| Challenge: | Existing models that use VTs as their backbone model are based on UTs that share parameters across layers and have better compositional generalization. |
| Approach: | They propose to use Sparse Mixture of Experts to reduce UT's computation complexity while retaining its parameter efficiency and generalization ability. |
| Outcome: | The proposed model achieves strong generalization results on formal language tasks and impressive parameter and computation efficiency on standard natural language benchmarks. |
Similar Papers
Towards Being Parameter-Efficient: A Stratified Sparsely Activated Transformer with Dynamic Capacity (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Recent studies have established that Mixture-of-experts models are parameter-inefficient as the improvement in performance diminishes with an increasing number of experts. |
| Approach: | They propose a mix-of-experts model that uses sparse activation to increase the number of parameters while maintaining low computational requirements per token. |
| Outcome: | The proposed models outperform state-of-the-art models on three multilingual machine translation benchmarks with 4, 15, and 94 language pairs. |
Towards A Unified View of Sparse Feed-Forward Network in Pretraining Large Language Model (2023.emnlp-main)
Copied to clipboard
| Challenge: | Large and sparse feed-forward layers (S-FFN) have proven effective in scaling up the model size for pretraining large language models. |
| Approach: | They compare S-FFN architectures for language modeling and compare their performance and efficiency . they found a simpler selection method that selects blocks through their mean aggregated hidden states . |
| Outcome: | The proposed model size and selection method achieve lower perplexity in language model pretraining compared to existing MoE architectures. |
SparseFlow: Accelerating Transformers by Sparsifying Information Flows (2024.acl-long)
Copied to clipboard
| Challenge: | SparseFlow is an efficient method to sparsify the dense information flows within transformers. |
| Approach: | They propose a method to sparsify the dense pathways of token representations across all transformer blocks by parameterizing them to be sparse. |
| Outcome: | The proposed method reduces computational costs by half on average without compromising task accuracy. |
Subformer: Exploring Weight Sharing for Parameter Efficiency in Generative Transformers (2021.findings-emnlp)
Copied to clipboard
| Challenge: | Recent improvements in NLP tasks can be attributed to the Transformer model. |
| Approach: | They propose to use parameter-sharing methods to reduce parameter budgets in generative models by using sandwich-style parameter sharing and self-attentive embedding factorization. |
| Outcome: | The proposed model outperforms the current RNN model even with significantly fewer parameters. |
Less is KEN: a Universal and Simple Non-Parametric Pruning Algorithm for Large Language Models (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing pruning algorithms suffer from limitations such as architecture specificity and reliance on demanding calculations. |
| Approach: | They propose a pruning algorithm based on Kernel Density Estimation . it preserves most significant parameters while restoring others to their pre-training state . |
| Outcome: | The proposed pruning algorithm achieves better performance than the original unpruned version. |
Polymorphic Universal Transformer (2026.acl-long)
Copied to clipboard
Yilong Chen, Zitian Gao, Yihao Xiao, Jason Klein Liu, Xinyu Yang, Yifan Luo, Haoming Luo, Zhengmao Ye, Tingwen Liu, Ran Tao, Bryan Dai
| Challenge: | Compute Distribution Skew is a pathological phenomenon in ultra-deep recurrent models . it causes over-smoothing, representation rank collapse, and degraded reasoning performance. |
| Approach: | They propose a dynamic architecture that redefines recursive computation by decoupling parameter count from depth. |
| Outcome: | The proposed model significantly improves representation rank and reasoning robustness while reducing computation by 64.7%. |
Sparser Mixture-of-Adapters with Cross-Layer Generalization (2025.naacl-long)
Copied to clipboard
| Challenge: | Existing methods for training large language models do not allow sharing adapters across layers . existing methods do not support sharing adapter pools, leading to redundancy and poor generalization . |
| Approach: | They propose a mixture-of-adapter framework that trains a pool of lightweight adapters at each layer and selects the most suitable ones for each input. |
| Outcome: | The proposed framework reduces active adapters by over 85% while boosting task accuracy. |
Approximating Two-Layer Feedforward Networks for Efficient Transformers (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Recent work uses sparse Mixtures of Experts (MoEs) to build resource-efficient large language models. |
| Approach: | They propose a general framework that unifies various methods to build two-layer NNs . they propose methods to improve both MoEs and PKMs based on their results . |
| Outcome: | The proposed framework improves both MoEs and product-key memories (PKMs) it shows that MoE's are competitive with dense Transformer-XL on two different scales while being much more resource efficient. |
AutoMoE: Heterogeneous Mixture-of-Experts with Adaptive Computation for Efficient Neural Machine Translation (2023.findings-acl)
Copied to clipboard
Ganesh Jawahar, Subhabrata Mukherjee, Xiaodong Liu, Young Jin Kim, Muhammad Abdul-Mageed, Laks Lakshmanan, V.S., Ahmed Hassan Awadallah, Sebastien Bubeck, Jianfeng Gao
| Challenge: | Existing MoE designs do not consider computational constraints (e.g., FLOPs, latency) Existing works in MoE consider homogeneous design where the same number of experts of the same size are placed uniformly throughout the network. |
| Approach: | They propose a framework for designing heterogeneous MoEs under computational constraints. |
| Outcome: | The proposed framework achieves 4x inference speedup and FLOPs reduction over manual models and within 1 BLEU point of MoE SwitchTransformer over benchmark datasets for NMT. |
Easy and Efficient Transformer: Scalable Inference Solution For Large NLP Model (2022.naacl-industry)
Copied to clipboard
Gongzheng Li, Yadong Xi, Jingzhen Ding, Duan Wang, Ziyang Luo, Rongsheng Zhang, Bai Liu, Changjie Fan, Xiaoxi Mao, Zeng Zhao
| Challenge: | Recent studies show that transformer-based models are effective over many tasks, but they are expensive to deploy in the industrial application. |
| Approach: | They propose a transformer-based inference solution that optimizes kernels for long inputs and large hidden sizes and a flexible CUDA memory manager to reduce the memory footprint when deploying a large model. |
| Outcome: | The proposed solution achieves an average speedup of 1.40-4.20x on the transformer decoder layer with an A100 GPU. |