paddle tutorials e2-问答模型

This commit is contained in:
x54-729 2022-06-29 00:06:22 +08:00
parent 40b8016e98
commit 780bf77ef7
2 changed files with 1511 additions and 1 deletions

View File

@ -57,7 +57,7 @@
"source": [ "source": [
"#### 1.2 语义理解框架 ERNIE\n", "#### 1.2 语义理解框架 ERNIE\n",
"\n", "\n",
"``ERNIEEnhanced Representation from kNowledge IntEgration`` 是百度提出的基于知识增强的持续学习语义理解框架,至今已有 ``ERNIE 2.0``、``ERNIE 3.0``、``ERNIE-M``、``ERNIE-tiny`` 等多种预训练模型。``ERNIE 1.0`` 采用``Transformer Encoder`` 作为其语义表示的骨架,并改进了两种``mask`` 策略,分别为基于**短语**和**实体**(人名、组织等)的策略。在 ``ERNIE`` 中,由多个字组成的短语或者实体将作为一个统一单元,在训练的时候被统一地 ``mask`` 掉,这样可以潜在地学习到知识的依赖以及更长的语义依赖来让模型更具泛化性。\n", "``ERNIEEnhanced Representation from kNowledge IntEgration`` 是百度提出的基于知识增强的持续学习语义理解框架,至今已有 ``ERNIE 2.0``、``ERNIE 3.0``、``ERNIE-M``、``ERNIE-tiny`` 等多种预训练模型。``ERNIE 1.0`` 采用``Transformer Encoder`` 作为其语义表示的骨架,并改进了两种 ``mask`` 策略,分别为基于**短语**和**实体**(人名、组织等)的策略。在 ``ERNIE`` 中,由多个字组成的短语或者实体将作为一个统一单元,在训练的时候被统一地 ``mask`` 掉,这样可以潜在地学习到知识的依赖以及更长的语义依赖来让模型更具泛化性。\n",
"\n", "\n",
"<img src=\"./figures/paddle-ernie-1.0-masking.png\" width=\"50%\" height=\"50%\" align=\"center\"></img>\n", "<img src=\"./figures/paddle-ernie-1.0-masking.png\" width=\"50%\" height=\"50%\" align=\"center\"></img>\n",
"\n", "\n",

File diff suppressed because it is too large Load Diff