Hugging Face Trending Papers
Interpreting Model Reasoning through Behavioral Counterfactuals
Read the original on Hugging Face Trending Papers →The Flow has not summarised this story yet — read it at Hugging Face Trending Papers.
Related stories
arXiv Machine Learning
5d ago
Interpreting Model Reasoning through Behavioral Counterfactuals
The paper introduces a method for analyzing how interventions in a prompt affect a language model’s reasoning by using behavioral counterfactuals. It trains a module to edit the residual stream so that the next-token distributions for a fixed response match between a source and a counterfactual prompt, outperforming standard activation reconstruction. The resulting attribution maps and visualizations help identify causally important locations for counterfactual recovery and guide hypothesis testing with activation patching, as demonstrated in a case study on Gemma-4-E4B-it and DeepSeek-R1-0528-Qwen3-8B.
arXiv AI
Aug 18
Would this change your answer? Evaluating Explanations of LLM Behavior In The Wild with Counterfactual Experiments
arXiv:2608. 16747v1 Announce Type: cross Abstract: Many areas of AI research, such as language model interpretability and chain of thought faithfulness, seek to explain model behaviors.
arXiv AI
Aug 10
Counterfactual Simulation Training for Chain-of-Thought Faithfulness
arXiv:2602. 20710v2 Announce Type: replace Abstract: Inspecting Chain-of-Thought reasoning is among the most common means of understanding why an LLM produced its output.
arXiv AI
Jul 1
Introspective Coupling: Self-Explanation Training Tracks Behavioral Change Despite Fixed Supervision
arXiv:2606. 32038v1 Announce Type: cross Abstract: When does training language models (LMs) to generate explanations of their predictions yield faithful introspection, rather than superficial imitation?
arXiv AI
Jul 24
Post-Hoc Reasoning in Chain of Thought: Decoding and Steering Pre-Committed Answers
arXiv:2603. 01437v2 Announce Type: replace Abstract: As chain of thought (CoT) has become central to scaling reasoning capabilities in large language models (LLMs), it has also emerged as a promising tool for interpretability, suggesting the opportunity to understand model decisions through verbalized reasoning.
arXiv AI
Oct 12
The Evidence Anchoring Boundary: Where Visual Interventions Matter and When They Help in Multimodal LLMs
The paper investigates how visual interventions—training‑free edits applied to multimodal large language models—affect hallucinations and prediction quality. By applying a visual‑residual edit across successive decoder layers and matching its RMS strength, the authors identify an Evidence Anchoring Boundary (EAB) where the intervention’s influence drops sharply; effects before the EAB are 13.78–26.10× larger than after it. They further compare counterfactual subtraction across different question types and introduce calibrated selective routing, which only alters predictions when calibration supports an intervention, improving attribute accuracy by 1.56–1.81 percentage points while avoiding the harms of universal subtraction.