|
| 1 | +# coding=utf-8 |
| 2 | +# Copyright 2024 Microsoft and the HuggingFace Inc. team. All rights reserved. |
| 3 | +# |
| 4 | +# Licensed under the Apache License, Version 2.0 (the "License"); |
| 5 | +# you may not use this file except in compliance with the License. |
| 6 | +# You may obtain a copy of the License at |
| 7 | +# |
| 8 | +# http://www.apache.org/licenses/LICENSE-2.0 |
| 9 | +# |
| 10 | +# Unless required by applicable law or agreed to in writing, software |
| 11 | +# distributed under the License is distributed on an "AS IS" BASIS, |
| 12 | +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 13 | +# See the License for the specific language governing permissions and |
| 14 | +# limitations under the License. |
| 15 | + |
| 16 | +""" Phi-3 model configuration""" |
| 17 | + |
| 18 | +from ...configuration_utils import PretrainedConfig |
| 19 | +from ....utils import logging |
| 20 | + |
| 21 | + |
| 22 | +logger = logging.get_logger(__name__) |
| 23 | + |
| 24 | +PHI3_PRETRAINED_CONFIG_ARCHIVE_MAP = { |
| 25 | + "microsoft/Phi-3-mini-4k-instruct": "https://huggingface.co/microsoft/Phi-3-mini-4k-instruct/resolve/main/config.json", |
| 26 | + "microsoft/Phi-3-mini-128k-instruct": "https://huggingface.co/microsoft/Phi-3-mini-128k-instruct/resolve/main/config.json", |
| 27 | +} |
| 28 | + |
| 29 | + |
| 30 | +class Phi3Config(PretrainedConfig): |
| 31 | + r""" |
| 32 | + This is the configuration class to store the configuration of a [`Phi3Model`]. It is used to instantiate a Phi-3 |
| 33 | + model according to the specified arguments, defining the model architecture. Instantiating a configuration with the |
| 34 | + defaults will yield a similar configuration to that of the |
| 35 | + [microsoft/Phi-3-mini-4k-instruct](https://huggingface.co/microsoft/Phi-3-mini-4k-instruct). |
| 36 | +
|
| 37 | + Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the |
| 38 | + documentation from [`PretrainedConfig`] for more information. |
| 39 | +
|
| 40 | + Args: |
| 41 | + vocab_size (`int`, *optional*, defaults to 32064): |
| 42 | + Vocabulary size of the Phi-3 model. Defines the number of different tokens that can be represented by the |
| 43 | + `inputs_ids` passed when calling [`Phi3Model`]. |
| 44 | + hidden_size (`int`, *optional*, defaults to 3072): |
| 45 | + Dimension of the hidden representations. |
| 46 | + intermediate_size (`int`, *optional*, defaults to 8192): |
| 47 | + Dimension of the MLP representations. |
| 48 | + num_hidden_layers (`int`, *optional*, defaults to 32): |
| 49 | + Number of hidden layers in the Transformer decoder. |
| 50 | + num_attention_heads (`int`, *optional*, defaults to 32): |
| 51 | + Number of attention heads for each attention layer in the Transformer decoder. |
| 52 | + num_key_value_heads (`int`, *optional*): |
| 53 | + This is the number of key_value heads that should be used to implement Grouped Query Attention. If |
| 54 | + `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if |
| 55 | + `num_key_value_heads=1 the model will use Multi Query Attention (MQA) otherwise GQA is used. When |
| 56 | + converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed |
| 57 | + by meanpooling all the original heads within that group. For more details checkout [this |
| 58 | + paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to |
| 59 | + `num_attention_heads`. |
| 60 | + resid_pdrop (`float`, *optional*, defaults to 0.0): |
| 61 | + Dropout probability for mlp outputs. |
| 62 | + embd_pdrop (`int`, *optional*, defaults to 0.0): |
| 63 | + The dropout ratio for the embeddings. |
| 64 | + attention_dropout (`float`, *optional*, defaults to 0.0): |
| 65 | + The dropout ratio after computing the attention scores. |
