Upload pose_configuration_llama.py with huggingface_hub
Browse files- pose_configuration_llama.py +177 -0
pose_configuration_llama.py
ADDED
@@ -0,0 +1,177 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
# Modification Copyright 2023 Dawei Zhu
|
3 |
+
# Copyright 2022 EleutherAI and the HuggingFace Inc. team. All rights reserved.
|
4 |
+
#
|
5 |
+
# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
|
6 |
+
# and OPT implementations in this library. It has been modified from its
|
7 |
+
# original forms to accommodate minor architectural differences compared
|
8 |
+
# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
|
9 |
+
#
|
10 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
11 |
+
# you may not use this file except in compliance with the License.
|
12 |
+
# You may obtain a copy of the License at
|
13 |
+
#
|
14 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
15 |
+
#
|
16 |
+
# Unless required by applicable law or agreed to in writing, software
|
17 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
18 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
19 |
+
# See the License for the specific language governing permissions and
|
20 |
+
# limitations under the License.
|
21 |
+
""" LLaMA model configuration"""
|
22 |
+
|
23 |
+
# from ...configuration_utils import PretrainedConfig
|
24 |
+
# from ...utils import logging
|
25 |
+
from transformers.configuration_utils import PretrainedConfig
|
26 |
+
from transformers.utils import logging
|
27 |
+
|
28 |
+
|
29 |
+
logger = logging.get_logger(__name__)
|
30 |
+
|
31 |
+
LLAMA_PRETRAINED_CONFIG_ARCHIVE_MAP = {}
|
32 |
+
|
33 |
+
|
34 |
+
class LlamaConfig(PretrainedConfig):
|
35 |
+
r"""
|
36 |
+
This is the configuration class to store the configuration of a [`LlamaModel`]. It is used to instantiate an LLaMA
|
37 |
+
model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
|
38 |
+
defaults will yield a similar configuration to that of the LLaMA-7B.
|
39 |
+
|
40 |
+
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
|
41 |
+
documentation from [`PretrainedConfig`] for more information.
|
42 |
+
|
43 |
+
|
44 |
+
Args:
|
45 |
+
vocab_size (`int`, *optional*, defaults to 32000):
|
46 |
+
Vocabulary size of the LLaMA model. Defines the number of different tokens that can be represented by the
|
47 |
+
`inputs_ids` passed when calling [`LlamaModel`]
|
48 |
+
hidden_size (`int`, *optional*, defaults to 4096):
|
49 |
+
Dimension of the hidden representations.
|
50 |
+
intermediate_size (`int`, *optional*, defaults to 11008):
|
51 |
+
Dimension of the MLP representations.
|
52 |
+
num_hidden_layers (`int`, *optional*, defaults to 32):
|
53 |
+
Number of hidden layers in the Transformer encoder.
|
54 |
+
num_attention_heads (`int`, *optional*, defaults to 32):
|
55 |
+
Number of attention heads for each attention layer in the Transformer encoder.
|
56 |
+
num_key_value_heads (`int`, *optional*):
|
57 |
+
This is the number of key_value heads that should be used to implement Grouped Query Attention. If
|
58 |
+
`num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
|
59 |
+
`num_key_value_heads=1 the model will use Multi Query Attention (MQA) otherwise GQA is used. When
|
60 |
+
converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
|
61 |
+
by meanpooling all the original heads within that group. For more details checkout [this
|
62 |
+
paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
|
63 |
+
`num_attention_heads`.
|
64 |
+
pretraining_tp (`int`, *optional*, defaults to `1`):
|
65 |
+
Experimental feature. Tensor parallelism rank used during pretraining. Please refer to [this
|
66 |
+
document](https://huggingface.co/docs/transformers/parallelism) to understand more about it. This value is
|
67 |
+
necessary to ensure exact reproducibility of the pretraining results. Please refer to [this
|
68 |
+
issue](https://github.com/pytorch/pytorch/issues/76232).
|
69 |
+
hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
|
70 |
+
The non-linear activation function (function or string) in the decoder.
|
71 |
+
max_position_embeddings (`int`, *optional*, defaults to 2048):
|
72 |
+
The maximum sequence length that this model might ever be used with. Typically set this to something large
|
73 |
+
just in case (e.g., 512 or 1024 or 2048).
|
74 |
+
initializer_range (`float`, *optional*, defaults to 0.02):
|
75 |
+
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
|
76 |
+
rms_norm_eps (`float`, *optional*, defaults to 1e-12):
|
77 |
+
The epsilon used by the rms normalization layers.
|
78 |
+
use_cache (`bool`, *optional*, defaults to `True`):
|
79 |
+
Whether or not the model should return the last key/values attentions (not used by all models). Only
|
80 |
+
relevant if `config.is_decoder=True`.
