2025-11-10

Models

• [ ]

Papers

• [2511.08923] TiDAR: Think in Diffusion, Talk in Autoregression
• [2511.09554] RF-DETR: Neural Architecture Search for Real-Time Detection Transformers
• [2511.10643] Black-Box On-Policy Distillation of Large Language Models
• [2511.07384] Teaching Pretrained Language Models to Think Deeper with Retrofitted Recurrence

Code

• GitHub - marin-community/marin: Open-source framework for the research and development of foundation models.
• GitHub - erogol/BlaGPT: Experimental playground for benchmarking language model (LM) architectures, layers, and tricks on smaller datasets. Designed for flexible experimentation and exploration.

Articles

• the-art-of-debugging/pytorch at master · stas00/the-art-of-debugging · GitHub
• Newton-Schulz - docs.modula.systems
• Beyond Standard LLMs - by Sebastian Raschka, PhD
• fleuret.org/public/slides-talk-free-transformer.pdf
• Beyond Quantization: Bringing Sparse Inference to PyTorch – PyTorch
• Accelerating Large-Scale Mixture-of-Experts Training in PyTorch | NVIDIA Technical Blog
• SGLang Diffusion: Accelerating Video and Image Generation | LMSYS Org
• Notion
• Text-to-image Architectural Experiments

Videos

• [M2L 2025] 5.2 Diffusion models - Sander Dieleman - YouTube
• Ray Summit 2025 Keynote: Building Cursor Composer with Sasha Rush - YouTube
• Kimi K2 and Our Contributions to Open Source - Yuxin Wu, Moonshot AI - YouTube
• Helion: A High-level DSL for Kernel Authoring - Jason Ansel, Meta - YouTube
• Self-Speculative Masked Diffusions | Andrew Campbell - YouTube
• Hybrid Models as First-Class Citizens in vLLM – PyTorch
• Making GPUs Actually Fast: A Deep Dive into Training Performance - YouTube

Other

• nvidia/Nemotron-ClimbLab · Datasets at Hugging Face
• nvidia/Nemotron-ClimbMix · Datasets at Hugging Face

Tweets