cornzz

[BUG] Device error when running on other cuda device than cuda:0

1

### Python -VV ```shell Python 3.11.9 (main, Apr 19 2024, 16:48:06) [GCC 11.2.0] ``` ### Pip Freeze conda env export ```shell name: test channels: - defaults dependencies: - _libgcc_mutex=0.1=main -...

Fixes https://github.com/asnunes/notion-page-to-html/issues/40 Page rendering broke because the `NodeHttpGetClient.get` function could not handle redirects. Since notion seems to have changed the domain of their user content servers, the package broke if...

Add support for sha256-rsa-MGF1 signing algorithm (#328)

4

Solves #328

@iofu728 For longllmlingua this is not a 100% correct fix, especially for short prompts with longllmlingua errors occur. I hope one of the maintainers can give some feedback whether this...

[Bug]: Achieved compression rate with (Long)LLMLingua overshot

2

I was evaluating how well the (Long)LLMLingua is able to achieve the requested compression rate (focusing on the `rate` parameter, not `target_tokens`) and came to these conclusions: - For smaller...

Fix prompts smaller than iterative_size not being compressed

1

# What does this PR do? Fixes #196 Prompts smaller than `iterative_size / 3` were not compressed at all and only partially compressed if `> iterative_size / 3` and `<...

[Question]: How was edit sim for code tasks calculated?

1

@pzs19 In the longbench evaluation `fuzzywuzzy.ratio` was used to calculate edit similarity (Levenshtein distance) for the predicted code lines ([see here](https://github.com/microsoft/LLMLingua/blob/main/experiments/llmlingua2/evaluation/metrics.py#L94)). However I noticed that this function returns different results...

[Bug]: Prompts smaller than iterative_size are not compressed

1

# Describe the bug This concerns only LLMLingua / LongLLMLingua. As part of https://github.com/microsoft/LLMLingua/issues/195 I noticed prompts smaller than `iterative_size` were not being compressed. The `iterative_size` parameter is 200 by...

[Question]: Reproduce end2end latency results of LLMLingua-2

3

### Describe the issue @pzs19 I would like to reproduce and expand the end2end latency benchmark results of the LLMLingua-2 paper and was therefore wondering if you could provide more...

[LongBench] How was Edit Sim for code tasks calculated?

1

Hi, I just noticed that the `fuzzywuzzy.ratio` function, used for [edit sim calculation here](https://github.com/THUDM/LongBench/blob/main/LongBench/metrics.py#L87), calculates different scores, depending on whether `python-Levenshtein` is installed or not. Also, it does not actually...