Beanocean

When to update the feature embedding layer ?

1

Hi, @karpathy Maybe it is not appropriate to add the feature embedding layer in `misc/net_utils.lua`( [Line 38](https://github.com/karpathy/neuraltalk2/blob/master/misc/net_utils.lua#L38) ). If the CNN part is not fine-tuned while training, the parameters of...

unmatched shapes in tf.concat()

4

In my distribution, the example works well: ``` python -m src.flownet2.test --input_a data/samples/0img0.ppm --input_b data/samples/0img1.ppm --out ./ ``` But when I tried to use my own data, there was a...

Is the [Azure openai-service](https://azure.microsoft.com/en-au/products/cognitive-services/openai-service) supported by Pandora? It may be a more accessible and cost-effective option compared to the original OpenAI API. Can I switch to the Azure backend through...

Why use -6.58 as the initialization of bias

3

https://github.com/s-gupta/visual-concepts/blob/0e223639df399dc973f8c7005ee68c228afc9784/output/v1/mil_finetune.prototxt#L184 Hi Saurabh, Recently, I am trying to export this code from Caffe to Pytorch. When tuning the model in Fully-Convolutional Network, I found a very interesting trick in the...

In the deformable attention, the spatial shapes represented by height (H) and width (W). Why is it necessary to exchange the `offset_normalizer` to the witdh(W) height(H) order here? https://github.com/open-mmlab/mmcv/blob/c46684c30f988607d5fb4e863841118d00ed8702/mmcv/ops/multi_scale_deform_attn.py#L352-L353 _Originally...