Change the repository type filter
All
Repositories list
20 repositories
- For OpenMOSS Mechanistic Interpretability Team's Sparse Autoencoder (SAE) research.
- MOSS-TTSD is a spoken dialogue generation model that enables expressive dialogue speech synthesis in both Chinese and English, supporting zero-shot multi-speaker voice cloning, and long-form speech generation.
UnifiedToolHub
Public- LongLLaDA: Unlocking Long Context Capabilities in Diffusion LLMs
TransformerLens
Public- a survey of long-context LLMs from four perspectives, architecture, infrastructure, training, and evaluation
ReAttention
Public[ICLR2025] ReAttention, a training-free approach to break the maximum context length in length extrapolationUltra-Innerthought
PublicUltra-Innerthought is a bilingual (Chinese and English) open-domain R1/o1 style SFT dataset.SpeechGPT-2.0-preview
PublicCoLLiE
PublicAnyGPT
PublicMOSS
PublicAn open-source tool-augmented conversational language model from Fudan UniversityHalluQA
PublicDataset and evaluation script for "Evaluating Hallucinations in Chinese Large Language Models"GAOKAO-MM
PublicSay-I-Dont-Know
Public[ICML'2024] Can AI Assistants Know What They Don't Know?