LILA: Calibration-Free Structured Pruning of Large Language Models via Latent Spectral Geometry
Updated 15 min ago · first seen 11 Sept 2026
paper_01M294FP1VVTCW15ZGNK0HCF54
- Published
- 11 Sept 2026
- T1 · 15 min ago
- arXiv
- 2609.11163
- T1 · 15 min ago
- Category
- cs.LG
- T1 · 15 min ago
Abstract
Structured pruning of large language models (LLMs) offers hardware-efficient compression, yet existing methods require calibration data, gradient computation, or large auxiliary policy networks at pruning time. LILA (\emph{Latent-Informed Layer Analysis}) scores neuron importance via the Kolmogorov--Smirnov (KS) distance between empirical singular value distributions of the full and neuron-ablated feed-forward network (FFN) weight matrix, providing a closed-form spectral rule requiring no training, calibration data, or auxiliary network. Without any fine-tuning, LILA surpasses PruneNet (45M-parameter RL policy) by 1.57~pp in zero-shot accuracy on LLaMA-2-7B at 25\% sparsity, and outperforms WikiText-2-calibrated SliceGPT by up to 6.0~pp across all sparsity levels, while preserving the original architecture. After one epoch of LoRA recovery fine-tuning, LILA achieves highly competitive performance, matching the heavily calibrated SliceGPT baseline to within a 0.48~pp margin across LLaMA-2-7B and Phi-2, despite using zero calibration data. A Neural Tangent Kernel analysis confirms a 22$\times$ reduction in functional distortion versus random pruning, providing theoretical grounding for the spectral importance criterion. Finally, extending LILA to dynamically allocate sparsity budgets via KS-scores yields state-of-the-art generative preservation at moderate compression, while uncovering fundamental single-layer architectural bottlenecks at higher compression regimes.
Authors 6
Sankar Behera, Dhruv Singh, Anshika Agnihotri, Raj Kumar Choudhary, Satyadev Ahlawat, Yamuna Prasad
Specification
- Official page
Source:arXiv (Atom API + RSS)T1observed 15 min agohigh
- Arxiv announce type
- cross
Source:arXiv (Atom API + RSS)T1observed 15 min agohigh
- arXiv id
- 2609.11163
Source:arXiv (Atom API + RSS)T1observed 15 min agohigh
- Categories
- cs.LG, cs.CL
Source:arXiv (Atom API + RSS)T1observed 15 min agohigh
Source:arXiv (Atom API + RSS)T1observed 15 min agohigh
- Primary category
- cs.LG
Source:arXiv (Atom API + RSS)T1observed 15 min agohigh
- Published
- 11 Sept 2026
Source:arXiv (Atom API + RSS)T1observed 15 min agohigh
Each value shows its source, tier and observation time. Conflicting claims are kept side by side and flagged — never averaged. How AI Atlas records facts →
Provenance
Attributed facts
9
Source tiers
T19
Freshest observation
15 min ago
Conflicts
None
No models linked to this paper yet.
- Authors
- Sankar Behera, Dhruv Singh, Anshika Agnihotri
As of
Rewind the record: see this entity's attributes exactly as AI Atlas knew them on a given day.
Claim history
Official pageofficial_url1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| https://arxiv.org/abs/2609.11163 | → current | current | arXiv (Atom API + RSS)T1 | high | deterministic |
Abstractabstract1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| Structured pruning of large language models (LLMs) offers hardware-efficient compression, yet existing methods require calibration data, gradient computation, or large auxiliary policy networks at pruning time. LILA (\emph{Latent-Informed Layer Analysis}) scores neuron importance via the Kolmogorov--Smirnov (KS) distance between empirical singular value distributions of the full and neuron-ablated feed-forward network (FFN) weight matrix, providing a closed-form spectral rule requiring no training, calibration data, or auxiliary network. Without any fine-tuning, LILA surpasses PruneNet (45M-parameter RL policy) by 1.57~pp in zero-shot accuracy on LLaMA-2-7B at 25\% sparsity, and outperforms WikiText-2-calibrated SliceGPT by up to 6.0~pp across all sparsity levels, while preserving the original architecture. After one epoch of LoRA recovery fine-tuning, LILA achieves highly competitive performance, matching the heavily calibrated SliceGPT baseline to within a 0.48~pp margin across LLaMA-2-7B and Phi-2, despite using zero calibration data. A Neural Tangent Kernel analysis confirms a 22$\times$ reduction in functional distortion versus random pruning, providing theoretical grounding for the spectral importance criterion. Finally, extending LILA to dynamically allocate sparsity budgets via KS-scores yields state-of-the-art generative preservation at moderate compression, while uncovering fundamental single-layer architectural bottlenecks at higher compression regimes. | → current | current | arXiv (Atom API + RSS)T1 | high | deterministic |
Arxiv announce typearxiv_announce_type2
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| cross | → current | current | arXiv (Atom API + RSS)T1 | high | deterministic |
| new | → | superseded | arXiv (Atom API + RSS)T1 | high | deterministic |
arXiv idarxiv_id1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| 2609.11163 | → current | current | arXiv (Atom API + RSS)T1 | high | deterministic |
Categoriescategories1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| cs.LG, cs.CL | → current | current | arXiv (Atom API + RSS)T1 | high | deterministic |
PDFpdf_url1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| https://arxiv.org/pdf/2609.11163 | → current | current | arXiv (Atom API + RSS)T1 | high | deterministic |
Primary categoryprimary_category1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| cs.LG | → current | current | arXiv (Atom API + RSS)T1 | high | deterministic |
Publishedpublished_at1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| 11 Sept 2026 | → current | current | arXiv (Atom API + RSS)T1 | high | deterministic |
Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →
- Property changedPaperLILA: Calibration-Free Structured Pruning of Large Language Models via Latent Spectral Geometry
LILA: Calibration-Free Structured Pruning of Large Language Models via Latent Spectral Geometry: arxiv announce type changed from new to cross
Arxiv announce typenew→crossarxiv - New paperPaperLILA: Calibration-Free Structured Pruning of Large Language Models via Latent Spectral Geometry
New paper: LILA: Calibration-Free Structured Pruning of Large Language Models via Latent Spectral Geometry
arxiv
| Source | Document | Type | Tier | Last observed | Snapshots |
|---|---|---|---|---|---|
| arXiv (Atom API + RSS) | rss.arxiv.org/rss/cs.CL | feed | T1· Official | 15 min ago | 1 |
| arXiv (Atom API + RSS) | rss.arxiv.org/rss/cs.LG | feed | T1· Official | 15 min ago | 1 |
Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.