Kimi-K3.1-Pro-SuperExpert-12GB / configuration_kimi_k31_superexpert.py
resuurcct1's picture shashanth-27xA9's picture
Duplicate from shashanth-27xA9/Kimi-K3.1-Pro-SuperExpert-12GB
776beef
Raw History Blame Contribute Delete
1.93 kB
from transformers.configuration_utils import PretrainedConfig
class KimiK31SuperExpertConfig(PretrainedConfig):
model_type = "kimi_k31_superexpert"
keys_to_ignore_at_loading = ["model.language_model.vision_tower"]
def __init__(
self,
vocab_size=163840,
hidden_size=7168,
intermediate_size=18432,
moe_intermediate_size=3072,
num_hidden_layers=36,
num_attention_heads=96,
num_key_value_heads=96,
hidden_act="silu",
max_position_embeddings=1048576,
initializer_range=0.02,
rms_norm_eps=1e-5,
use_cache=True,
pad_token_id=0,
bos_token_id=1,
eos_token_id=2,
tie_word_embeddings=False,
compression_method="AW-SECD",
compressed_footprint_gb=9.2,
retained_intelligence_ratio=0.948,
**kwargs
):
self.vocab_size = vocab_size
self.hidden_size = hidden_size
self.intermediate_size = intermediate_size
self.moe_intermediate_size = moe_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.max_position_embeddings = max_position_embeddings
self.initializer_range = initializer_range
self.rms_norm_eps = rms_norm_eps
self.use_cache = use_cache
self.compression_method = compression_method
self.compressed_footprint_gb = compressed_footprint_gb
self.retained_intelligence_ratio = retained_intelligence_ratio
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
)
def get_text_config(self, decoder=False):
return self