With the increasing size of large language models, layer pruning has gained increased attention as a hardware-friendly approach for model compression. However, existing layer pruning methods struggle to simultaneously address key practical deployment challenges, including performance degradation, high training costs, and limited acceleration. To overcome these limitations, we propose \name, a task-\underline{E}ffective, training-\underline{E}conomical and inference-\underline{E}fficient layer pruning framework. \namespace introduces two key innovations: (1) a differentiable mask optimization method using a Gumbel-TopK sampler, enabling efficient and precise pruning mask search; and (2) an entropy-aware adaptive knowledge distillation strategy that enhances task performance. Extensive experiments over diverse model architectures and benchmarks demonstrate the superiority of our method over state-of-the-art approaches. Notably, \namespace achieves 96\% accuracy, a mere 0.8\% drop from the original model (96.8\%) on MATH-500 when pruning 25\% layers of Qwen3-32B, outperforming existing SOTA (95\%), with a 1.33$\times$ inference speedup by consuming merely 0.5B tokens (0.5\% of the post-training data volume).

Large language models (LLMs) have transformed many areas of natural language processing, including machine translation. However, efficient deployment of LLMs remains challenging due to their intensive computational requirements. In this paper, we address this challenge and present our submissions to the Model Compression track at the Conference on Machine Translation (WMT 2025). In our experiments, we investigate iterative layer pruning guided by layer importance analysis. We evaluate this method using the Aya-Expanse-8B model for translation from Czech to German, and from English to Egyptian Arabic. Our approach achieves substantial reductions in model size and inference time, while maintaining the translation quality of the baseline models.

Layer pruning has emerged as a widely adopted technique for improving the efficiency of large language models (LLMs). Although existing methods demonstrate strong performance retention on general knowledge tasks, their effect on long-chain reasoning, a more brittle yet crucial capability, remains largely unexplored. In this work, we study the impact of layer pruning on long-chain reasoning through the lens of test-time scaling, a key mechanism in modern LLMs that enables strong reasoning capacity by allocating more computation at inference time. With extensive experiments, we demonstrate that pruning even one or two layers can severely impair test-time scaling, with performance collapsing drastically on long reasoning benchmarks even when performance on knowledge-intensive and shallow reasoning tasks remains stable. Furthermore, we find that standard supervised fine-tuning remedies fail to recover test-time scaling once it has deteriorated. Through in-depth analyses, we identify the mechanisms underlying this fragility of test-time scaling and highlight the fundamental risks of applying layer pruning to reasoning-intensive LLMs. These findings call for a rethinking of layer pruning strategies and provide insights for developing methods that preserve the robustness of reasoning. We open-source the codebase in \href{https://github.com/keyu-wang-2002/Layer-Pruning-Harms-Inference-Scaling}{https://github.com/keyu-wang-2002/Layer-Pruning-Harms-Inference-Scaling}.

Large Language Models excel at natural language processing tasks, but their massive size leads to high computational and storage demands. Recent works have sought to reduce their model size through layer-wise structured pruning. However, they tend to ignore retaining the capabilities in the pruned part. In this work, we re-examine structured pruning paradigms and uncover several key limitations: 1) notable performance degradation due to direct layer removal, 2) incompetent linear weight layer aggregation, and 3) the lack of effective post-training recovery mechanisms. To address these limitations, we propose CoMe, including a progressive layer pruning framework with a Concatenation-based Merging technology and a hierarchical distillation post-training process. Specifically, we introduce a channel sensitivity metric that utilizes activation intensity and weight norms for fine-grained channel selection. Subsequently, we employ a concatenation-based layer merging method to fuse the most critical channels across adjacent layers, enabling progressive model size reduction. Finally, we propose a hierarchical distillation protocol that leverages the correspondences between the original and pruned model layers established during pruning, thereby enabling efficient knowledge transfer. Experiments on seven benchmarks show that CoMe achieves state-of-the-art performance; when pruning 30% of LLaMA-2-7b's parameters, the pruned model retains 83% of its original average accuracy. Our code is available at https://github.com/MPI-Lab/CoMe.

Convolutional Neural Networks (CNNs) have achieved significant breakthroughs in various fields. However, these advancements have led to a substantial increase in the complexity and size of these networks. This poses a challenge when deploying large and complex networks on edge devices. Consequently, model compression has emerged as a research field aimed at reducing the size and complexity of CNNs. One prominent technique in model compression is model pruning. This paper will present a new technique of pruning that combines both channel and layer pruning in what is called a "hybrid pruning framework". Inspired by EfficientNet, a renowned CNN architecture known for scaling up networks from both channel and layer perspectives, this hybrid approach applies the same principles but in reverse, where it scales down the network through pruning. Experiments on the hybrid approach demonstrated a notable decrease in the overall complexity of the model, with only a minimal reduction in accuracy compared to the baseline model. This complexity reduction translates into reduced latency when deploying the pruned models on an NVIDIA JETSON TX2 embedded AI device.

