| 13 | } |
| 14 | |
| 15 | PLDAModelObject::PLDAModelObject(size_t tw, size_t minCnt, size_t minDf, size_t rmTop, |
| 16 | size_t latentTopics, size_t topicsPerLabel, PyObject* alpha, float eta, |
| 17 | PyObject* seed, PyObject* corpus, PyObject* transform) |
| 18 | { |
| 19 | tomoto::PLDAArgs mArgs; |
| 20 | mArgs.numLatentTopics = latentTopics; |
| 21 | mArgs.numTopicsPerLabel = topicsPerLabel; |
| 22 | if (alpha) |
| 23 | { |
| 24 | mArgs.alpha = broadcastObj<tomoto::Float>(alpha, mArgs.k, |
| 25 | [&]() { return "`alpha` must be an instance of `float` or `List[float]` with length `k` (given " + py::repr(alpha) + ")"; } |
| 26 | ); |
| 27 | } |
| 28 | mArgs.eta = eta; |
| 29 | if (seed && seed != Py_None && !py::toCpp<size_t>(seed, mArgs.seed)) |
| 30 | { |
| 31 | throw py::ValueError{ "`seed` must be an integer or None." }; |
| 32 | } |
| 33 | |
| 34 | inst = tomoto::IPLDAModel::create((tomoto::TermWeight)tw, mArgs); |
| 35 | if (!inst) throw py::ValueError{ "unknown tw value" }; |
| 36 | isPrepared = false; |
| 37 | seedGiven = !!seed; |
| 38 | minWordCnt = minCnt; |
| 39 | minWordDf = minDf; |
| 40 | removeTopWord = rmTop; |
| 41 | |
| 42 | insertCorpus(corpus, transform); |
| 43 | } |
| 44 | |
| 45 | std::optional<size_t> PLDAModelObject::addDoc(PyObject* words, PyObject* labels, bool ignoreEmptyWords) |
| 46 | { |