{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/spex-scaling-feature-interaction-explanations","title":"SPEX: Scaling Feature Interaction Explanations for LLMs","arxiv_id":"2502.13870","date":"2025-02-19","proceeding":null,"authors":["Justin Singh Kang","Landon Butler","Abhineet Agarwal","Yigit Efe Erginbas","Ramtin Pedarsani","Kannan Ramchandran","Bin Yu"],"abstract":"Large language models (LLMs) have revolutionized machine learning due to their ability to capture complex interactions between input features. Popular post-hoc explanation methods like SHAP provide marginal feature attributions, while their extensions to interaction importances only scale to small input lengths ($\\approx 20$). We propose Spectral Explainer (SPEX), a model-agnostic interaction attribution algorithm that efficiently scales to large input lengths ($\\approx 1000)$. SPEX exploits underlying natural sparsity among interactions -- common in real-world data -- and applies a sparse Fourier transform using a channel decoding algorithm to efficiently identify important interactions. We perform experiments across three difficult long-context datasets that require LLMs to utilize interactions between inputs to complete the task. For large inputs, SPEX outperforms marginal attribution methods by up to 20% in terms of faithfully reconstructing LLM outputs. Further, SPEX successfully identifies key features and interactions that strongly influence model output. For one of our datasets, HotpotQA, SPEX provides interactions that align with human annotations. Finally, we use our model-agnostic approach to generate explanations to demonstrate abstract reasoning in closed-source LLMs (GPT-4o mini) and compositional reasoning in vision-language models.","url_abs":"https://arxiv.org/abs/2502.13870v1","url_pdf":"https://arxiv.org/pdf/2502.13870v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"spex-scaling-feature-interaction-explanations","repo_url":"https://github.com/basics-lab/spectral-explain","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[],"methods":[{"method_slug":"align","method_name":"ALIGN"},{"method_slug":"shap","method_name":"SHAP"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2502.13870","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2502.13870"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/basics-lab/spectral-explain","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"unverified":10},"by_repo_kind":{"official":{"samples":10,"ran":0,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"8306ee06d44ca5f2","entry":"estimate_r2","repo":"basics-lab/spectral-explain","repo_kind":"official","path":"src/spectralexplain/utils.py","file_url":"https://github.com/basics-lab/spectral-explain/blob/HEAD/src/spectralexplain/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"8306ee06d44ca5f2"}},{"code_sha256_prefix":"6c55879dd8a805c3","entry":"eval_function","repo":"basics-lab/spectral-explain","repo_kind":"official","path":"experiments/removal.py","file_url":"https://github.com/basics-lab/spectral-explain/blob/HEAD/experiments/removal.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"6c55879dd8a805c3"}},{"code_sha256_prefix":"7d19628adb34ce36","entry":"fourier_to_mobius","repo":"basics-lab/spectral-explain","repo_kind":"official","path":"src/spectralexplain/utils.py","file_url":"https://github.com/basics-lab/spectral-explain/blob/HEAD/src/spectralexplain/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"7d19628adb34ce36"}},{"code_sha256_prefix":"4f594e42f5b0b233","entry":"get_num_samples","repo":"basics-lab/spectral-explain","repo_kind":"official","path":"src/spectralexplain/support_recovery.py","file_url":"https://github.com/basics-lab/spectral-explain/blob/HEAD/src/spectralexplain/support_recovery.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"4f594e42f5b0b233"}},{"code_sha256_prefix":"9107e21a5ad6e23e","entry":"make_dependancy_tree","repo":"basics-lab/spectral-explain","repo_kind":"official","path":"experiments/removal.py","file_url":"https://github.com/basics-lab/spectral-explain/blob/HEAD/experiments/removal.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"9107e21a5ad6e23e"}},{"code_sha256_prefix":"b1f0be90b2b91abd","entry":"powerset","repo":"basics-lab/spectral-explain","repo_kind":"official","path":"src/spectralexplain/utils.py","file_url":"https://github.com/basics-lab/spectral-explain/blob/HEAD/src/spectralexplain/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"b1f0be90b2b91abd"}},{"code_sha256_prefix":"c70c75284b91e46f","entry":"run_and_evaluate_method","repo":"basics-lab/spectral-explain","repo_kind":"official","path":"experiments/faithfulness.py","file_url":"https://github.com/basics-lab/spectral-explain/blob/HEAD/experiments/faithfulness.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"c70c75284b91e46f"}},{"code_sha256_prefix":"1d7a8fb3797b5107","entry":"scaler_classification","repo":"basics-lab/spectral-explain","repo_kind":"official","path":"experiments/dataloader.py","file_url":"https://github.com/basics-lab/spectral-explain/blob/HEAD/experiments/dataloader.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"1d7a8fb3797b5107"}},{"code_sha256_prefix":"335822cf082d847d","entry":"scaler_regression","repo":"basics-lab/spectral-explain","repo_kind":"official","path":"experiments/dataloader.py","file_url":"https://github.com/basics-lab/spectral-explain/blob/HEAD/experiments/dataloader.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"335822cf082d847d"}},{"code_sha256_prefix":"f949f31b5f8130a7","entry":"vec_to_index","repo":"basics-lab/spectral-explain","repo_kind":"official","path":"experiments/removal.py","file_url":"https://github.com/basics-lab/spectral-explain/blob/HEAD/experiments/removal.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"f949f31b5f8130a7"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}