Huiqiang Jiang

Hi @DomStan, thanks for your support. Currently, LLMLingua-2 does not support query conditioning and dynamic ratio. We will consider adding support for these features in future versions. Thanks again for...

Hi @saucebing, thanks for your support. We'll check it.

Hi @MrTBH, it looks like out of memory. You might want to try using a smaller model like `lgaalves/gpt2-dolly` or a quantized version of the model.

Hi @amrosalehms, I suspect the issue is due to loading the tokenizer for GPT-3.5-turbo. You can comment out [this line](https://github.com/microsoft/LLMLingua/blob/main/llmlingua/prompt_compressor.py#L17) and related code to resolve the issue.

Hi @Dorish, the code invoking `self.oai_tokenizer`, such as in #712, needs to be modified.

Hi @Liangyx2, thanks for your support and the very detailed experiment information. Although we haven't tried using LongLLMLingua with Mistral as the LLM, my intuition tells me that it should...

Hi @yunlongia, Thank you for your interest and support. For the compression parameters, you can follow the guidelines at https://github.com/microsoft/LLMLingua/blob/main/Transparency_FAQ.md#how-to-reproduce-the-result-in-llmlingua--longllmlingua. For the LongChat experiments, we used the code base from...

Hi @Liangyx2, The last issue is fixed. You can update to the latest version using `pip install git+https://github.com/microsoft/LLMLingua.git`.

Hi @yunlongia, we plan to release these scripts in the upcoming weeks. Initially, you can try splitting the context using "\n\n" or another delimiter.

Hi @alexandreteles, thank you for your interest in our project. In fact, we have released the entire data collection pipeline and scripts at https://github.com/microsoft/LLMLingua/tree/main/experiments/llmlingua2/data_collection. You can define your own compressor...