Skip to content

A
Annotated Deep Learning Paper Implementations

Greenplum / Annotated Deep Learning Paper Implementations
11 个月 前同步成功

通知 6
Star 0
Fork 0
A
Annotated Deep Learning Paper Implementations

前往新版Gitcode，体验更适合开发者的 AI 搜索 >>

提交 63022e66 编写于 作者: V Varuna Jayasiri
浏览文件
操作

fix

上级 d7af20a6
隐藏空白更改
内联 并排
Showing with 12 addition and 12 deletion
labml_nn/neox/model.py
浏览文件 @ 63022e66
......@@ -521,18 +521,18 @@ class LayerGenerator:
#
with monit.section('Covert to int8'):
layer.attention.output = make_llm_int8_linear(layer.attention.output,
device=self.device,
threshold=self.llm_int8_threshold)
layer.attention.qkv_lin = make_llm_int8_linear(layer.attention.qkv_lin,
device=self.device,
threshold=self.llm_int8_threshold)
layer.ffn.dense_h_h4 = make_llm_int8_linear(layer.ffn.dense_h_h4,
device=self.device,
threshold=self.llm_int8_threshold)
layer.ffn.dense_h4_h = make_llm_int8_linear(layer.ffn.dense_h4_h,
device=self.device,
threshold=self.llm_int8_threshold)
layer.attention._modules['output'] = make_llm_int8_linear(layer.attention.output,
device=self.device,
threshold=self.llm_int8_threshold)
layer.attention._modules['qkv_lin'] = make_llm_int8_linear(layer.attention.qkv_lin,
device=self.device,
threshold=self.llm_int8_threshold)
layer.ffn._modules['dense_h_h4'] = make_llm_int8_linear(layer.ffn.dense_h_h4,
device=self.device,
threshold=self.llm_int8_threshold)
layer.ffn._modules['dense_h4_h'] = make_llm_int8_linear(layer.ffn.dense_h4_h,
device=self.device,
threshold=self.llm_int8_threshold)
#
return layer
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或