Update modeling_codeshell.py

call the "quantize_online" method.

modeling_codeshell.py

@@ -809,10 +809,10 @@ class CodeShellForCausalLM(CodeShellPreTrainedModel):
  def quantize(self, bits: int):
  try:
  import bitsandbytes
- from .quantizer import quantize
+ from .quantizer import quantize_online
  except ImportError:
  raise ImportError(f"Needs bitsandbytes to run quantize.")
- return quantize(self, bits)
+ return quantize_online(self, bits)
 
  def get_output_embeddings(self):
  return self.lm_head