kidist-amde/ddro
We introduce the direct document relevance optimization (DDRO) for training a pairwise ranker model. DDRO encourages the model to focus on document-level relevance during generation
This project helps information retrieval specialists improve how their generative search models rank documents. By taking an existing generative retrieval model and a dataset of relevant and less relevant document pairs for queries, it trains the model to directly prefer relevant documents. The output is a more accurate generative retrieval model that ranks search results better.
Use this if you are an information retrieval specialist working with generative search models and need to fine-tune them to rank documents more accurately based on their relevance.
Not ideal if you are looking for a general-purpose large language model fine-tuning tool or if your primary goal is open-ended text generation rather than structured document ranking.
Stars
36
Forks
3
Language
Python
License
Apache-2.0
Category
Last pushed
Jan 10, 2026
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/generative-ai/kidist-amde/ddro"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
soran-ghaderi/torchebm
🍓 Build and train energy-based and diffusion models in PyTorch ⚡.
spqb/adabmDCApy
Pytorch implementation of adabmDCA
opendilab/GenerativeRL
Python library for solving reinforcement learning (RL) problems using generative models (e.g....
AstraZeneca/DiffAbXL
The official implementation of DiffAbXL benchmarked in the paper "Exploring Log-Likelihood...
G-U-N/UniRL
a unified reinforcement learning toolbox for joint RL on language models and diffusion models