-
I try to use clip text guidance instead of CFG in SD-Style, but the result seems not satisfatory, could u please help me find what's going wrong? Here is the code, I only change three lines in the fil…
-
I tested the conservation property for some LRP rules, and all results seem okay, except those for the alpha beta rule. The maximum relative errors are:
- LRP alpha=2 beta=1, ignore bias: 0.1658904…
-
Hello, I have used adaptive loss implementation on a neural network, however after training a model long enough, I am getting negative loss values. Any help/suggestion would be highly appreciated! Ple…
-
I am working on a problem that has a multivariate outcome with negative skew. It seemed like a skew MVN model could be a good choice so I've been exploring that. In some initial simulation testing, I …
-
In the original splatting [repo](https://github.com/graphdeco-inria/gaussian-splatting), spherical harmonics parameters are represented as `features_rest` of shape (n, 15, 3) and `features_dc` of shap…
-
Hi @shyamupa ,
Thanks for your attention model!!
I can get the alpha value to visualize the machine attention level for my task.
But I found a strange phenomenon about alpha value.
The following p…
-
Minor suggestion:
Using all the weights (including bias) in regularization might end up in constraining aformentioned bias for non-normilized training data.
e.g:
```
class l1_regularization():
…
-
Is it somehow possible to have the deltas of the model optimized for best fitting all targets when using Multiple-kernel ridge with scikit-learn API?
Meaning that the deltas_ output (or another outpu…
-
### Describe the issue
I’m encountering a RuntimeError: expected scalar type BFloat16 but found Float error during fine-tuning LLAVA with LoRA enabled. This error occurs when I run the model on a mac…
-
As defined in SigmaLoss, sigmaloss = -torch.log(weights) * torch.exp(-(z_vals - depths[:, None]) ** 2 / (2 * 1)) * dists, and weights = alpha * torch.cumprod(torch.cat([torch.ones((alpha.shape[0], 1))…