• DeepSpeed Ulysses: System Optimizations for Enabling Training of Extreme Long Sequence Transformer Models
• Aligning Large Multimodal Models with Factually Augmented RLHF
• LAVIE: High-Quality Video Generation with Cascaded Latent Diffusion Models

Support the show

• CoRF : Colorizing Radiance Fields using Knowledge Distillation
• The Cambridge Law Corpus: A Corpus for Legal AI Research
• CodePlan: Repository-level Coding using LLMs and Planning
• DualToken-ViT: Position-aware Efficient Vision Transformer with Dual Token Fusion

Support the show

• Parallelizing non-linear sequential models over the sequence length
• Fast Feedforward Networks
• LongLoRA: Efficient Fine-tuning of Long-Context Large Language Models
• A Paradigm Shift in Machine Translation: Boosting Translation Performance of Large Language Models
• Boolformer: Symbolic Regression of Logic Functions with Transformers

Support the show

• FreeU: Free Lunch in Diffusion U-Net
• Neurons in Large Language Models: Dead, N-gram, Positional
• DreamLLM: Synergistic Multimodal Comprehension and Creation
• Kosmos-2.5: A Multimodal Literate Model
• End-to-End Speech Recognition Contextualization with Large Language Models
• The Languini Kitchen: Enabling Language Modelling Research at Different Scales of Compute

Support the show

• Graph Neural Networks Use Graphs When They Shouldn't
• Large Language Models for Compiler Optimization
• OpenBA: An Open-sourced 15B Bilingual Asymmetric seq2seq Model Pre-trained from Scratch
• Baichuan 2: Open Large-scale Language Models
• Language Modeling Is Compression
• FoleyGen: Visually-Guided Audio Generation

Support the show

• Textbooks Are All You Need II: phi-1.5 technical report
• DiffBIR: Towards Blind Image Restoration with Generative Diffusion Prior
• When Less is More: Investigating Data Pruning for Pretraining LLMs at Scale
• MADLAD-400: A Multilingual And Document-Level Large Audited Dataset
• FIAT: Fusing learning paradigms with Instruction-Accelerated Tuning
• Optimize Weight Rounding via Signed Gradient Descent for the Quantization of LLMs

Support the show

• Large-Scale Automatic Audiobook Creation
• CityDreamer: Compositional Generative Model of Unbounded 3D Cities
• From Sparse to Dense: GPT-4 Summarization with Chain of Density Prompting
• Mobile V-MoEs: Scaling Down Vision Transformers via Sparse Mixture-of-Experts
• High-Quality Entity Segmentation

Support the show

• Large Language Models as Optimizers
• FLM-101B: An Open LLM and How to Train It with $100K Budget
• XGen-7B Technical Report
• Tracking Anything with Decoupled Video Segmentation
• DoLa: Decoding by Contrasting Layers Improves Factuality in Large Language Models

Support the show

• SLiMe: Segment Like Me
• Matcha-TTS: A fast TTS architecture with conditional flow matching
• Physically Grounded Vision-Language Models for Robotic Manipulation
• Scaling Autoregressive Multi-Modal Models: Pretraining and Instruction Tuning

Support the show

• One Wide Feedforward is All You Need
• Efficient RLHF: Reducing the Memory Usage of PPO
• PromptTTS 2: Describing and Generating Voices with Text Prompt
• AniPortraitGAN: Animatable 3D Portrait Generation from 2D Image Collections

Support the show