ayo/ipex-llm
1
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

fix import ipex problem (#9323)

Browse source 
* fix import ipex problem

* fix style
This commit is contained in:
Yang Wang 2023-11-01 11:31:34 +08:00 committed by GitHub
parent 9f3d4676c6
commit e1bc18f8eb
1 changed files with 10 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

18
python/llm/src/bigdl/llm/transformers/models/llama.py
View file

@ -59,21 +59,23 @@ def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
KV_CACHE_ALLOC_BLOCK_LENGTH = 256 KV_CACHE_ALLOC_BLOCK_LENGTH = 256
_ipex_version = None
def get_ipex_version(): def get_ipex_version():
if importlib.util.find_spec("intel_extension_for_pytorch") is not None: global _ipex_version
import intel_extension_for_pytorch as ipex if _ipex_version is not None:
return ipex.__version__ return _ipex_version
else:
return None
import intel_extension_for_pytorch as ipex
ipex_version = get_ipex_version() _ipex_version = ipex.__version__
return _ipex_version
def llama_rms_norm_forward(self, hidden_states): def llama_rms_norm_forward(self, hidden_states):
if hidden_states.device.type == "xpu" and not (self.training and hidden_states.requires_grad): if hidden_states.device.type == "xpu" and not (self.training and hidden_states.requires_grad):
if ipex_version == "2.0.110+xpu": if get_ipex_version() == "2.0.110+xpu":
hidden_states, _ = torch.ops.torch_ipex.rms_norm(hidden_states, hidden_states, _ = torch.ops.torch_ipex.rms_norm(hidden_states,
[self.weight.size(0)], self.weight) [self.weight.size(0)], self.weight)
else: else: