Concept Influence: Leveraging Interpretability to Improve Performance and Efficiency in Training Data Attribution
cs.AI, stat.ML
Submitted: 2026-02-16
Updated: 2026-11-16
Terminology
Sources
- The Llama 3 Herd of Models
- Towards Tracing Factual Knowledge in Language Models Back to the Training Data
- Training Feature Attribution for Vision Models
- Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs
- Exemplary Natural Images Explain CNN Activations Better than State-of-the-Art Feature Visualization
- Persona Vectors: Monitoring and Controlling Character Traits in Language Models
- IF-GUIDE: Influence Function-Guided Detoxification of LLMs
- What I Cannot Predict, I Do Not Understand: A Human-Centered Evaluation Framework for Explainability Methods
- TinyStories: How Small Can Language Models Be and Still Speak Coherent English?
- Studying Large Language Model Generalization with Influence Functions
- Explaining Black Box Predictions and Unveiling Data Artifacts through Influence Functions
- Position: Curvature Matrices Should Be Democratized via Linear Operators
- Evaluating Explainable AI: Which Algorithmic Explanations Help Users Predict Model Behavior?
- Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training
- MT-Eval: A Multi-Turn Capabilities Evaluation Benchmark for Large Language Models
- Do Influence Functions Work on Large Language Models?
- Evaluating Saliency Explanations in NLP by Crowdsourcing
- Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps
- Do Users Benefit From Interpretable Vision? A User Study, Baseline, And Dataset
- Convergent Linear Representations of Emergent Misalignment
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents