Understanding of Pytorch NLLLOSS
PyTorch's negative log-likelihood loss, nn.NLLLoss is defined as:
So, if the loss is calculated with the standard weight of one in a single batch the formula for the loss is always:
-1 * (prediction of model for correct class)
Example:
Correct Class = 0
prediction of model for correct class = 0.5
loss = -1 * 0.5
So, why is it called the "negative log-likelihood loss", if there isn't a log function involved in calculating the loss?
Indeed no log is being used to compute the result of nn.NLLLoss so this can be a little confusing. However, I believe the reason why it was called this way is because it expects to receive log-probabilities:
The input given through a forward call is expected to contain log-probabilities of each class. - docs
In the end it does not make much sense to have it in the name since you might as well want to apply this function on non-log-probabilities...
- What do BatchNorm2d's running_mean / running_var mean in PyTorch?
- Map each element of torch.Tensor with it's value in the dict
- Pytroch clamp for complex values
- Can batch normalization be considered a linear transformation?
- Doing PyWavelets calculation on GPU
- What is the difference between an Embedding Layer with a bias immediately afterwards and a Linear Layer in PyTorch
- How do I display a single image in PyTorch?
- Examples or explanations of pytorch dataloaders?
- How to map values from a 3D tensor to a 1D tensor in PyTorch?
- Traceback (most recent call last) in Colab when looping through dataloader in pytorch
- Is there a way to use list of indices to simultaneously access the modules of nn.ModuleList in python?
- How to multiply 2x3x3x3 matrix by 2x3 matrix to get 2x3 matrix
- How does one set the pad token correctly (not to eos) during fine-tuning to avoid model not predicting EOS?
- HuggingFace Model - OnnxRuntime - Jupyter Notebook Print Model Summary
- The “Forward/Backward Passage Size” is too large for the pytorch model (Yolov3)
- How to solve the pytorch RuntimeError: Numpy is not available without upgrading numpy to the latest version because of other dependencies
- Torch Euclidian Norm (L2)
- Does order of transforms applied for data augmentation matter in Torchvision transforms?
- Problem in Backpropagation through a sample in Beta distribution in pytorch
- Reinforcement Learning Gymnasium ValueError
- Difference between torch.as_tensor() and torch.asarray()
- Why KL divergence is negative in Pytorch?
- Neural network learning to sum two numbers
- Forward pass with all samples
- Pytorch: how to (efficiently) apply a function without a “dim” argument to each row of a 2D tensor?
- override pytorch Dataset efficiently
- Implementation of F1-score, IOU and Dice Score
- Why bilinear scaling of images with PIL and pytorch produces different results?
- Pytorch Python Distributed Multiprocessing: Gather/Concatenate tensor arrays of different lengths/sizes
- CUDA not available