Ken Luo comments

Results 6 comments of


                                            Ken Luo

vgg16 has no dropout

Waiting..

Seems that setting logprobs=100 is not useful now.

Wondering the same thing here. I'm guessing should have at least some effect.

RuntimeError: expected scalar type Half but found Float

> This error occurs when the 2 matrices you are multiplying are not of same dtype. > > Half means dtype = torch.float16 while, Float means dtype = torch.float32 >...

deconv = visualize(model, img_array, layer_name, feature_to_visualize, visualize_mode)

Same error here. I wonder if the author could be so kind to explain why `np.squeeze` and `numpy.expand_dims` were used consecutively? And It does seem some operation creates extra dimension...

yarn start worked first time for Tesco , but fails with " error: Auth failed." on subsequent tries

> This is the reason - by capturing the screenshot at the start of the assertLoginSuccess function, it indicates Tesco consider the browser to have failed some security checks. See...

The EOS and BOS token setting when contine pretraining Llama3.1

> [@init27](https://github.com/init27) Thank you for your response. I've reviewed the information provided about the special tokens: > > * ``: Specifies the start of the prompt > * ``: Indicates...