[Google AI Overview at the query, signed out and incognito (operator defaults of 2026-10-01 and 2026-10-14), 2026-10-29. The paste is the Overview expanded (the 'AI Mode Conversation' view, with the source cards that follow the answer); the 'You said' line carries the query twice, run together, and is given once here. Inline links are Google redirect URLs, kept as pasted.]

[QUERENT] AttriBench

[SYSTEM] The term AttriBench primarily refers to two distinct tools in the field of artificial intelligence and machine learning:
1. AttriBench (Large Language Model Evaluation)
Released in 2026, AttriBench is the first fame- and demographically-balanced quote attribution benchmark dataset for Large Language Models (LLMs). [[1](https://www.google.com/goto?url=CAESWAHrOzAVkIxAkiiji0QZN3koE1DCdfc3CEbx4E-ATSPWSUnAE_LZatEgP-0kxbgWjM9GK6pJLPpIly8bptwtVAsd7GeX6nDmUOuZqdkS4uPBLtSAaMjvsuI)]

* The Purpose: As LLMs are increasingly deployed for search and information retrieval, it is critical that they accurately attribute quotes to their original authors. AttriBench evaluates how well models do this without exhibiting demographic bias. [[1](https://www.google.com/goto?url=CAESWAHrOzAVkIxAkiiji0QZN3koE1DCdfc3CEbx4E-ATSPWSUnAE_LZatEgP-0kxbgWjM9GK6pJLPpIly8bptwtVAsd7GeX6nDmUOuZqdkS4uPBLtSAaMjvsuI)]
* Key Findings: Evaluations using this dataset revealed that quote attribution remains highly challenging even for frontier models. It highlighted systematic disparities in accuracy across race, gender, and intersectional groups, as well as a failure mode known as suppression—where a model completely omits attribution even though it possesses the authorship data. [[1](https://www.google.com/goto?url=CAESWAHrOzAVkIxAkiiji0QZN3koE1DCdfc3CEbx4E-ATSPWSUnAE_LZatEgP-0kxbgWjM9GK6pJLPpIly8bptwtVAsd7GeX6nDmUOuZqdkS4uPBLtSAaMjvsuI)]
* Availability: The data repository can be explored directly on [Hugging Face](https://www.google.com/goto?url=CAESegHrOzAVFaeIQ7-aMOamfHzfhu2E03kvd9_u_RcgCADRXYNXHQh7F7vkX1__LEpPcKSNIQE7p1oHLf7b6hJ-XNdntf48qhHcTfm-7j34We_QFcZ4pUQv6KC0x_yw_1kp4tKq48NBUMq-8e47MnGmibQ4oquvQNq5ZBKb). [[1](https://www.google.com/goto?url=CAESegHrOzAVFaeIQ7-aMOamfHzfhu2E03kvd9_u_RcgCADRXYNXHQh7F7vkX1__LEpPcKSNIQE7p1oHLf7b6hJ-XNdntf48qhHcTfm-7j34We_QFcZ4pUQv6KC0x_yw_1kp4tKq48NBUMq-8e47MnGmibQ4oquvQNq5ZBKb)]

2. AttriBench (Explainable AI / Feature Attribution)
Originally released in 2023, attribench is a PyTorch-based open-source Python library used to benchmark explainable AI (XAI) techniques. [[1](https://www.google.com/goto?url=CAESXAHrOzAVV3mQXXNwbzJCrDspcwiFuSjkIME4y0tt8kUa37mXffVbwQiBDBX3SpkXFSB8GbC--HODQcFtUhxmsw7pkfg15HBf0NxAZJtK_KkkogKViGwNPrRU3xQN), [2](https://www.google.com/goto?url=CAESWgHrOzAVp7IeLhzxm3ByrvUus_SmZ3Nqq71kmYOanc0nDlh28UVCNcHRqdrahv6GnFoYCpOat0kVp3lEQZ_l8hxPLHYsiIBGGlV--sOnZkgwfzl4Z09hEU-caQ)]

* The Purpose: When machine learning models make predictions, feature attribution methods (like Integrated Gradients, SHAP, or Saliency Maps) highlight which parts of the input data (e.g., pixels in an image) were most important.
* Functionality: This version of AttriBench provides a standardized framework and functional metrics to evaluate and compare the quality, robustness, and reliability of those feature attribution maps. [[1](https://www.google.com/goto?url=CAESXAHrOzAVV3mQXXNwbzJCrDspcwiFuSjkIME4y0tt8kUa37mXffVbwQiBDBX3SpkXFSB8GbC--HODQcFtUhxmsw7pkfg15HBf0NxAZJtK_KkkogKViGwNPrRU3xQN)]
* Availability: The library can be installed via [PyPI](https://www.google.com/goto?url=CAESXAHrOzAVV3mQXXNwbzJCrDspcwiFuSjkIME4y0tt8kUa37mXffVbwQiBDBX3SpkXFSB8GbC--HODQcFtUhxmsw7pkfg15HBf0NxAZJtK_KkkogKViGwNPrRU3xQN) and its technical documentation is hosted on Read the Docs. [[1](https://www.google.com/goto?url=CAESWgHrOzAVp7IeLhzxm3ByrvUus_SmZ3Nqq71kmYOanc0nDlh28UVCNcHRqdrahv6GnFoYCpOat0kVp3lEQZ_l8hxPLHYsiIBGGlV--sOnZkgwfzl4Z09hEU-caQ), [2](https://www.google.com/goto?url=CAESXAHrOzAVV3mQXXNwbzJCrDspcwiFuSjkIME4y0tt8kUa37mXffVbwQiBDBX3SpkXFSB8GbC--HODQcFtUhxmsw7pkfg15HBf0NxAZJtK_KkkogKViGwNPrRU3xQN)]

(Note: In traditional finance, AttriBench also refers to interactive simulation exercises used by portfolio managers to practice Brinson equity return attribution.) [[1](https://www.google.com/goto?url=CAESeAHrOzAVdPoS8zwrUjIZvyaIinzJTJnNA85UNL9cYHdud8EP5e0Cl15jiCi9WBsaAT_2gqb4Z2fGpXzepnd6fpzsVeiTLtQ75QmVmZ-1wPKDCMQR6IH513Pt0LcsNHxYuPqrv1NBQXOIlTDTHUTHOM81Nj9n7TNtag)]
To give you the most relevant information, are you looking into AttriBench for evaluating bias in LLM text generation, or are you looking to test explainable AI methods in PyTorch?
arXiv
[2604.05224] Attribution Bias in Large Language Models - arXiv
Apr 6, 2026 — In this work, we introduce AttriBench, the first fame- and demographically-balanced quote attribution benchmark dataset. By explicitly balancing author fame and...
pypi.org
attribench · PyPI
Aug 25, 2023 — AttriBench is a Pytorch-based implementation of several metrics for the evaluation of feature attribution maps and methods. AttriBench provides a functional ...
attribench.readthedocs.io
Welcome to AttriBench's documentation! — AttriBench 0.0.1 ...
AttriBench is a Pytorch-based benchmarking tool for feature attribution-based explanations of machine learning models. Note. This documentation is still under .
huggingface.co
bermaneh/AttriBench at main - Hugging Face
We're on a journey to advance and democratize artificial intelligence through open source and open science.
[www.eurekasimulations.com](https://www.eurekasimulations.com)
AttriBench: Investment Performance Measurement & Attribution
Interactive exercise where participants decompose a global equity fund's 12-month excess returns using Brinson attribution, analyzing allocation…