scaled-llama-7b-lora-16k-rp2 / configuration_llama.py
jordiclive's picture
Upload 2 files
30131af
raw
history blame contribute delete
547 Bytes
from transformers.models.llama.configuration_llama import \
LlamaConfig as LlamaConfigOriginal
class LlamaConfig(LlamaConfigOriginal):
def __init__(
self,
use_xpos=False,
position_interpolation_scale=1,
ntk_alpha=None,
transformer_engine=None,
**kwargs
):
self.use_xpos = use_xpos
self.position_interpolation_scale = position_interpolation_scale
self.transformer_engine = transformer_engine
self.ntk_alpha = ntk_alpha
super().__init__(**kwargs)