五老师
五老师
> When I use "BertForQuestionAnswering.from_pretrained" to import my pytorch_model.bin, there having a error "tarfile.ReadError: not a gzip file".So, what should I do to use my pytorch_model.bin? Thanks. Just like these:...
> I went and asked Kapa.ai, and here's the answer: > > To update an existing FAISS vector store with a new version of your document, you can follow these...
> i met the same error, did you solve it No, not yet.
> Hi @wulaoshi, > > We only use the `KMeansClassifier` and `KMeansRegressor` from `sklearn` which you can find [here](https://github.com/automl/auto-sklearn/blob/a978478f6053f2e966955347e32b20d6d35c0a61/autosklearn/pipeline/components/classification/k_nearest_neighbors.py#L25-L27) and [here](https://github.com/automl/auto-sklearn/blob/a978478f6053f2e966955347e32b20d6d35c0a61/autosklearn/pipeline/components/regression/k_nearest_neighbors.py#L27-L29). So we do pass the labels to fit if...
> So I'm not really sure what `XTrain1` represents in `fit(X0, y0, X1)` but I assume this is part of the typical setup of semi-supervised. So the answer then is...
same problem
> 你好，NCE（噪音对比估计）就是负样本采样，就是在做二分类问题，不知道这样解释有没有说清楚 谢谢，我搞清楚了。
> 为啥用用的是BCE而不是CE，BCE相当于在做多标签分类，会造成类别极度不均衡吗？正例B，负例B*2 - B > > 看InfoNCE的公式 是softmax，BCE是sigmoid 应该是CE，这里应该是写错了，我记得二分类一直用的都是CE