-
![image](https://github.com/user-attachments/assets/302b1c1e-5924-4c2b-9a32-5b74516b8ec5)
![image](https://github.com/user-attachments/assets/9e979ac8-3442-4ba2-b711-ada5a01d0509)
![image](https://g…
-
Hi, thanks for such a great package!
I'm currently training a simple U-net on a binary image classification problem (in this case, identifying water in satellite imagery). I am exploring different …
-
Can you please support the IV regression for probit and logit in `feglm`? The basic idea for getting the IV estimators is to include the residuals from the first stage OLS into the second stage probit…
-
Are there any plans to support logit processors via an additional API parameter? For instance, the OpenAI API provides a `logit_bias` parameter, that is applied to the token distribution during genera…
-
```
# Exclude the last step as it does not appear in the targets.
logits = logits[0, :-1]
# Similarly, the first token cannot be predicteds.
target_tokens = input_tokens[0, 1:]
target…
-
Feature request:
How can i get logits (probabilites of each next token), during generation, just like I can do it in Open AI API (logprobs)? This feature will be helpfull for apps, that use logprobs …
-
### Feature request
The current NeuronX TGI server does not return logits probabilities for generated tokens.
### Motivation
This makes the `best_of` parameter useless (the first sequence is alwa…
-
When running `run_eval.sh` --> `eval_main.py` --> `eval_suite.py`, I get the following error:
```AttributeError: 'Tensor' object has no attribute 'logits'. Did you mean: 'logit'?```
The [`_mode…
-
We are using the visual part (ViT) of BioClip to process images. However, there is an issue with the forward method in BaseCAM.
In the following line of code:
`self.outputs = outputs = self.activati…
-
The code in `_glm_predictive_samples` always applies `torch.softmax` to the results under classification.
For numerical stability supporting `torch.log_softmax` here would be helpful. Similarly, it…