Code for "LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding", ACL 2024
-
Updated
Jul 20, 2026 - Python
#
early-exit
Here are 23 public repositories matching this topic...
A curated list of early exiting (LLM, CV, NLP, etc)
-
Updated
Aug 21, 2024
Dynamic per-token early exit for LLM inference. Skip layers tokens don't need
-
Updated
Mar 18, 2026 - Python
Artifact for "Apparate: Rethinking Early Exits to Tame Latency-Throughput Tensions in ML Serving" [SOSP '24]
-
Updated
Nov 21, 2024 - Python
[NeurIPS'25] FreqExit: Enabling Early-Exit Inference for Visual Autoregressive Models via Frequency-Aware Guidance
-
Updated
Dec 15, 2025 - Python
Official PyTorch implementation of "LGViT: Dynamic Early Exiting for Accelerating Vision Transformer" (ACM MM 2023)
-
Updated
Nov 18, 2024 - Python
Code for paper "EdgeKE: An On-Demand Deep Learning IoT System for Cognitive Big Data on Industrial Edge Devices"
iot deep-neural-networks deep-learning inference resnet knowledge-distillation edge-computing edge-devices resnets inference-optimization early-exit jetson-nano branchynet edge-applications accuracy-requirements efficient-ai
-
Updated
Jul 22, 2026 - Python
A deep learning framework that implements Early Exit strategies in Convolutional Neural Networks (CNNs) using Deep Q-Learning (DQN). This project enhances computational efficiency by dynamically determining the optimal exit point in a neural network for image classification tasks on CIFAR-10.
reinforcement-learning deep-learning cnn pytorch dqn image-classification cifar10 cifar-10 pytorch-cnn cnn-pytorch cifar10-classification early-exit model-optimization efficient-ai
-
Updated
Feb 23, 2025 - Jupyter Notebook
Official repository of Busolin et al., "Learning Early Exit Strategies for Additive Ranking Ensembles", ACM SIGIR 2021.
-
Updated
May 10, 2021 - Jupyter Notebook
This repository provides an implementation of Early Exit Neural Network (EENN) and provides a way to profile the processing time of the model.
-
Updated
Mar 19, 2025 - Python
Deployment-oriented early-exit toolkit for PyTorch vision models, with per-head calibration, fair benchmarks, and staged inference.
-
Updated
Jul 22, 2026 - Python
The project aim to experiment implementing a modular architecture: an early-exit model and testing it using Tensorflow.
-
Updated
Feb 21, 2023 - Jupyter Notebook
This repository is dedicated to self-learning about early exit papers, including relevant code and documentation.
-
Updated
May 18, 2024
Relational Time Engine (RTE): runtime density regulation for compute-efficient transformer inference. Demonstrates up to 75% layer reduction with improved latency and throughput.
python benchmark machine-learning deep-learning transformer event-driven energy-efficiency ai-systems inference-optimization early-exit transformer-architecture green-ai ai-optimization runtime-systems efficient-ai energy-efficient-ai transformer-inference-overhead runtime-gating relational-time
-
Updated
Mar 12, 2026 - Python
VerDee: Vertical Deep Network on Mamba-2 — staged LoRA (shallow/mid/deep), early-exit routing, and domain experts. Pilot on 370M; 16GB GPU friendly. Research/experiment repo.
lora state-space-model peft early-exit huggingface qlora llm-finetuning mamba2 vertical-deep-network
-
Updated
May 25, 2026 - Python
This project focuses on the automatic classification of corn leaf diseases using deep neural networks. The dataset includes over 4000 images categorized into four classes: Common Rust, Gray Leaf Spot, Blight, and Healthy. Through the use of Convolutional Neural Networks and advanced techniques, the model achieves a classification accuracy of 91.5%
-
Updated
Oct 6, 2024 - Jupyter Notebook
[ICML 2026] Think Less, Act Early: Reinforced Latent Reasoning with Early Exit in Vision-Language-Action Models
reinforcement-learning robotics vla robotics-algorithms early-exit vision-language-action openvla latent-reasoning icml-2026
-
Updated
May 15, 2026 - Python
C implementation of a SHA-1 cracker with various optimizations
-
Updated
Dec 19, 2018 - C
Early exit inference framework for HuggingFace LLMs — skip unnecessary transformer layers when the model is already confident. Supports LLaMA, Mistral, Phi, Gemma, Qwen, Pythia and more.
deep-learning transformers pytorch language-models knowledge-distillation inference-optimization early-exit huggingface model-optimization llm-inference
-
Updated
Mar 29, 2026 - Python
Confidence-gated early-exit inference for Llama 3.1 8B, with distillation, calibration, segmented multi-GPU runtime, and evaluation.
-
Updated
Jul 10, 2026 - Python
Improve this page
Add a description, image, and links to the early-exit topic page so that developers can more easily learn about it.
Add this topic to your repo
To associate your repository with the early-exit topic, visit your repo's landing page and select "manage topics."