| 66 | + hidden_act (`str` or `function`, *optional*, defaults to `"silu"`): |
| 67 | + The non-linear activation function (function or string) in the decoder. |
| 68 | + max_position_embeddings (`int`, *optional*, defaults to 4096): |
| 69 | + The maximum sequence length that this model might ever be used with. |
| 70 | + original_max_position_embeddings (`int`, *optional*, defaults to 4096): |
| 71 | + The maximum sequence length that this model was trained with. This is used to determine the size of the |
| 72 | + original RoPE embeddings when using long scaling. |
| 73 | + initializer_range (`float`, *optional*, defaults to 0.02): |
| 74 | + The standard deviation of the truncated_normal_initializer for initializing all weight matrices. |
| 75 | + rms_norm_eps (`float`, *optional*, defaults to 1e-05): |
| 76 | + The epsilon value used for the RMSNorm. |
| 77 | + use_cache (`bool`, *optional*, defaults to `True`): |
| 78 | + Whether or not the model should return the last key/values attentions (not used by all models). Only |
| 79 | + relevant if `config.is_decoder=True`. Whether to tie weight embeddings or not. |
| 80 | + tie_word_embeddings (`bool`, *optional*, defaults to `False`): |
| 81 | + Whether to tie weight embeddings |
| 82 | + rope_theta (`float`, *optional*, defaults to 10000.0): |
| 83 | + The base period of the RoPE embeddings. |
| 84 | + rope_scaling (`dict`, *optional*): |
| 85 | + The scaling strategy for the RoPE embeddings. If `None`, no scaling is applied. If a dictionary, it must |
| 86 | + contain the following keys: `type`, `short_factor` and `long_factor`. The `type` must be either `su` or `yarn` and |
| 87 | + the `short_factor` and `long_factor` must be lists of numbers with the same length as the hidden size |
| 88 | + divided by the number of attention heads divided by 2. |
| 89 | + bos_token_id (`int`, *optional*, defaults to 1): |
| 90 | + The id of the "beginning-of-sequence" token. |
| 91 | + eos_token_id (`int`, *optional*, defaults to 32000): |
| 92 | + The id of the "end-of-sequence" token. |
| 93 | + pad_token_id (`int`, *optional*, defaults to 32000): |
| 94 | + The id of the padding token. |
| 95 | + sliding_window (`int`, *optional*): |
| 96 | + Sliding window attention window size. If `None`, no sliding window is applied. |
| 97 | +
|
| 98 | + Example: |
| 99 | +
|
| 100 | + ```python |
| 101 | + >>> from transformers import Phi3Model, Phi3Config |
| 102 | +
|
| 103 | + >>> # Initializing a Phi-3 style configuration |
| 104 | + >>> configuration = Phi3Config.from_pretrained("microsoft/Phi-3-mini-4k-instruct") |
| 105 | +
|
| 106 | + >>> # Initializing a model from the configuration |
| 107 | + >>> model = Phi3Model(configuration) |
| 108 | +
|
| 109 | + >>> # Accessing the model configuration |
| 110 | + >>> configuration = model.config |
| 111 | + ```""" |
| 112 | + |
| 113 | + model_type = "phi3" |
| 114 | + keys_to_ignore_at_inference = ["past_key_values"] |
| 115 | + |
| 116 | + def __init__( |
| 117 | + self, |
| 118 | + vocab_size=32064, |
| 119 | + hidden_size=3072, |
| 120 | + intermediate_size=8192, |
| 121 | + num_hidden_layers=32, |
| 122 | + num_attention_heads=32, |
| 123 | + num_key_value_heads=None, |
| 124 | + resid_pdrop=0.0, |
| 125 | + embd_pdrop=0.0, |
| 126 | + attention_dropout=0.0, |
| 127 | + hidden_act="silu", |
| 128 | + max_position_embeddings=4096, |
| 129 | + original_max_position_embeddings=4096, |
| 130 | + initializer_range=0.02, |
| 131 | + rms_norm_eps=1e-5, |
| 132 | + use_cache=True, |
| 133 | + tie_word_embeddings=False, |
| 134 | + rope_theta=10000.0, |
| 135 | + rope_scaling=None, |
| 136 | + bos_token_id=1, |
| 137 | + eos_token_id=32000, |
