FlashRank

FlashRank copied to clipboard

Option to Use GPU, CUDA

4

I really appreciate this repository. I hope the rerank model can optionally use a GPU to fully utilize the performance increase, potentially even with multi-GPU support. Thank you.

hieuhthh

Minor typographic corrections

srimouli04

How to deploy a FlashRank Reranker model on AWS LAMBDA as specified?

2

NeevrajKB

I am having issue while initializing Reranker

1

``` ranker = Ranker() rerankrequest = RerankRequest(query=query, passages=initial_docs) reordered_docs = ranker.rerank(rerankrequest) ``` ERROR:__main__:Error processing query: [WinError 2] The system cannot find the file specified: '\\opt'

abhigrazitti

Request new reranker model cross-encoder/ms-marco-MiniLM-L-6-v2.

1

Hi everyone, Fist of all, thanks to PrithivirajDamodaran for your great works ! My question is: how can I use the model cross-encoder/ms-marco-MiniLM-L-6-v2 because it's performance is almost as good...

congdaoduy298

tokens_map = json.load(open(str(self.model_dir / "special_tokens_map.json") You have opened special token file, used it, but then forgot to close the file. when Python closes it for you when it notices that...

sudhanshu746

Hi. I would like to know if I can use customized local reranker model in FlashRank? or... is there any documentaion regarding this? I would like to know how to...

skmanzg

This update fixes the rank extraction logic in the results appending process by replacing `rank[1]` with `rank.strip("[]")` to handle rank values more effectively. Refer to the links below for the...

prabhkaran

How can I select top_n nodes that I want to be returned from reranker?

alexandruvesa

Fix rank extraction when using LLMs for reranking >10 docs

2

See the [following PR](https://github.com/PrithivirajDamodaran/FlashRank/pull/31) and [this comment](https://github.com/PrithivirajDamodaran/FlashRank/pull/31). This PR addresses the following issue when using the LLM Reranker: - I tried to use the llm based reranking for > 10...

nmohr192