Cover Image

Surgical Feature-Space Decomposition of LLMs: Why, When and How?

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Chavan, Arnav, Lele, Nahush, Gupta, Deepak
Format:	Preprint
Published:	2024
Subjects:	Computation and Language Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Faster and Lighter LLMs: A Survey on Current Challenges and Way Forward
by: Chavan, Arnav, et al.
Published: (2024)

S2D: Selective Spectral Decay for Quantization-Friendly Conditioning of Neural Activations
by: Chavan, Arnav, et al.
Published: (2026)

Beyond Uniform Scaling: Exploring Depth Heterogeneity in Neural Architectures
by: T, Akash Guna R., et al.
Published: (2024)

Why LLMs Fail at Causal Discovery and How Interventional Agents Escape
by: Roy, Amartya, et al.
Published: (2026)

How Do LLMs Use Their Depth?
by: Gupta, Akshat, et al.
Published: (2025)

Knowing When to Abstain: Medical LLMs Under Clinical Uncertainty
by: Machcha, Sravanthi, et al.
Published: (2026)

When Attention Closes: How LLMs Lose the Thread in Multi-Turn Interaction
by: Dongre, Vardhan, et al.
Published: (2026)

Why are LLMs' abilities emergent?
by: Havlík, Vladimír
Published: (2025)

Mind the Pause: Disfluency-Aware Objective Tuning for Multilingual Speech Correction with LLMs
by: Kumar, Deepak, et al.
Published: (2026)

Exploring Multilingual Unseen Speaker Emotion Recognition: Leveraging Co-Attention Cues in Multitask Learning
by: Goel, Arnav, et al.
Published: (2024)

Why We Need World Models for AGI: Where LLMs Fail and How World Models May Outperform
by: Alaswad, Feisal, et al.
Published: (2026)

Look Within, Why LLMs Hallucinate: A Causal Perspective
by: Li, He, et al.
Published: (2024)

DOT-MoE: Differentiable Optimal Transport for MoEfication
by: Bamba, Udbhav, et al.
Published: (2026)

SELT: Self-Evaluation Tree Search for LLMs with Task Decomposition
by: Wu, Mengsong, et al.
Published: (2025)

Generative Adversarial Reviews: When LLMs Become the Critic
by: Bougie, Nicolas, et al.
Published: (2024)

Say Anything but This: When Tokenizer Betrays Reasoning in LLMs
by: Ayoobi, Navid, et al.
Published: (2026)

When Benchmarks Leak: Inference-Time Decontamination for LLMs
by: Chai, Jianzhe, et al.
Published: (2026)

When Attention Collapses: How Degenerate Layers in LLMs Enable Smaller, Stronger Models
by: Sanyal, Sunny, et al.
Published: (2024)

When Bias Pretends to Be Truth: How Spurious Correlations Undermine Hallucination Detection in LLMs
by: Wang, Shaowen, et al.
Published: (2025)

Why Does New Knowledge Create Messy Ripple Effects in LLMs?
by: Qin, Jiaxin, et al.
Published: (2024)

Forecasting Time Series with LLMs via Patch-Based Prompting and Decomposition
by: Bumb, Mayank, et al.
Published: (2025)

How LLMs Might Think
by: Gottlieb, Joseph, et al.
Published: (2026)

Unnatural Languages Are Not Bugs but Features for LLMs
by: Duan, Keyu, et al.
Published: (2025)

From What to Why: Thought-Space Recommendation with Small Language Models
by: Biswas, Prosenjit, et al.
Published: (2025)

Why are all LLMs Obsessed with Japanese Culture? On the Hidden Cultural and Regional Biases of LLMs
by: de Landa, Joseba Fernandez, et al.
Published: (2026)

When LLMs Team Up: The Emergence of Collaborative Affective Computing
by: Lai, Wenna, et al.
Published: (2025)

When Personalization Misleads: Understanding and Mitigating Hallucinations in Personalized LLMs
by: Sun, Zhongxiang, et al.
Published: (2026)

MORPHOGEN: A Multilingual Benchmark for Evaluating Gender-Aware Morphological Generation
by: Agarwal, Mehul, et al.
Published: (2026)

Can LLMs Generate High-Quality Task-Specific Conversations?
by: Li, Shengqi, et al.
Published: (2025)

When, What, and How: Rethinking Retrieval-Enhanced Speculative Decoding
by: Fang, Min, et al.
Published: (2025)

Presumed Cultural Identity: How Names Shape LLM Responses
by: Pawar, Siddhesh, et al.
Published: (2025)

What Layers When: Learning to Skip Compute in LLMs with Residual Gates
by: Laitenberger, Filipe, et al.
Published: (2025)

Retracing the Past: LLMs Emit Training Data When They Get Lost
by: Ko, Myeongseob, et al.
Published: (2025)

When Facts Change: Probing LLMs on Evolving Knowledge with evolveQA
by: Nakshatri, Nishanth Sridhar, et al.
Published: (2025)

When LLMs Benchmark Themselves: Deconstructing Self-Bias in Automated Evaluation
by: Xu, Wenda, et al.
Published: (2025)

When Weak LLMs Speak with Confidence, Preference Alignment Gets Stronger
by: Afzali, Amirabbas, et al.
Published: (2026)

Question Answering on Patient Medical Records with Private Fine-Tuned LLMs
by: Kothari, Sara, et al.
Published: (2025)

Learning When to Retrieve, What to Rewrite, and How to Respond in Conversational QA
by: Roy, Nirmal, et al.
Published: (2024)

When Verification Fails: How Compositionally Infeasible Claims Escape Rejection
by: Liu, Muxin, et al.
Published: (2026)

Systematic Evaluation of Long-Context LLMs on Financial Concepts
by: Gupta, Lavanya, et al.
Published: (2024)