|
|
|
|
|
from transformers.configuration_utils import PretrainedConfig |
|
|
|
|
|
class BaichuanConfig(PretrainedConfig): |
|
model_type = "baichuan" |
|
keys_to_ignore_at_inference = ["past_key_values"] |
|
|
|
def __init__( |
|
self, |
|
vocab_size=64000, |
|
hidden_size=5120, |
|
intermediate_size=13696, |
|
num_hidden_layers=40, |
|
num_attention_heads=40, |
|
hidden_act="silu", |
|
model_max_length=4096, |
|
initializer_range=0.02, |
|
rms_norm_eps=1e-6, |
|
use_cache=True, |
|
pad_token_id=0, |
|
bos_token_id=1, |
|
eos_token_id=2, |
|
tie_word_embeddings=False, |
|
gradient_checkpointing=False, |
|
z_loss_weight=0, |
|
**kwargs, |
|
): |
|
self.vocab_size = vocab_size |
|
self.model_max_length = model_max_length |
|
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.hidden_act = hidden_act |
|
self.initializer_range = initializer_range |
|
self.rms_norm_eps = rms_norm_eps |
|
self.use_cache = use_cache |
|
self.z_loss_weight = z_loss_weight |
|
self.gradient_checkpointing = (gradient_checkpointing,) |
|
super().__init__( |
|
pad_token_id=pad_token_id, |
|
bos_token_id=bos_token_id, |
|
eos_token_id=eos_token_id, |
|
tie_word_embeddings=tie_word_embeddings, |
|
**kwargs, |
|
) |
|
|
