Method train

src/python/handler/py_LDA.cpp:277–320

}

void LDAModelObject::train(size_t iteration, size_t workers, size_t ps, bool freezeTopics, size_t callbackInterval, PyObject* callback)
{
	if (seedGiven && workers != 1 && PyErr_WarnEx(PyExc_RuntimeWarning, "The training result may differ even with fixed seed if `workers` != 1.", 1)) throw py::ExcPropagation{};
	auto* inst = getInst<tomoto::ILDAModel>();
	if (!isPrepared)
	{
		inst->prepare(true, minWordCnt, minWordDf, removeTopWord);
		isPrepared = true;
	}

	if (callback == Py_None) callback = nullptr;
	if (callback && !PyCallable_Check(callback)) throw py::ValueError{ "`callback` should be a callable object" };
	if (!callback || callbackInterval <= 0)
	{
		callbackInterval = iteration;
	}

	for (size_t it = 0; it < iteration; it += callbackInterval)
	{
		if (callback)
		{
			py::UniqueObj args{ py::buildPyTuple(getObject(), it, iteration)};
			if (callback)
			{
				py::UniqueObj ret{ PyObject_CallObject(callback, args.get()) };
				if (!ret) throw py::ExcPropagation{};
			}
		}

		if (inst->train(std::min(callbackInterval, iteration - it), workers, (tomoto::ParallelScheme)ps, !!fixed) < 0)
		{
			throw py::RuntimeError{ "Train failed" };
		}
	}
	if (callback)
	{
		py::UniqueObj args{ py::buildPyTuple(getObject(), iteration, iteration) };
		if (callback)
		{
			py::UniqueObj ret{ PyObject_CallObject(callback, args.get()) };
			if (!ret) throw py::ExcPropagation{};
		}
	}
}

py::UniqueObj LDAModelObject::getTopicWords(size_t topicId, size_t topN, bool returnId) const
{