DNeo: Your Parallel Draft Model Needs Better Target Feature Encoding Paper • 2611.21925 • Published 3 days ago • 3
view article Article Assisted Generation: a new direction toward low-latency text generation joaogante • May 11, 2023 • 83
Running Featured 100 Flow Matching for Image Generation 🌊 100 Visualize how noise transforms into images in your browser
Gated Recurrent Transformer (GRT) Collection Pre-trained checkpoints for the Gated Recurrent Transformer (GRT). arXiv:2608.15062 • 5 items • Updated Sep 1 • 2
Unlocking Lossless Speedups in LLMs via Discrete Diffusion Paper • 2609.04010 • Published Sep 3 • 112
Don't Drop Dropout: Optimizing Layer Sparsity for Efficient LLM Training and Inference Paper • 2609.05275 • Published Sep 4 • 23
Don't Drop Dropout: Optimizing Layer Sparsity for Efficient LLM Training and Inference Paper • 2609.05275 • Published Sep 4 • 23
Unlocking Lossless Speedups in LLMs via Discrete Diffusion Paper • 2609.04010 • Published Sep 3 • 112
Don't Drop Dropout: Optimizing Layer Sparsity for Efficient LLM Training and Inference Paper • 2609.05275 • Published Sep 4 • 23 • 2
Calibrating Beyond English: Language Diversity for Better Quantized Multilingual LLM Paper • 2601.18306 • Published Jan 26
Gated Recurrent Transformers: Expressive Depth through Recurrent Modulation Paper • 2608.15062 • Published Aug 26 • 12