File size: 2,947 Bytes
8cc7157 51cf4ed 8cc7157 51cf4ed 8cc7157 a85fd6b 8cc7157 2fa9200 8cc7157 2d2e45d 8cc7157 51cf4ed 8cc7157 51cf4ed 8cc7157 a85fd6b 51cf4ed a85fd6b 8cc7157 a85fd6b 51cf4ed 2d2e45d a85fd6b 51cf4ed | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 | # ==================== configuration_neollm.py ====================
from transformers.configuration_utils import PretrainedConfig
from transformers.modeling_rope_utils import rope_config_validation
from transformers.utils import logging
logger = logging.get_logger(__name__)
class NeoLLMConfig(PretrainedConfig):
r"""
This is the configuration class to store the configuration of a `NeoLLMModel`. It is used to instantiate a
NeoLLM model according to the specified arguments, defining the model architecture.
Configuration objects inherit from `PretrainedConfig` and can be used to control the model outputs.
"""
model_type = "neollm"
keys_to_ignore_at_inference = []
def __init__(
self,
vocab_size=200005,
hidden_size=512,
intermediate_size=1536,
num_hidden_layers=12,
num_attention_heads=8,
num_key_value_heads=2,
hidden_act="xielu",
max_position_embeddings=32768,
initializer_range=0.02,
rms_norm_eps=1e-6,
tie_word_embeddings=True,
rope_theta=10000.0,
rope_scaling=None,
partial_rotary_factor=0.25,
attention_bias=False,
attention_dropout=0.1,
head_dim=64,
fan_ratio=0.125,
fan_ratio_ffn=0.0625,
dropout_rate=0.1,
use_stack=True,
num_stack_heads=4,
stack_slots=24,
stack_d_model=16,
**kwargs,
):
super().__init__(tie_word_embeddings=tie_word_embeddings, **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.num_key_value_heads = num_key_value_heads
self.hidden_act = hidden_act
self.initializer_range = initializer_range
self.rms_norm_eps = rms_norm_eps
self.rope_theta = rope_theta
self.rope_scaling = rope_scaling
self.partial_rotary_factor = partial_rotary_factor
self.attention_bias = attention_bias
self.attention_dropout = attention_dropout
self.head_dim = head_dim
rope_config_validation(self)
# FANformer parameters
self.fan_ratio = fan_ratio
self.fan_ratio_ffn = fan_ratio_ffn
self.dropout_rate = dropout_rate
# StackMemory parameters
self.use_stack = use_stack
self.num_stack_heads = num_stack_heads
self.stack_slots = stack_slots
self.stack_d_model = stack_d_model
self.auto_map = {
"AutoConfig": "configuration_neollm.NeoLLMConfig",
"AutoModel": "modeling_neollm.NeoLLMModel",
"AutoModelForCausalLM": "modeling_neollm.NeoLLMForCausalLM"
}
__all__ = ["NeoLLMConfig"] |