FLUX 3 Action Collection Open weights 7B world action model: base and shared encoders, SO-101 and DROID policies. Code: https://github.com/black-forest-labs/flux-action • 3 items • Updated Oct 2 • 47
Not All Prompts Are Equal: Exploration-Guided Prompt Scaffolding for Multimodal Reinforcement Post-Training Paper • 2609.15051 • Published Sep 14 • 11
PlaidQ Collection Continuous diffusion LM for code: the 0.7B teacher plus its few-step and one-step distilled students. • 3 items • Updated Sep 7 • 1
ForgeWM: Progressive Causal Training for Few-Step Action-Conditioned Video World Models Paper • 2608.14022 • Published Aug 14 • 19
Moebius: 0.2B Lightweight Image Inpainting Framework with 10B-Level Performance Paper • 2606.19195 • Published Jun 17 • 66
Foundation Text-Generation Models Below 360M Parameters Collection Great candidates for fine-tuning targeting Wllama and Transformers.js for mobile devices, ordered by number of parameters. • 101 items • Updated Sep 14 • 47
Mistral Small 4 Collection A state-of-the-art model, open-weight, with a granular Mixture-of-Experts architecture that fuses instruct, reasoning and agentic skills. • 3 items • Updated 29 days ago • 84
FastRTC Custom UIs Collection A collection of FastRTC demos that showcase how to built a Custom UI for your server • 4 items • Updated Apr 7, 2025 • 2
MOE/Mixture of Experts Models (see also "source" cll) Collection Mixture of Expert Models by me. This leverages the power of multiple models at the same time during generation for next level performance. • 61 items • Updated 20 days ago • 22
MobileLLM Collection Optimizing Sub-billion Parameter Language Models for On-Device Use Cases (ICML 2024) https://arxiv.org/abs/2402.14905 • 49 items • Updated Mar 2 • 141
LLM in a flash: Efficient Large Language Model Inference with Limited Memory Paper • 2312.11514 • Published Dec 12, 2023 • 265