emozilla commited on
Commit
a89c6d6
1 Parent(s): 39956c6

Upload configuration_llama.py

Browse files
Files changed (1) hide show
  1. configuration_llama.py +184 -0
configuration_llama.py ADDED
@@ -0,0 +1,184 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # coding=utf-8
2
+ # Copyright 2022 EleutherAI and the HuggingFace Inc. team. All rights reserved.
3
+ #
4
+ # This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
5
+ # and OPT implementations in this library. It has been modified from its
6
+ # original forms to accommodate minor architectural differences compared
7
+ # to GPT-NeoX and OPT used by the Meta AI team that trained the model.
8
+ #
9
+ # Licensed under the Apache License, Version 2.0 (the "License");
10
+ # you may not use this file except in compliance with the License.
11
+ # You may obtain a copy of the License at
12
+ #
13
+ # http://www.apache.org/licenses/LICENSE-2.0
14
+ #
15
+ # Unless required by applicable law or agreed to in writing, software
16
+ # distributed under the License is distributed on an "AS IS" BASIS,
17
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
18
+ # See the License for the specific language governing permissions and
19
+ # limitations under the License.
20
+ """ LLaMA model configuration"""
21
+
22
+ from transformers.configuration_utils import PretrainedConfig
23
+ from transformers.utils import logging
24
+
25
+
26
+ logger = logging.get_logger(__name__)
27
+
28
+ LLAMA_PRETRAINED_CONFIG_ARCHIVE_MAP = {}
29
+
30
+
31
+ class LlamaConfig(PretrainedConfig):
32
+ r"""
33
+ This is the configuration class to store the configuration of a [`LlamaModel`]. It is used to instantiate an LLaMA
34
+ model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
35
+ defaults will yield a similar configuration to that of the LLaMA-7B.
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
+
41
+ Args:
42
+ vocab_size (`int`, *optional*, defaults to 32000):
43
+ Vocabulary size of the LLaMA model. Defines the number of different tokens that can be represented by the
44
+ `inputs_ids` passed when calling [`LlamaModel`]
45
+ hidden_size (`int`, *optional*, defaults to 4096):
46
+ Dimension of the hidden representations.
47
+ intermediate_size (`int`, *optional*, defaults to 11008):
48
+ Dimension of the MLP representations.
49
+ num_hidden_layers (`int`, *optional*, defaults to 32):
50
+ Number of hidden layers in the Transformer encoder.
51
+ num_attention_heads (`int`, *optional*, defaults to 32):
52
+ Number of attention heads for each attention layer in the Transformer encoder.
53
+ num_key_value_heads (`int`, *optional*):
54
+ This is the number of key_value heads that should be used to implement Grouped Query Attention. If
55
+ `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
56
+ `num_key_value_heads=1 the model will use Multi Query Attention (MQA) otherwise GQA is used. When
57
+ converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
58
+ by meanpooling all the original heads within that group. For more details checkout [this
59
+ paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
60
+ `num_attention_heads`.
61
+ pretraining_tp (`int`, *optional*, defaults to `1`):
62
+ Experimental feature. Tensor parallelism rank used during pretraining. Please refer to [this
63
+ document](https://huggingface.co/docs/transformers/parallelism) to understand more about it. This value is
64
+ necessary to ensure exact reproducibility of the pretraining results. Please refer to [this
65
+ issue](https://github.com/pytorch/pytorch/issues/76232).
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 2048):
69
+ The maximum sequence length that this model might ever be used with. Typically set this to something large
70
+ just in case (e.g., 512 or 1024 or 2048).
71
+ initializer_range (`float`, *optional*, defaults to 0.02):
72
+ The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
73
+ rms_norm_eps (`float`, *optional*, defaults to 1e-12):
74
+ The epsilon used by the rms normalization layers.
75
+ use_cache (`bool`, *optional*, defaults to `True`):
76
+ Whether or not the model should return the last key/values attentions (not used by all models). Only
77
+ relevant if `config.is_decoder=True`.
78
+ tie_word_embeddings(`bool`, *optional*, defaults to `False`):
79
+ Whether to tie weight embeddings
80
+ rope_scaling (`Dict`, *optional*):
81
+ Dictionary containing the scaling configuration for the RoPE embeddings. Currently supports three scaling
82
+ strategies: linear and dynamic. Their scaling factor must be an float greater than 1. The expected format
83
+ is `{"type": strategy name, "factor": scaling factor}`. When using this flag, don't update
84
+ `max_position_embeddings` to the expected new maximum. See the following thread for more information on how
85
+ these scaling strategies behave:
86
+ https://www.reddit.com/r/LocalLLaMA/comments/14mrgpr/dynamically_scaled_rope_further_increases/. This is an
87
+ experimental feature, subject to breaking API changes in future versions.