Efficient deployment of large audio-language models for speech translation remains challenging due to their significant computational requirements. In this paper, we address this challenge through our system submissions to the "Model Compression" track at the International Conference on Spoken Language Translation (IWSLT 2025). We experiment with a combination of approaches including iterative layer pruning based on layer importance evaluation, low-rank adaptation with 4-bit quantization (QLoRA), and knowledge distillation. In our experiments, we use Qwen2-Audio-7B-Instruct for speech translation into German and Chinese. Our pruned (student) models achieve up to a 50% reduction in both model parameters and storage footprint, while retaining 97-100% of the translation quality of the in-domain (teacher) models.

Layer pruning has emerged as a promising technique for compressing large language models (LLMs) while achieving acceleration proportional to the pruning ratio. In this work, we identify that removing any layer induces a significant magnitude gap in hidden states, resulting in substantial performance degradation. To address this issue, we propose Prune&Comp, a novel plug-and-play layer pruning scheme that leverages magnitude compensation to mitigate such gaps in a training-free manner. Specifically, we first estimate the magnitude gap caused by layer removal and then eliminate this gap by rescaling the remaining weights offline, with zero runtime overhead incurred. We further demonstrate the advantages of Prune&Comp through an iterative pruning strategy. When integrated with an iterative prune-and-compensate loop, Prune&Comp consistently enhances existing layer pruning metrics. For instance, when 5 layers of LLaMA-3-8B are pruned using the prevalent block influence metric, Prune&Comp nearly halves the perplexity and retains 93.19\% of the original model's question-answering performance, outperforming the baseline by 4.01%.

AI's exponential growth intensifies computational demands and energy challenges. While practitioners employ various optimization techniques, that we refer as "knobs" in this paper, to tune model efficiency, these are typically afterthoughts and reactive ad-hoc changes applied in isolation without understanding their combinatorial effects on energy efficiency. This paper emphasizes on treating energy efficiency as the first-class citizen and as a fundamental design consideration for a compute-intensive pipeline. We show that strategic selection across five AI pipeline phases (data, model, training, system, inference) creates cascading efficiency. Experimental validation shows orthogonal combinations reduce energy consumption by up to $94.6$% while preserving $95.95$% of the original F1 score of non-optimized pipelines. This curated approach provides actionable frameworks for informed sustainable AI that balance efficiency, performance, and environmental responsibility.

Early exits (EEs) offer a promising approach to reducing computational costs and latency by dynamically terminating inference once a satisfactory prediction confidence on a data sample is achieved. Although many works integrate EEs into encoder-only Transformers, their application to decoder-only architectures and, more importantly, Mamba models, a novel family of state-space architectures in the LLM realm, remains insufficiently explored. This work introduces DYNAMAX, the first framework to exploit the unique properties of Mamba architectures for early exit mechanisms. We not only integrate EEs into Mamba but also repurpose Mamba as an efficient EE classifier for both Mamba-based and transformer-based LLMs, showcasing its versatility. Our experiments employ the Mistral 7B transformer compared to the Codestral 7B Mamba model, using data sets such as TruthfulQA, CoQA, and TriviaQA to evaluate computational savings, accuracy, and consistency. The results highlight the adaptability of Mamba as a powerful EE classifier and its efficiency in balancing computational cost and performance quality across NLP tasks. By leveraging Mamba's inherent design for dynamic processing, we open pathways for scalable and efficient inference in embedded applications and resource-constrained environments. This study underscores the transformative potential of Mamba in redefining dynamic computing paradigms for LLMs.

Large Language Models (LLMs) possess outstanding capabilities in addressing various natural language processing (NLP) tasks. However, the sheer size of these models poses challenges in terms of storage, training and inference due to the inclusion of billions of parameters through layer stacking. While traditional approaches such as model pruning or distillation offer ways for reducing model size, they often come at the expense of performance retention. In our investigation, we systematically explore the approach of reducing the number of layers in LLMs. Surprisingly, we observe that even with fewer layers, LLMs maintain similar or better performance levels, particularly in prompt-based fine-tuning for text classification tasks. Remarkably, in certain cases, models with a single layer outperform their fully layered counterparts. These findings offer valuable insights for future work aimed at mitigating the size constraints of LLMs while preserving their performance, thereby opening avenues for significantly more efficient use of LLMs.