Opensource

We are conducting research with opensource policy. So far, we have engaged many opensource projects for datasets and algorithms. See the details below.

Large Language Models

SummLlama: Faithful Summarization Model [Link]

Ext2Gen: Robust Generation Model for RAG [Link]

Benchmark Datasets

UniSumEval: Benchmark Data for Text Summarization [Link]

ToFuEval: Hallucination Benchmark Data [Link]

ANIMAL-10N Data: Real-world Data with Noisy Labels [Link]

Algorithms

Retrieval for RAG

Word2Passage: Accurate Retrieval Method [Link]

Transformers

ViDT: A Fully Transformer-based Object Detector [Link]

Fast Autoregressive Decoding [Link]

DisCal: Calibrated Distillation [Link]

MEDUSA: RGB-D Transformer-based Object Detector [Link]

Robust Deep Learning

Awesome Noisy Labels [Link]

SELFIE: Robust Training against Noisy Labels [Link]

Prune4Rel: Robust Data Pruning [Link]

MQNet: Open-set Active Learning [Link]

FedRN: Federated Learning with Noisy Labels [Link]

Continual Learning

DAP: Instance-level Prompt-based CL with Transformers [Link]

SDP: Scheduled Data Prior for Continual Learning [Link]