{"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/emb-gam-an-interpretable-and-efficient","title":"Augmenting Interpretable Models with LLMs during Training","arxiv_id":"2209.11799","date":"2022-09-23","proceeding":null,"authors":["Chandan Singh","Armin Askari","Rich Caruana","Jianfeng Gao"],"abstract":"Recent large language models (LLMs) have demonstrated remarkable prediction performance for a growing array of tasks. However, their proliferation into high-stakes domains (e.g. medicine) and compute-limited settings has created a burgeoning need for interpretability and efficiency. We address this need by proposing Augmented Interpretable Models (Aug-imodels), a framework for leveraging the knowledge learned by LLMs to build extremely efficient and interpretable models. Aug-imodels use LLMs during fitting but not during inference, allowing complete transparency and often a speed/memory improvement of greater than 1,000x for inference compared to LLMs. We explore two instantiations of Aug-imodels in natural-language processing: (i) Aug-GAM, which augments a generalized additive model with decoupled embeddings from an LLM and (ii) Aug-Tree, which augments a decision tree with LLM feature expansions. Across a variety of text-classification datasets, both outperform their non-augmented counterparts. Aug-GAM can even outperform much larger models (e.g. a 6-billion parameter GPT-J model), despite having 10,000x fewer parameters and being fully transparent. We further explore Aug-imodels in a natural-language fMRI study, where they generate interesting interpretations from scientific data. All code for using Aug-imodels and reproducing results is made available on Github.","url_abs":"https://arxiv.org/abs/2209.11799v3","url_pdf":"https://arxiv.org/pdf/2209.11799v3.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":"emb-gam-an-interpretable-and-efficient","repo_url":"https://github.com/csinva/emb-gam","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"emb-gam-an-interpretable-and-efficient","repo_url":"https://github.com/csinva/imodelsX","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"emb-gam-an-interpretable-and-efficient","repo_url":"https://github.com/microsoft/augmented-interpretable-models","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"emb-gam-an-interpretable-and-efficient","repo_url":"https://github.com/microsoft/automated-explanations","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"additive-models","task_name":"Additive models"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"sentiment-analysis","task_name":"Sentiment Analysis"},{"task_slug":"text-classification","task_name":"Text Classification"},{"task_slug":"text-classification-1","task_name":"text-classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=2209.11799","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2209.11799"}},"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/csinva/imodelsX","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/microsoft/automated-explanations","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/csinva/emb-gam","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/microsoft/augmented-interpretable-models","reach":null}],"summary":{"ran_draft_wrong":2,"unverified":10},"by_repo_kind":{"official":{"samples":12,"ran":2,"repositories":2}},"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":10,"samples":[{"code_sha256_prefix":"19363da9b2e0a8e4","entry":"add_computational_args","repo":"microsoft/augmented-interpretable-models","repo_kind":"official","path":"augtree/experiments/01_train_model.py","file_url":"https://github.com/microsoft/augmented-interpretable-models/blob/HEAD/augtree/experiments/01_train_model.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"19363da9b2e0a8e4"}},{"code_sha256_prefix":"9a9d5e7dbc61cc12","entry":"add_main_args","repo":"microsoft/augmented-interpretable-models","repo_kind":"official","path":"augtree/experiments/01_train_model.py","file_url":"https://github.com/microsoft/augmented-interpretable-models/blob/HEAD/augtree/experiments/01_train_model.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"9a9d5e7dbc61cc12"}},{"code_sha256_prefix":"406e7b11b9a38bf6","entry":"add_computational_args","repo":"csinva/emb-gam","repo_kind":"official","path":"augdistill/experiments/01_eval.py","file_url":"https://github.com/csinva/emb-gam/blob/HEAD/augdistill/experiments/01_eval.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":false,"mcp_get_code":{"code_sha256":"406e7b11b9a38bf6"}},{"code_sha256_prefix":"63743868776d4aa1","entry":"add_main_args","repo":"csinva/emb-gam","repo_kind":"official","path":"augdistill/experiments/01_eval.py","file_url":"https://github.com/csinva/emb-gam/blob/HEAD/augdistill/experiments/01_eval.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":false,"mcp_get_code":{"code_sha256":"63743868776d4aa1"}},{"code_sha256_prefix":"94fc12cee266d05f","entry":"average_seeds","repo":"csinva/emb-gam","repo_kind":"official","path":"auggam/auggam/analyze_helper.py","file_url":"https://github.com/csinva/emb-gam/blob/HEAD/auggam/auggam/analyze_helper.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":false,"mcp_get_code":{"code_sha256":"94fc12cee266d05f"}},{"code_sha256_prefix":"4d8fce0e5de37264","entry":"corrplot_max_abs_unigrams","repo":"csinva/emb-gam","repo_kind":"official","path":"auggam/auggam/analyze_helper.py","file_url":"https://github.com/csinva/emb-gam/blob/HEAD/auggam/auggam/analyze_helper.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":false,"mcp_get_code":{"code_sha256":"4d8fce0e5de37264"}},{"code_sha256_prefix":"1bf3bdbe0145f356","entry":"embed_and_sum_function","repo":"csinva/emb-gam","repo_kind":"official","path":"auggam/auggam/embed.py","file_url":"https://github.com/csinva/emb-gam/blob/HEAD/auggam/auggam/embed.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":false,"mcp_get_code":{"code_sha256":"1bf3bdbe0145f356"}},{"code_sha256_prefix":"1999a51d3bcebcb3","entry":"generate_ngrams_list","repo":"csinva/emb-gam","repo_kind":"official","path":"auggam/auggam/embed.py","file_url":"https://github.com/csinva/emb-gam/blob/HEAD/auggam/auggam/embed.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":false,"mcp_get_code":{"code_sha256":"1999a51d3bcebcb3"}},{"code_sha256_prefix":"6fbf6dc523fe795c","entry":"get_dir_name","repo":"csinva/emb-gam","repo_kind":"official","path":"auggam/auggam/data.py","file_url":"https://github.com/csinva/emb-gam/blob/HEAD/auggam/auggam/data.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":false,"mcp_get_code":{"code_sha256":"6fbf6dc523fe795c"}},{"code_sha256_prefix":"cddd3f8fab150afb","entry":"load_fitted_results","repo":"csinva/emb-gam","repo_kind":"official","path":"auggam/auggam/data.py","file_url":"https://github.com/csinva/emb-gam/blob/HEAD/auggam/auggam/data.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":false,"mcp_get_code":{"code_sha256":"cddd3f8fab150afb"}},{"code_sha256_prefix":"ce5f966b7b6f6781","entry":"preprocess_gpt_token_batch","repo":"csinva/emb-gam","repo_kind":"official","path":"auggam/auggam/embed.py","file_url":"https://github.com/csinva/emb-gam/blob/HEAD/auggam/auggam/embed.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":false,"mcp_get_code":{"code_sha256":"ce5f966b7b6f6781"}},{"code_sha256_prefix":"751633c4c0a353c8","entry":"process_data_and_args","repo":"csinva/emb-gam","repo_kind":"official","path":"auggam/auggam/data.py","file_url":"https://github.com/csinva/emb-gam/blob/HEAD/auggam/auggam/data.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":false,"mcp_get_code":{"code_sha256":"751633c4c0a353c8"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}