jiapingW

In the process of using triton, it only supports index expansion but does not support index and array slicing, making it very difficult to retrieve data in triton like torch....

### Describe the bug I pip installed llmlingua in the conda environment. Its version is llmlingua-0.2.2. I created a new python file with the code: ```python from llmlingua import PromptCompressor...

AttributeError: 'Tensor' object has no attribute 'mask_mod'

3

Line 100 of generate.py file of main branch. The model_forward function is called here, but there are only 3 parameters, namely model, x, input_pos, without mask. Then the model_forward function...