From b1ea5449361a02bf0064ca6d314d40ca54ccbe9e Mon Sep 17 00:00:00 2001
From: Frank <16040285+FrankYinXF@users.noreply.github.com>
Date: Sat, 8 Feb 2020 10:17:09 +1100
Subject: [PATCH] Update numpy_manual_grad.py
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

16行求导与目标函数（\min_{a, b} L(a, b) = \sum_{i=1}^n(ax_i + b - y_i)^2）不符。虽然程序可以这样，但是如果这样的相当于变相更改了learning_rate。
---
 source/_static/code/zh/basic/example/numpy_manual_grad.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/source/_static/code/zh/basic/example/numpy_manual_grad.py b/source/_static/code/zh/basic/example/numpy_manual_grad.py
index f874901e..de868967 100644
--- a/source/_static/code/zh/basic/example/numpy_manual_grad.py
+++ b/source/_static/code/zh/basic/example/numpy_manual_grad.py
@@ -13,9 +13,9 @@
 for e in range(num_epoch):
     # 手动计算损失函数关于自变量（模型参数）的梯度
     y_pred = a * X + b
-    grad_a, grad_b = (y_pred - y).dot(X), (y_pred - y).sum()
+    grad_a, grad_b = 2*(y_pred - y).dot(X), 2*(y_pred - y).sum()
 
     # 更新参数
     a, b = a - learning_rate * grad_a, b - learning_rate * grad_b
 
-print(a, b)
\ No newline at end of file
+print(a, b)