current_device()),
+                device=layer.mlp.up_proj.weight.device
+            )
+
+        return self

quantization.py

@@ -0,0 +1,124 @@
+import bz2
+import torch
+import base64
+import ctypes
+from transformers.utils import logging
+from typing import List
+
+logger = logging.get_logger(__name__)
+
+try:
+    from cpm_kernels.kernels.base import LazyKernelCModule, KernelFunction, round_up
+
+    class Kernel:
+        def __init__(self, code: bytes, function_names: List[str]):
+            self.code = code
+            self._function_names = function_names
+            self._cmodule = LazyKernelCModule(self.code)
+
+            for name in self._function_names:
+                setattr(self, name, KernelFunction(self._cmodule, name))
+
+    quantization_code = "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"
+
+    kernels = Kernel(
+        bz2.decompress(base64.b64decode(quantization_code)),
+        [
+            "weightInt8_int4",
+            "weightInt4_fp16",
+            "weightInt4_bf16"
+        ],
+    )
+except Exception as exception:
+    kernels = None
+    logger.warning("Failed to load cpm_kernels:" + str(exception))
+
+
+def quantize_int8(weight: torch.Tensor, bit_length: int):
+    weight_scale = weight.abs().max(dim=-1).values / ((2 ** (bit_length - 1)) - 1)
+    weight_scale = weight_scale.to(torch.float32)
+
+    weight = torch.round(weight.to(weight_scale.dtype) / weight_scale[:, None]).to(torch.int8)
+    return weight, weight_scale
+
+
+def compress_int4_weight(weight: torch.Tensor):
+    with torch.cuda.device(weight.device):
+        num_row, num_chan = weight.size(0), weight.size(1)
+        num_chan = num_chan // 2
+
+        int8_weight = torch.empty(num_row, num_chan, dtype=torch.int8, device="cuda")
+        stream = torch.cuda.current_stream()
+        dim_grid = (num_row, 1, 1)
+        dim_block = (min(round_up(num_chan, 32), 1024), 1, 1)
+
+        kernels.weightInt8_int4(
+            dim_grid,
+            dim_block,
+            0,
+            stream,
+            [
+                ctypes.c_void_p(weight.data_ptr()), 
+                ctypes.c_void_p(int8_weight.data_ptr()), 
+                ctypes.c_int32(num_row), 
+                ctypes.c_int32(num_chan)
+            ],
+        )
+
+        return int8_weight
+
+
+def dequantize_float(weight: torch.Tensor, weight_scale: torch.Tensor, bit_length: int, input: torch.Tensor):
+    if bit_length == 8:
+        float_weight = weight.to(input.dtype) * weight_scale.to(input.dtype)[:, None]
+        return float_weight
+    
+    assert bit_length == 4, f"unsupported bit length: {bit_length}"
+    
+    func = (
+        kernels.weightInt4_fp16 if input.dtype == torch.half else kernels.weightInt4_bf16
+    )
+    with torch.cuda.device(weight.device):
+        num_row, num_chan = weight.size(0), weight.size(1)
+
+        float_weight = torch.empty(num_row, num_chan * 2, dtype=input.dtype, device="cuda")
+        stream = torch.cuda.current_stream()
+        dim_grid = (num_row, 1, 1)
+        dim_block = (min(round_up(num_chan, 32), 1024), 1, 1)
+
+        func(
+            dim_grid,
+            dim_block,
+            0,
+            stream,
+            [
+                ctypes.c_void_p(weight.data_ptr()),
+                ctypes.c_void_p(weight_scale.data_ptr()),
+                ctypes.c_void_p(float_weight.data_ptr()),
+                ctypes.c_int32(num_row),
+                ctypes.c_int32(num_chan),
+            ],
+        )
+        return float_weight
+
+class QuantizationLinear(torch.nn.Module):
+    def __init__(self, bit_length: int, weight: torch.Tensor, device="cuda"):
+        super().__init__()
+
+        self.bit_length = bit_length
+
+        weight, weight_scale = quantize_int8(weight=weight, bit_length=bit_length)
+        if bit_length == 4:
+            weight = compress_int4_weight(weight)
+
+        self.weight = torch.nn.Parameter(weight.to(device), requires_grad=False)
+        self.weight_scale = torch.nn.Parameter(weight_scale.to(device), requires_grad=False)
+
+    def forward(self, input: torch.Tensor):
+        input_size = input.size()
+
+        input = input.contiguous().view(-1, input.size(-1))
+        original_weight = dequantize_float(self.weight, self.weight_scale, self.bit_length, input)
+
+        output = torch.matmul(input, original_weight.t())
+        return output.view(*(input_size[:-1] + (self.weight.size(0),)))

special_tokens_map.json

@@ -0,0 +1,23 @@
+{
+    "bos_token": {
+      "content": "<|startoftext|>",
+      "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
+    }
+  }

tokenizer_config.json

@@ -0,0 +1,5 @@
+{
+  "clean_up_tokenization_spaces": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "tokenizer_class": "PreTrainedTokenizerFast"
+}