学习pycm3遇到问题，这个错误如何修改呢？

KeyError                                  Traceback (most recent call last)
<ipython-input-7-7587d3b8778f> in <module>
----> 1 p = trace.get_values('p', burn=niter//2, combine=True, chains=[0,2])
      2 p.shape

~\anaconda3\lib\site-packages\pymc3\backends\base.py in get_values(self, varname, burn, thin, combine, chains, squeeze)
    485         varname = get_var_name(varname)
    486         try:
--> 487             results = [self._straces[chain].get_values(varname, burn, thin) for chain in chains]
    488         except TypeError:  # Single chain passed.
    489             results = [self._straces[chains].get_values(varname, burn, thin)]

~\anaconda3\lib\site-packages\pymc3\backends\base.py in <listcomp>(.0)
    485         varname = get_var_name(varname)
    486         try:
--> 487             results = [self._straces[chain].get_values(varname, burn, thin) for chain in chains]
    488         except TypeError:  # Single chain passed.
    489             results = [self._straces[chains].get_values(varname, burn, thin)]

错误信息“KeyError：'p'”表明变量'p'在跟踪中不存在。

要解决此问题，您应该检查模型是否正确定义和采样，以及变量'p'是否在模型中指定为参数。如果未将'p'定义为参数，则应将其添加到模型中并重新运行采样。如果模型未正确定义或采样，则应检查代码并尝试确定问题。