88
+ attention_bias (`bool`, defaults to `False`, *optional*, defaults to `False`):
89
+ Whether to use a bias in the query, key, value and output projection layers during self-attention.
90
+
91
+ Example:
92
+
93
+ ```python
94
+ >>> from transformers import LlamaModel, LlamaConfig
95
+
96
+ >>> # Initializing a LLaMA llama-7b style configuration
97
+ >>> configuration = LlamaConfig()
98
+
99
+ >>> # Initializing a model from the llama-7b style configuration
100
+ >>> model = LlamaModel(configuration)
101
+
102
+ >>> # Accessing the model configuration
103
+ >>> configuration = model.config
104
+ ```"""
105
+ model_type = "llama"
106
+ keys_to_ignore_at_inference = ["past_key_values"]
107
+
108
+ def __init__(
109
+ self,
110
+ vocab_size=32000,
111
+ hidden_size=4096,
112
+ intermediate_size=11008,
113
+ num_hidden_layers=32,
114
+ num_attention_heads=32,
115
+ num_key_value_heads=None,
116
+ hidden_act="silu",
117
+ max_position_embeddings=2048,
118
+ initializer_range=0.02,
119
+ rms_norm_eps=1e-6,
120
+ use_cache=True,
121
+ pad_token_id=0,
122
+ bos_token_id=1,
123
+ eos_token_id=2,
124
+ pretraining_tp=1,
125
+ tie_word_embeddings=False,
126
+ rope_theta=10000,
127
+ rope_scaling=None,
128
+ attention_bias=False,
129
+ **kwargs,
130
+ ):
131
+ self.vocab_size = vocab_size
132
+ self.max_position_embeddings = max_position_embeddings
133
+ self.hidden_size = hidden_size
134
+ self.intermediate_size = intermediate_size
135
+ self.num_hidden_layers = num_hidden_layers
136
+ self.num_attention_heads = num_attention_heads
137
+
138
+ # for backward compatibility
139
+ if num_key_value_heads is None:
140
+ num_key_value_heads = num_attention_heads
141
+
142
+ self.num_key_value_heads = num_key_value_heads
143
+ self.hidden_act = hidden_act
144
+ self.initializer_range = initializer_range
145
+ self.rms_norm_eps = rms_norm_eps
146
+ self.pretraining_tp = pretraining_tp
147
+ self.use_cache = use_cache
148
+ self.rope_theta = rope_theta
149
+ self.rope_scaling = rope_scaling
150
+ self._rope_scaling_validation()
151
+ self.attention_bias = attention_bias
152
+
153
+ super().__init__(
154
+ pad_token_id=pad_token_id,
155
+ bos_token_id=bos_token_id,
156
+ eos_token_id=eos_token_id,
157
+ tie_word_embeddings=tie_word_embeddings,
158
+ **kwargs,
159
+ )
160
+
161
+ def _rope_scaling_validation(self):
162
+ """
163
+ Validate the `rope_scaling` configuration.
164
+ """
165
+ if self.rope_scaling is None:
166
+ return
167
+
168
+ if not isinstance(self.rope_scaling, dict):
169
+ raise ValueError(
170
+ "`rope_scaling` must be a dictionary, "
171
+ f"got {self.rope_scaling}"
172
+ )
173
+ rope_scaling_type = self.rope_scaling.get("type", None)
174
+ rope_scaling_factor = self.rope_scaling.get("factor", None)
175
+ if rope_scaling_type is None or rope_scaling_type not in ["linear", "dynamic", "ntk-by-parts", "yarn", "dynamic-yarn"]:
176
+ raise ValueError(
177
+ f"`rope_scaling`'s name field must be one of ['linear', 'dynamic', 'ntk-by-parts', 'yarn', 'dynamic-yarn'], got {rope_scaling_type}"
178
+ )
179
+ if rope_scaling_factor is None or not isinstance(rope_scaling_factor, float) or rope_scaling_factor <= 1.0:
180
+ raise ValueError(f"`rope_scaling`'s factor field must be an float > 1, got {rope_scaling_factor}")
181
+ if rope_scaling_type == "yarn" or rope_scaling_type == "dynamic-yarn" or rope_scaling_type == "ntk-by-parts":
182
+ original_max_position_embeddings = self.rope_scaling.get("original_max_position_embeddings", None)
183
+ if original_max_position_embeddings is None or not isinstance(original_max_position_embeddings, int):
184
+ raise ValueError(f"`rope_scaling.original_max_position_embeddings` must be set to an int when using ntk-by-parts, yarn, or dynamic-yarn")