Research map / Architectures and efficiency / Quantization
DictLLM: Post-training Compression of Large Language Model with Dictionary Kernels
Jinho Park, Se Young Chun, Mingoo Seok
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes DictLLM, a post-training LLM compression method that decomposes weight matrices within grouped layers into a shared dictionary of kernels for mobile deployment.
Problem
Deploying LLMs on mobile devices under compute and memory constraints; post-training compression is needed to trade off performance against resource use.
Approach
Decomposes weight matrices within grouped layers into a shared dictionary of kernels (dictionary-based compression), applied post-training without retraining.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use DictLLM instead of Basis Sharing at 2x compression | Basis Sharing · LLaMA-7B WikiText-2 | perplexity: 19.99 -> 10.59 | Improved |
| prune only the coefficients, keep dictionary dense (ablation) | DictLLM · LLaMA-7B WikiText-2 at 1.43x | perplexity: 9.25 -> 7.83 | Improved |
| prune the dictionary instead of coefficients (ablation) | DictLLM · LLaMA-7B WikiText-2 at 1.43x | perplexity: 7.83 -> 9.04 | Hurt |
| use dictionary-aware Hessian lazy-update pruning instead of magnitude pruning (ablation) | DictLLM · LLaMA-7B WikiText-2 | perplexity: 47.43 -> 9.89 | Improved |
| use lazy update instead of full Hessian inversion (ablation) | DictLLM · LLaMA-7B WikiText-2 | pruning time: 13.49 h -> 55.6 min, same perplexity 9.89 | Improved |
| add dictionary refit after coefficient pruning (ablation) | DictLLM · LLaMA-7B WikiText-2 | perplexity: 11.53 -> 10.59 | Improved |
| add dictionary refit at higher sparsity (ablation) | DictLLM · LLaMA-7B WikiText-2 | perplexity reduction: 1.04 PPL at 50% sparsity vs 0.06 PPL at 20% | Improved |
| quantize both dictionary and coefficients to NVFP4 versus NVFP4/INT8 (ablation) | DictLLM · LLaMA-7B WikiText-2 | compression gain and perplexity: 16.05 -> 16.82 ppl; 2.60x -> 3.56x | Mixed |
| add NVFP4 quantization after factorization, pruning and refit (ablation) | DictLLM · LLaMA-7B | average zero-shot accuracy: 0.43 -> 0.40; 2x -> 7x compression | Hurt |
| add post-compression LoRA fine-tuning | DictLLM · LLaMA-7B WikiText-2 | perplexity: up to 1.9 points lower | Improved |
| increase calibration set size from 32 to 256 (ablation) | DictLLM and Basis Sharing · LLaMA-7B WikiText-2 at 1.43x | perplexity | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Basis Sharing — Basis Sharing: Cross-Layer Parameter Sharing for Large Language Model Compression (ICLR 2025) · better on 9 of 9 shared results
- SVD-LLM — SVD-LLM: Truncation-aware Singular Value Decomposition for Large Language Model Compression (ICLR 2025) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Compact Language Models via Pruning and Knowledge Distillation (NeurIPS 2024 · similarity 0.90)
- SoLA: Leveraging Soft Activation Sparsity and Low-Rank Decomposition for Large Language Model Compression (AAAI 2025 · similarity 0.89)
- LowRankArena: A Standardized Evaluation Platform for SVD-Based LLM Compression (NeurIPS 2026 · similarity 0.89)
- SparseLLM: Towards Global Pruning of Pre-trained Language Models (NeurIPS 2024 · similarity 0.89)
- Compressing Large Language Models using Low Rank and Low Precision Decomposition (NeurIPS 2024 · similarity 0.89)
- 3BASiL: An Algorithmic Framework for Sparse plus Low-Rank Compression of LLMs (NeurIPS 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
