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

use fuse mlp in qwen (#9672)

Browse source
This commit is contained in:
Yishuo Wang 2023-12-13 17:20:08 +08:00 committed by GitHub
parent c7741c4e84
commit 09ca540f9b
2 changed files with 18 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
python/llm/src/bigdl/llm/transformers/convert.py
View file

@ -590,6 +590,7 @@ def _optimize_post(model, lightweight_bmm=False):
modeling_module_name = model.__class__.__module__ modeling_module_name = model.__class__.__module__
module = importlib.import_module(modeling_module_name) module = importlib.import_module(modeling_module_name)
from bigdl.llm.transformers.models.qwen import qwen_attention_forward from bigdl.llm.transformers.models.qwen import qwen_attention_forward
from bigdl.llm.transformers.models.qwen import qwen_mlp_forward
from bigdl.llm.transformers.models.chatglm2 import chatglm_rms_norm_forward from bigdl.llm.transformers.models.chatglm2 import chatglm_rms_norm_forward
convert_forward(model, convert_forward(model,
module.QWenAttention, module.QWenAttention,
@ -598,6 +599,9 @@ def _optimize_post(model, lightweight_bmm=False):
convert_forward(model, convert_forward(model,
module.RMSNorm, module.RMSNorm,
chatglm_rms_norm_forward) chatglm_rms_norm_forward)
convert_forward(model,
module.QWenMLP,
qwen_mlp_forward)
elif model.config.model_type == "aquila": elif model.config.model_type == "aquila":
modeling_module_name = model.__class__.__module__ modeling_module_name = model.__class__.__module__
module = importlib.import_module(modeling_module_name) module = importlib.import_module(modeling_module_name)

14
python/llm/src/bigdl/llm/transformers/models/qwen.py
View file

@ -210,3 +210,17 @@ def qwen_attention_forward(
outputs += (attn_weight,) outputs += (attn_weight,)
return outputs return outputs
def qwen_mlp_forward(self, x: torch.Tensor) -> torch.Tensor:
if x.shape[1] == 1 and x.dtype == torch.float32 and x.device.type == 'xpu' \
and not (self.training and x.requires_grad):
import linear_q4_0
x_2d = x.view(-1, x.shape[-1])
if not x_2d.is_contiguous():
x_2d = x_2d.contiguous()
return self.c_proj(linear_q4_0.mlp_forward_q4_0_xpu(
x_2d, self.w2.weight.data, self.w1.weight.data,
x_2d.shape[0], x_2d.shape[1], self.w2.out_len,
))
return self.c_proj(F.silu(self.w2(x)) * self.w1(x))