File size: 1,822 Bytes
072784c fd1489d 072784c fd1489d 072784c |
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 |
from transformers import PretrainedConfig
class AriaConfig(PretrainedConfig):
model_type = "aria"
keys_to_ignore_at_inference = ["past_key_values"]
def __init__(
self,
vocab_size: int = 17731,
hidden_size: int = 1536,
num_hidden_layers: int = 16,
num_attention_heads: int = 64,
intermediate_size: int = 6144,
max_position_embeddings: int = 8192,
use_cache: bool = True,
bos_token_id: int = 0,
eos_token_id: int = 1,
tie_word_embeddings: bool = False,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = False,
**kwargs,
):
super().__init__(bos_token_id=bos_token_id, eos_token_id=eos_token_id, **kwargs)
self.vocab_size = vocab_size
self.hidden_size = hidden_size
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.intermediate_size = intermediate_size
self.max_position_embeddings = max_position_embeddings
self.use_cache = use_cache
self.tie_word_embeddings = tie_word_embeddings
self.output_attentions = output_attentions
self.output_hidden_states = output_hidden_states
self.return_dict = return_dict
if self.intermediate_size % self.hidden_size != 0:
raise ValueError(
"The intermediate size needs to be divisible by hidden size."
)
if self.hidden_size % self.num_attention_heads != 0:
raise ValueError(
"The hidden size needs to be divisible by the number of attention heads."
)
@property
def ff_mult(self):
return self.intermediate_size // self.hidden_size
__all__ = ["AriaConfig"]
|