# Hidden Heroes and Gradient Bloats: Layer-Wise Redundancy Inverts Attribution in Transformers

arXiv: https://arxiv.org/abs/2602.01442 · HTML: https://mlsanity.com/paper/2602.01442/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Circuits and mechanistic analysis, one of 13 directions in Mechanistic interpretability.
- Builds on 10 of its 12 references that are in the map.
- First posted on arXiv 1 Feb 2026.

## Placed in

- Mechanistic interpretability / [Circuits and mechanistic analysis](https://mlsanity.com/mechinterp/circuits-and-mechanistic-analysis/)

## Builds on (most cited first, 10 of 10)

- [Learning both Weights and Connections for Efficient Neural Networks](https://mlsanity.com/paper/1506.02626/) (2015)
- [Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small](https://mlsanity.com/paper/2211.00593/) (2022)
- [Sanity Checks for Saliency Maps](https://mlsanity.com/paper/1810.03292/) (2018)
- [Progress measures for grokking via mechanistic interpretability](https://mlsanity.com/paper/2301.05217/) (2023)
- [Are Sixteen Heads Really Better than One?](https://mlsanity.com/paper/1905.10650/) (2019)
- [A Simple and Effective Pruning Approach for Large Language Models](https://mlsanity.com/paper/2306.11695/) (2023)
- [Towards Automated Circuit Discovery for Mechanistic Interpretability](https://mlsanity.com/paper/2304.14997/) (2023)
- [A Benchmark for Interpretability Methods in Deep Neural Networks](https://mlsanity.com/paper/1806.10758/) (2018)
- [Massive Activations in Large Language Models](https://mlsanity.com/paper/2402.17762/) (2024)
- [Explaining grokking through circuit efficiency](https://mlsanity.com/paper/2309.02390/) (2023)

## Built on by (most cited first, 1 of 1)

- [Conditional Co-Ablation: Recovering Self-Repair Backups in Transformer Circuits](https://mlsanity.com/paper/2607.01940/) (2026)

## Abstract (arXiv)

Gradient-based attribution is the workhorse of mechanistic interpretability, yet whether it reliably tracks causal importance at the component level remains largely untested. We causally evaluate this assumption across two algorithmic tasks and up to 10 random seeds, uncovering a systematic, layer-wise failure: gradient attribution consistently overvalues early-layer Gradient Bloats and undervalues late-layer Hidden Heroes. Rank correlation collapses from ρ = 0.72 on sequence reversal to 0.27 on sequence sorting, reaching ρ = -0.18 in individual seeds. This failure stems from first-order gradient attribution's inability to detect collective redundancy: joint Bloat ablation causes 14× greater damage than individual results predict. Consequently, Bloats dominate gradient rankings despite negligible functional impact, while ablating Hidden Heroes destroys OOD accuracy (-36.4% ± 22.8%). This systematic inversion of early-layer feature extraction and late-layer computation motivates causal validation as a prerequisite for circuit-level claims.