Open Source¶
We support reproducible research by sharing open-source implementations, model checkpoints, experimental resources, and reusable software developed through II-Research projects.
Open Source Projects
EasyRAG
A beginner-friendly RAG framework implementing five algorithms with interactive tools for training, evaluation, and inference.
- RAG
- Information Retrieval
- Interactive AI
ComGTIR
Official implementation of generative text–image retrieval with dual identifiers and hybrid retrieval strategies.
- Generative Retrieval
- Text–Image Retrieval
- Multimodal AI
ARF-RAG
Adaptive and retriever-friendly RAG that learns when to retrieve and how to generate effective retrieval queries through reinforcement learning.
- RAG
- Reinforcement Learning
- Adaptive Retrieval
DrIG
Source code and checkpoints for generative universal multimodal retrieval with dual-role identifiers.
- Multimodal Retrieval
- Generative Retrieval
- Dual-role Identifiers
MCKT
A parallel Mamba-based model for content-aware knowledge tracing over exercise semantics and learning sequences.
- Knowledge Tracing
- Mamba
- Educational AI
MHTA
Multi-dimensional hierarchical temporal alignment for improving temporal commonsense reasoning in large language models.
- Temporal Reasoning
- Commonsense
- Large Language Models
PT-Ranking
A PyTorch framework for learning-to-rank research and reproducible experimentation.
- Learning to Rank
- PyTorch
- Information Retrieval