|
81 |
+
tie_word_embeddings(`bool`, *optional*, defaults to `False`):
|
82 |
+
Whether to tie weight embeddings
|
83 |
+
rope_scaling (`Dict`, *optional*):
|
84 |
+
Dictionary containing the scaling configuration for the RoPE embeddings. Currently supports three scaling
|
85 |
+
strategies: linear and dynamic. Their scaling factor must be an float greater than 1. The expected format
|
86 |
+
is `{"type": strategy name, "factor": scaling factor}`. When using this flag, don't update
|
87 |
+
`max_position_embeddings` to the expected new maximum. See the following thread for more information on how
|
88 |
+
these scaling strategies behave:
|
89 |
+
https://www.reddit.com/r/LocalLLaMA/comments/14mrgpr/dynamically_scaled_rope_further_increases/. This is an
|
90 |
+
experimental feature, subject to breaking API changes in future versions.
|
91 |
+
|
92 |
+
Example:
|
93 |
+
|
94 |
+
```python
|
95 |
+
>>> from transformers import LlamaModel, LlamaConfig
|
96 |
+
|
97 |
+
>>> # Initializing a LLaMA llama-7b style configuration
|
98 |
+
>>> configuration = LlamaConfig()
|
99 |
+
|
100 |
+
>>> # Initializing a model from the llama-7b style configuration
|
101 |
+
>>> model = LlamaModel(configuration)
|
102 |
+
|
103 |
+
>>> # Accessing the model configuration
|
104 |
+
>>> configuration = model.config
|
105 |
+
```"""
|
106 |
+
model_type = "llama"
|
107 |
+
keys_to_ignore_at_inference = ["past_key_values"]
|
108 |
+
|
109 |
+
def __init__(
|
110 |
+
self,
|
111 |
+
vocab_size=32000,
|
112 |
+
hidden_size=4096,
|
113 |
+
intermediate_size=11008,
|
114 |
+
num_hidden_layers=32,
|
115 |
+
num_attention_heads=32,
|
116 |
+
num_key_value_heads=None,
|
117 |
+
hidden_act="silu",
|
118 |
+
max_position_embeddings=2048,
|
119 |
+
initializer_range=0.02,
|
120 |
+
rms_norm_eps=1e-6,
|
121 |
+
use_cache=True,
|
122 |
+
pad_token_id=0,
|
123 |
+
bos_token_id=1,
|
124 |
+
eos_token_id=2,
|
125 |
+
pretraining_tp=1,
|
126 |
+
tie_word_embeddings=False,
|
127 |
+
rope_scaling=None,
|
128 |
+
**kwargs,
|
129 |
+
):
|
130 |
+
self.vocab_size = vocab_size
|
131 |
+
self.max_position_embeddings = max_position_embeddings
|
132 |
+
self.hidden_size = hidden_size
|
133 |
+
self.intermediate_size = intermediate_size
|
134 |
+
self.num_hidden_layers = num_hidden_layers
|
135 |
+
self.num_attention_heads = num_attention_heads
|
136 |
+
|
137 |
+
# for backward compatibility
|
138 |
+
if num_key_value_heads is None:
|
139 |
+
num_key_value_heads = num_attention_heads
|
140 |
+
|
141 |
+
self.num_key_value_heads = num_key_value_heads
|
142 |
+
self.hidden_act = hidden_act
|
143 |
+
self.initializer_range = initializer_range
|
144 |
+
self.rms_norm_eps = rms_norm_eps
|
145 |
+
self.pretraining_tp = pretraining_tp
|
146 |
+
self.use_cache = use_cache
|
147 |
+
self.rope_scaling = rope_scaling
|
148 |
+
self._rope_scaling_validation()
|
149 |
+
|
150 |
+
super().__init__(
|
151 |
+
pad_token_id=pad_token_id,
|
152 |
+
bos_token_id=bos_token_id,
|
153 |
+
eos_token_id=eos_token_id,
|
154 |
+
tie_word_embeddings=tie_word_embeddings,
|
155 |
+
**kwargs,
|
156 |
+
)
|
157 |
+
|
158 |
+
def _rope_scaling_validation(self):
|
159 |
+
"""
|
160 |
+
Validate the `rope_scaling` configuration.
|
161 |
+
"""
|
162 |
+
if self.rope_scaling is None:
|
163 |
+
return
|
164 |
+
|
165 |
+
if not isinstance(self.rope_scaling, dict):
|
166 |
+
raise ValueError(
|
167 |
+
"`rope_scaling` must be a dictionary with with two fields, `name` and `factor`, "
|
168 |
+
f"got {self.rope_scaling}"
|
169 |
+
)
|
170 |
+
rope_scaling_type = self.rope_scaling.get("type", None)
|
171 |
+
rope_scaling_factor = self.rope_scaling.get("factor", None)
|
172 |
+
if rope_scaling_type is None or rope_scaling_type not in ["linear", "dynamic", "vanilla_ntk", "yarn"]:
|
173 |
+
raise ValueError(
|
174 |
+
f"`rope_scaling`'s name field must be one of ['linear', 'dynamic', 'vanilla_ntk', 'yarn'], got {rope_scaling_type}"
|
175 |
+
)
|
176 |
+
if rope_scaling_factor is None or not isinstance(rope_scaling_factor, float) or rope_scaling_factor <= 1.0:
|
177 |
+
raise ValueError(f"`rope_scaling`'s factor field must be an float > 1, got {rope_scaling_factor}")
|