Six state-of-the-art LLMs systematically prefer Standard American English over AAE continuations, and a training-free activation steering method reduces this bias 5-20x more than prompting while preserving fluency.
Locating and Editing Factual Associations in
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
years
2026 6representative citing papers
Finite-answer projections of continuation probabilities stabilize before the answer is parseable, showing 17-31 token mean lead in delayed-verdict tasks with Qwen3-4B-Instruct.
A 4B compiler model generates LoRA adapters from natural-language specs, enabling a frozen 0.6B interpreter to match Qwen3-32B performance on fuzzy text tasks at 50× less memory.
MixSD uses dynamic mixing of the model's expert and naive conditionals to create distribution-aligned supervision that improves the memorization-retention tradeoff over standard SFT.
In LLM agents, memory routing circuits emerge at 0.6B scale while content circuits appear only at 4B, and write/read operations recruit a pre-existing late-layer context hub instead of creating a new one, enabling a 76% accurate unsupervised failure diagnostic.
LOCA induces refusal on jailbroken prompts by iteratively patching a handful of token-specific SAE features, with 6–8 patches for Llama and 12–16 for Gemma.
citing papers explorer
-
LLMs Silently Correct African American English: Auditing and Mitigating Dialect Bias via Activation Steering
Six state-of-the-art LLMs systematically prefer Standard American English over AAE continuations, and a training-free activation steering method reduces this bias 5-20x more than prompting while preserving fluency.
-
When Does a Language Model Commit? A Finite-Answer Theory of Pre-Verbalization Commitment
Finite-answer projections of continuation probabilities stabilize before the answer is parseable, showing 17-31 token mean lead in delayed-verdict tasks with Qwen3-4B-Instruct.
-
Program-as-Weights: A Programming Paradigm for Fuzzy Functions
A 4B compiler model generates LoRA adapters from natural-language specs, enabling a frozen 0.6B interpreter to match Qwen3-32B performance on fuzzy text tasks at 50× less memory.
-
MixSD: Mixed Contextual Self-Distillation for Knowledge Injection
MixSD uses dynamic mixing of the model's expert and naive conditionals to create distribution-aligned supervision that improves the memorization-retention tradeoff over standard SFT.
-
What Happens Inside Agent Memory? Circuit Analysis from Emergence to Diagnosis
In LLM agents, memory routing circuits emerge at 0.6B scale while content circuits appear only at 4B, and write/read operations recruit a pre-existing late-layer context hub instead of creating a new one, enabling a 76% accurate unsupervised failure diagnostic.
-
Minimal, Local, Causal Explanations for Jailbreak Success in Large Language Models
LOCA induces refusal on jailbroken prompts by iteratively patching a handful of token-specific SAE features, with 6–8 patches for Llama and 12–16 for Gemma.