Mixture-of-Recursions: Learning Dynamic Recursive Depths for Adaptive Token-Level Computation (NeurIPS 2025)
-
Updated
Sep 26, 2025 - Python
#
adaptive-computation
Here are 8 public repositories matching this topic...
A curated reading list of research in Adaptive Computation, Inference-Time Computation & Mixture of Experts (MoE).
nlp machine-learning computer-vision tensorflow transformers pytorch mixture-of-experts adaptive-computation
-
Updated
Jan 1, 2025
[ICLR 2025] Dynamic Mixture of Experts: An Auto-Tuning Approach for Efficient Transformer Models
-
Updated
Jul 9, 2025 - Python
(NeurIPS-2019 MicroNet Challenge - 3rd Winner) Open source code for "SIPA: A simple framework for efficient networks"
pruning model-compression micronet model-acceleration neurips-2019 micronet-challenge early-exiting adaptive-computation compact-neural-network
-
Updated
Dec 18, 2022 - Python
Bio-inspired LLM architecture achieving ~5x reduction in Attention FLOPs via Context-Aware Target-Sparsity Routing. Features custom OpenAI Triton kernels optimized for RTX 5090, delivering 1.41M tok/s throughput.
machine-learning research cuda efficiency pytorch triton adaptive-computation sparse-attention llm rtx-5090
-
Updated
Dec 25, 2025 - Python
The ARL Hierarchical MultiScale Framework (ARL-HMS) is a software library for development of multiscale models on heterogeneous high-performance computing systems.
-
Updated
Jul 18, 2024 - C++
Model implementation for "Adaptive computation as a new mechanism of dynamic human attention"
-
Updated
Jun 2, 2025 - Julia
Lightweight PyTorch implementation of Mixture-of-Recursions with Expert-Choice & Token-Choice routing | Runs on your laptop!
router neural-network python3 pytorch mor mixture-of-experts early-exiting adaptive-computation kv-cache llm recursive-transformers
-
Updated
Nov 25, 2025 - Python
Improve this page
Add a description, image, and links to the adaptive-computation topic page so that developers can more easily learn about it.
Add this topic to your repo
To associate your repository with the adaptive-computation topic, visit your repo's landing page and select "manage topics."