Localized model averaging with covariate-dependent weights achieves asymptotic optimality and weight consistency for combining pre-trained models under a general loss framework.
When do prompting and prefix-tuning work? a theory of capabilities and limitations
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
Fine-tuned decoder-only LLMs fall into a Semantic Trap on vulnerability detection, achieving high scores on unpaired normal code but failing on paired vulnerable-patched code, semantic perturbations, and gap analysis, while reasoning supervision reduces symptoms at the cost of recall.
PrefixMemory-Tuning decouples the prefix from attention to overcome performance limits of traditional prefix-tuning and reaches competitive results with modern PEFT methods on LLM adaptation benchmarks.
A comprehensive survey of PEFT algorithms for large models, covering their performance, overhead, applications, and real-world system implementations.
citing papers explorer
-
Combining pre-trained models via localized model averaging
Localized model averaging with covariate-dependent weights achieves asymptotic optimality and weight consistency for combining pre-trained models under a general loss framework.
-
Do Fine-Tuned LLMs Understand Vulnerabilities? An Investigation into the Semantic Trap
Fine-tuned decoder-only LLMs fall into a Semantic Trap on vulnerability detection, achieving high scores on unpaired normal code but failing on paired vulnerable-patched code, semantic perturbations, and gap analysis, while reasoning supervision reduces symptoms at the cost of recall.
-
PrefixMemory-Tuning: Modernizing Prefix-Tuning by Decoupling the Prefix from Attention
PrefixMemory-Tuning decouples the prefix from attention to overcome performance limits of traditional prefix-tuning and reaches competitive results with modern PEFT methods on LLM adaptation benchmarks.
-
Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Survey
A comprehensive survey of PEFT algorithms for large models, covering their performance, overhead, applications, and real-world system implementations.