You need to enable JavaScript to run this app.
优惠活动
大模型
产品
解决方案
定价
更多

Keras训练history绘图未默认使用ggplot2的问题求助

Fixing Keras History Plotting with ggplot2 in RStudio

Hey there! Let's get your Keras training history plots looking sharp with ggplot2. Here's what you need to know and fix:

1. You're Using the Wrong Parameter

First off, the method="ggplot2" argument you tried doesn't exist in the R Keras package's plot() function for history objects—that's why it didn't work! The correct parameter to trigger ggplot2 is use_ggplot = TRUE.

Before plotting, make sure you've loaded the ggplot2 package (installing it isn't enough—you need to load it into your session):

library(ggplot2)
library(keras)

# After training your model and getting the history object
plot(history, use_ggplot = TRUE)

2. If the Built-in Plot Still Doesn't Work, Manually Build the Plot

Sometimes, depending on your Keras version or environment quirks, the built-in ggplot integration might not behave as expected. No worries—you can easily extract the history data and build a custom ggplot from scratch, which gives you full control over the aesthetics.

Here's how:

# Convert history metrics to a data frame
history_data <- as.data.frame(history$metrics)
# Add an epoch column for x-axis
history_data$epoch <- 1:nrow(history_data)

# Plot training vs validation loss
ggplot(history_data, aes(x = epoch)) +
  geom_line(aes(y = loss, color = "Training Loss"), linewidth = 1) +
  geom_line(aes(y = val_loss, color = "Validation Loss"), linewidth = 1) +
  labs(title = "Model Loss Over Epochs",
       x = "Epoch",
       y = "Loss Value",
       color = "Metric") +
  scale_color_brewer(palette = "Set1") +
  theme_minimal()

# Plot training vs validation accuracy (adjust metric names if you used different ones)
ggplot(history_data, aes(x = epoch)) +
  geom_line(aes(y = accuracy, color = "Training Accuracy"), linewidth = 1) +
  geom_line(aes(y = val_accuracy, color = "Validation Accuracy"), linewidth = 1) +
  labs(title = "Model Accuracy Over Epochs",
       x = "Epoch",
       y = "Accuracy",
       color = "Metric") +
  scale_color_brewer(palette = "Set1") +
  theme_minimal()

3. Check Your Keras Package Version

Older versions of the R Keras package might not support the use_ggplot parameter. Verify your version with:

packageVersion("keras")

If it's outdated, update it using:

install.packages("keras")

That should get you the polished ggplot2-style plots you're looking for!

内容的提问来源于stack exchange,提问作者Turex

相关产品推荐
方舟 Agent Plan

超全模态模型 × Harness 升级,最新支持 Deepseek-V4.1-Flash、GLM-5.3 系列、Doubao-Seedream-5.0-pro、Kimi-K3 (部分), 限时 9.9 元起

最近更新时间:2026.04.27 19:37:43