如何用wandb.log保存一个地块？

Question

我想用wandb.log来保存一个情节。他们的文档说：

    wandb.log({"chart": plt})

但这对我来说是失败的。

我得到两个错误，第一个错误(当我在尝试执行plt.show()之前不执行wand.log时)：

Traceback (most recent call last):
  File "/Applications/PyCharm.app/Contents/plugins/python/helpers/pydev/_pydevd_bundle/pydevd_exec2.py", line 3, in Exec
    exec(exp, global_vars, local_vars)
  File "<input>", line 1, in <module>
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/wandb_run.py", line 256, in wrapper
    return func(self, *args, **kwargs)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/wandb_run.py", line 222, in wrapper
    return func(self, *args, **kwargs)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/wandb_run.py", line 1548, in log
    self._log(data=data, step=step, commit=commit)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/wandb_run.py", line 1339, in _log
    self._partial_history_callback(data, step, commit)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/wandb_run.py", line 1228, in _partial_history_callback
    self._backend.interface.publish_partial_history(
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/interface/interface.py", line 541, in publish_partial_history
    data = history_dict_to_json(run, data, step=user_step, ignore_copy_err=True)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/data_types/utils.py", line 54, in history_dict_to_json
    payload[key] = val_to_json(
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/data_types/utils.py", line 82, in val_to_json
    val = Plotly.make_plot_media(val)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/data_types/plotly.py", line 48, in make_plot_media
    val = util.matplotlib_to_plotly(val)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/util.py", line 560, in matplotlib_to_plotly
    return tools.mpl_to_plotly(obj)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/plotly/tools.py", line 112, in mpl_to_plotly
    matplotlylib.Exporter(renderer).run(fig)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/plotly/matplotlylib/mplexporter/exporter.py", line 53, in run
    self.crawl_fig(fig)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/plotly/matplotlylib/mplexporter/exporter.py", line 124, in crawl_fig
    self.crawl_ax(ax)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/plotly/matplotlylib/mplexporter/exporter.py", line 146, in crawl_ax
    self.draw_collection(ax, collection)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/plotly/matplotlylib/mplexporter/exporter.py", line 289, in draw_collection
    offset_order = offset_dict[collection.get_offset_position()]
AttributeError: 'LineCollection' object has no attribute 'get_offset_position'

我得到两个错误，第二个错误(在尝试执行plt.show()之前执行wand.log时)：

Traceback (most recent call last):
  File "/Applications/PyCharm.app/Contents/plugins/python/helpers/pydev/_pydevd_bundle/pydevd_exec2.py", line 3, in Exec
    exec(exp, global_vars, local_vars)
  File "<input>", line 1, in <module>
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/wandb_run.py", line 256, in wrapper
    return func(self, *args, **kwargs)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/wandb_run.py", line 222, in wrapper
    return func(self, *args, **kwargs)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/wandb_run.py", line 1548, in log
    self._log(data=data, step=step, commit=commit)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/wandb_run.py", line 1339, in _log
    self._partial_history_callback(data, step, commit)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/wandb_run.py", line 1228, in _partial_history_callback
    self._backend.interface.publish_partial_history(
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/interface/interface.py", line 541, in publish_partial_history
    data = history_dict_to_json(run, data, step=user_step, ignore_copy_err=True)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/data_types/utils.py", line 54, in history_dict_to_json
    payload[key] = val_to_json(
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/data_types/utils.py", line 82, in val_to_json
    val = Plotly.make_plot_media(val)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/sdk/data_types/plotly.py", line 48, in make_plot_media
    val = util.matplotlib_to_plotly(val)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/wandb/util.py", line 560, in matplotlib_to_plotly
    return tools.mpl_to_plotly(obj)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/plotly/tools.py", line 112, in mpl_to_plotly
    matplotlylib.Exporter(renderer).run(fig)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/plotly/matplotlylib/mplexporter/exporter.py", line 53, in run
    self.crawl_fig(fig)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/plotly/matplotlylib/mplexporter/exporter.py", line 122, in crawl_fig
    with self.renderer.draw_figure(fig=fig, props=utils.get_figure_properties(fig)):
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/contextlib.py", line 119, in __enter__
    return next(self.gen)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/plotly/matplotlylib/mplexporter/renderers/base.py", line 45, in draw_figure
    self.open_figure(fig=fig, props=props)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/plotly/matplotlylib/renderer.py", line 90, in open_figure
    self.mpl_x_bounds, self.mpl_y_bounds = mpltools.get_axes_bounds(fig)
  File "/Users/brandomiranda/opt/anaconda3/envs/meta_learning/lib/python3.9/site-packages/plotly/matplotlylib/mpltools.py", line 265, in get_axes_bounds
    x_min, y_min, x_max, y_max = min(x_min), min(y_min), max(x_max), max(y_max)
ValueError: min() arg is an empty sequence

请注意，它们的简单示例确实有效：

import matplotlib.pyplot as plt

plt.plot([1, 2, 3, 4])
plt.ylabel("some interesting numbers")
wandb.log({"chart": plt})

对我来说。

交叉张贴：https://community.wandb.ai/t/how-does-one-save-a-plot-in-wandb-with-wandb-log/2373

Answer 1

请查看此笔记本的Log Plots部分：

图

导入wandb并登录/身份验证后，尝试执行Log Plots单元格，然后单击指向visualize-predictions项目的链接。(当您在下面粘贴的代码片段的第5行中使用等于“可视化-预测”的项目参数调用wandb.init时，将自动创建该项目。)

Log_plots部分中的代码粘贴在下面，以便搜索引擎也能找到它：

import matplotlib.pyplot as plt
%matplotlib inline

# Initialize a new run
wandb.init(project="visualize-predictions", name="plots")

# Make the plot
fibonacci = [0, 1, 1, 2, 3, 5, 8, 13, 21, 34]
fig, ax = plt.subplots()
ax.plot(fibonacci)
ax.set_ylabel("some interesting numbers")

# Log the plot
wandb.log({"plot": fig})
wandb.finish()
fig