Dify is an open-source LLM app development platform. Dify's intuitive interface combines AI workflow, RAG pipeline, agent capabilities, model management, observability features and more, letting you quickly go from prototype to production.
[X] I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
[X] Pleas do not modify this template :) and fill in all the required fields.
1. Is this request related to a challenge you're experiencing?
The Cohere rearrangement model is not very effective for Chinese in certain situations, such as analyzing legal provisions, and is not as useful as BGE.
2. Describe the feature you'd like to see
Suggest dify to access the BGE RANK rearrangement model
3. How will this feature improve your workflow or experience?
In more rigorous language documents, such as internal rules and regulations of enterprises and institutional norms for social operation, a deep understanding of the language is required. Currently, the effectiveness of using Cohere through API calls is not optimistic, especially with a lack of support for Chinese, which may even lead to poor business performance and urgently need to be improved.
4. Additional context or comments
Here, as a supplement, we provide the relevant images of the bge run base:
Self Checks
1. Is this request related to a challenge you're experiencing?
The Cohere rearrangement model is not very effective for Chinese in certain situations, such as analyzing legal provisions, and is not as useful as BGE.
2. Describe the feature you'd like to see
Suggest dify to access the BGE RANK rearrangement model
3. How will this feature improve your workflow or experience?
In more rigorous language documents, such as internal rules and regulations of enterprises and institutional norms for social operation, a deep understanding of the language is required. Currently, the effectiveness of using Cohere through API calls is not optimistic, especially with a lack of support for Chinese, which may even lead to poor business performance and urgently need to be improved.
4. Additional context or comments
Here, as a supplement, we provide the relevant images of the bge run base:
Source code deployment method
Python 3.9, 3.10 CUDA 11.7
https://github.com/labring/FastGPT/tree/main/python/bge-rerank/bge-reranker-base https://github.com/labring/FastGPT/tree/main/python/bge-rerank/bge-reranker-large https://github.com/labring/FastGPT/tree/main/python/bge-rerank/bge-rerank-v2-m3
Installation dependencies
Pip install - r requirements. txt
Download the model The huggingface warehouse addresses for the three models are as follows:
Clone the model in the corresponding code directory. Directory structure:
Run code
Python app. py
Docker deployment
The image names are:
six thousand and six
environment variable
The auth token is mytoken
Docker run - d -- name reranker - p 6006:6006- e ACCESS-TOKEN=mytoken -- gpus all registry. cn hangzhou. aliyuncs. com/fastgpt/bge rank base: v0.1
Docker Compose.yml Example
Reference materials https://doc.fastai.site/docs/development/custom-models/bge-rerank/
5. Can you help us with this feature?