| 138 | + pad_token_id=32000, |
| 139 | + sliding_window=None, |
| 140 | + **kwargs, |
| 141 | + ): |
| 142 | + self.vocab_size = vocab_size |
| 143 | + self.hidden_size = hidden_size |
| 144 | + self.intermediate_size = intermediate_size |
| 145 | + self.num_hidden_layers = num_hidden_layers |
| 146 | + self.num_attention_heads = num_attention_heads |
| 147 | + |
| 148 | + if num_key_value_heads is None: |
| 149 | + num_key_value_heads = num_attention_heads |
| 150 | + |
| 151 | + self.num_key_value_heads = num_key_value_heads |
| 152 | + self.resid_pdrop = resid_pdrop |
| 153 | + self.embd_pdrop = embd_pdrop |
| 154 | + self.attention_dropout = attention_dropout |
| 155 | + self.hidden_act = hidden_act |
| 156 | + self.max_position_embeddings = max_position_embeddings |
| 157 | + self.original_max_position_embeddings = original_max_position_embeddings |
| 158 | + self.initializer_range = initializer_range |
| 159 | + self.rms_norm_eps = rms_norm_eps |
| 160 | + self.use_cache = use_cache |
| 161 | + self.rope_theta = rope_theta |
| 162 | + self.rope_scaling = rope_scaling |
| 163 | + self._rope_scaling_validation() |
| 164 | + self.sliding_window = sliding_window |
| 165 | + |
| 166 | + super().__init__( |
| 167 | + bos_token_id=bos_token_id, |
| 168 | + eos_token_id=eos_token_id, |
| 169 | + pad_token_id=pad_token_id, |
| 170 | + tie_word_embeddings=tie_word_embeddings, |
| 171 | + **kwargs, |
| 172 | + ) |
| 173 | + |
| 174 | + def _rope_scaling_validation(self): |
| 175 | + """ |
| 176 | + Validate the `rope_scaling` configuration. |
| 177 | + """ |
| 178 | + if self.rope_scaling is None: |
| 179 | + return |
| 180 | + |
| 181 | + if not isinstance(self.rope_scaling, dict) or len(self.rope_scaling) != 3: |
| 182 | + raise ValueError( |
| 183 | + "`rope_scaling` must be a dictionary with three fields, `type`, `short_factor` and `long_factor`, " |
| 184 | + f"got {self.rope_scaling}" |
| 185 | + ) |
| 186 | + rope_scaling_type = self.rope_scaling.get("type", None) |
| 187 | + rope_scaling_short_factor = self.rope_scaling.get("short_factor", None) |
| 188 | + rope_scaling_long_factor = self.rope_scaling.get("long_factor", None) |
| 189 | + if rope_scaling_type is None or rope_scaling_type not in ["su", "yarn"]: |
| 190 | + raise ValueError(f"`rope_scaling`'s type field must be one of ['su', 'yarn'], got {rope_scaling_type}") |
| 191 | + if not ( |
| 192 | + isinstance(rope_scaling_short_factor, list) |
| 193 | + and all(isinstance(x, (int, float)) for x in rope_scaling_short_factor) |
| 194 | + ): |
| 195 | + raise ValueError( |
| 196 | + f"`rope_scaling`'s short_factor field must be a list of numbers, got {rope_scaling_short_factor}" |
| 197 | + ) |
| 198 | + if not len(rope_scaling_short_factor) == self.hidden_size // self.num_attention_heads // 2: |
| 199 | + raise ValueError( |
| 200 | + f"`rope_scaling`'s short_factor field must have length {self.hidden_size // self.num_attention_heads // 2}, got {len(rope_scaling_short_factor)}" |
| 201 | + ) |
| 202 | + if not ( |
| 203 | + isinstance(rope_scaling_long_factor, list) |
| 204 | + and all(isinstance(x, (int, float)) for x in rope_scaling_long_factor) |
| 205 | + ): |
| 206 | + raise ValueError( |
| 207 | + f"`rope_scaling`'s long_factor field must be a list of numbers, got {rope_scaling_long_factor}" |
| 208 | + ) |
| 209 | + if not len(rope_scaling_long_factor) == self.hidden_size // self.num_attention_heads // 2: |
| 210 | + raise ValueError( |
| 211 | + f"`rope_scaling`'s long_factor field must have length {self.hidden_size // self.num_attention_heads // 2}, got {len(rope_scaling_long_factor)}" |
| 212 | + ) |
| 213 | + |
| 214 | +__all__ = ['Phi3Config'] |
0 commit comments