2024-12-16

Models

• Welcome to the Falcon 3 Family of Open Models!
• GitHub - foundation-model-stack/bamba: Train, tune, and infer Bamba model
• GitHub - AnswerDotAI/ModernBERT: Bringing BERT into modernity via both architecture changes and scaling

Papers

• [2412.07752] FlashRNN: Optimizing Traditional RNNs on Modern Hardware
• [2412.10360] Apollo: An Exploration of Video Understanding in Large Multimodal Models
• [2412.10117] CosyVoice 2: Scalable Streaming Speech Synthesis with Large Language Models
• [2410.18779] A Little Help Goes a Long Way: Efficient LLM Training by Leveraging Small LMs
• [2412.10302] DeepSeek-VL2: Mixture-of-Experts Vision-Language Models for Advanced Multimodal Understanding
• [2412.09607] Spectral Image Tokenizer
• [2412.12095] Causal Diffusion Transformers for Generative Modeling
• [2412.10437] SVGFusion: Scalable Text-to-SVG Generation via Vector Space Diffusion
• [2410.02899] FactCheckmate: Preemptively Detecting and Mitigating Hallucinations in LMs
• [2412.01951] Self-Improvement in Language Models: The Sharpening Mechanism
• [2412.13061] VidTok: A Versatile and Open-Source Video Tokenizer
• [[2412.14164] MetaMorph: Multimodal Understanding and Generation via Instruction Tuning](https://arxiv.org/abs/2412.14164
• [2412.12432] Three Things to Know about Deep Metric Learning
• GitHub - foundation-model-stack/bamba: Train, tune, and infer Bamba model
• [2412.13303] FastVLM: Efficient Vision Encoding for Vision Language Models
• [2412.15115] Qwen2.5 Technical Report
• [2412.15213] Flowing from Words to Pixels: A Framework for Cross-Modality Evolution
• [2412.13663] Smarter, Better, Faster, Longer: A Modern Bidirectional Encoder for Fast, Memory Efficient, and Long Context Finetuning and Inference
• [2412.14475] MegaPairs: Massive Data Synthesis For Universal Multimodal Retrieval

Code

• GitHub - hao-ai-lab/FastVideo: FastVideo is an open-source framework for accelerating large video diffusion model.
• GitHub - Genesis-Embodied-AI/Genesis: A generative world for general-purpose robotics & embodied AI learning.
• GitHub - huggingface/picotron: Minimalistic 4D-parallelism distributed training framework for education purpose

Articles

• [ ]

Videos

• [ ]

Other

• [ ]

Tweets