{"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/multi-property-steering-of-large-language","title":"Multi-property Steering of Large Language Models with Dynamic Activation Composition","arxiv_id":"2406.17563","date":"2024-06-25","proceeding":null,"authors":["Daniel Scalena","Gabriele Sarti","Malvina Nissim"],"abstract":"Activation steering methods were shown to be effective in conditioning language model generation by additively intervening over models' intermediate representations. However, the evaluation of these techniques has so far been limited to single conditioning properties and synthetic settings. In this work, we conduct a comprehensive evaluation of various activation steering strategies, highlighting the property-dependent nature of optimal parameters to ensure a robust effect throughout generation. To address this issue, we propose Dynamic Activation Composition, an information-theoretic approach to modulate the steering intensity of one or more properties throughout generation. Our experiments on multi-property steering show that our method successfully maintains high conditioning while minimizing the impact of conditioning on generation fluency.","url_abs":"https://arxiv.org/abs/2406.17563v1","url_pdf":"https://arxiv.org/pdf/2406.17563v1.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":"multi-property-steering-of-large-language","repo_url":"https://github.com/danielsc4/dynamic-activation-composition","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2406.17563","atlas_url":"https://app.syntology.ai/?focus=2406.17563","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2406.17563"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-25T09:33:49+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/danielsc4/dynamic-activation-composition","reach":{"status":"ok"}}],"summary":{"ran":5,"unverified":3},"by_repo_kind":{"official":{"samples":8,"ran":5,"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":8,"samples":[{"code_sha256_prefix":"94f7e73dbbc7745a","entry":"build_prompt_txt","repo":"danielsc4/dynamic-activation-composition","repo_kind":"official","path":"src/utils/prompt_helper.py","file_url":"https://github.com/danielsc4/dynamic-activation-composition/blob/HEAD/src/utils/prompt_helper.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"94f7e73dbbc7745a"}},{"code_sha256_prefix":"b1966b1d268c147b","entry":"calculate_accuracy","repo":"danielsc4/dynamic-activation-composition","repo_kind":"official","path":"src/utils/eval/general_evaluation_fun.py","file_url":"https://github.com/danielsc4/dynamic-activation-composition/blob/HEAD/src/utils/eval/general_evaluation_fun.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"b1966b1d268c147b"}},{"code_sha256_prefix":"2a309e3201fa645a","entry":"load_dataset","repo":"danielsc4/dynamic-activation-composition","repo_kind":"official","path":"src/utils/prompt_helper.py","file_url":"https://github.com/danielsc4/dynamic-activation-composition/blob/HEAD/src/utils/prompt_helper.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"2a309e3201fa645a"}},{"code_sha256_prefix":"c8a743848dd737a6","entry":"split_activation","repo":"danielsc4/dynamic-activation-composition","repo_kind":"official","path":"src/extraction.py","file_url":"https://github.com/danielsc4/dynamic-activation-composition/blob/HEAD/src/extraction.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"c8a743848dd737a6"}},{"code_sha256_prefix":"b5d0a88b37f44fb0","entry":"tokenize_from_template","repo":"danielsc4/dynamic-activation-composition","repo_kind":"official","path":"src/utils/prompt_helper.py","file_url":"https://github.com/danielsc4/dynamic-activation-composition/blob/HEAD/src/utils/prompt_helper.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"b5d0a88b37f44fb0"}},{"code_sha256_prefix":"59961b2712da82ad","entry":"filter_activations","repo":"danielsc4/dynamic-activation-composition","repo_kind":"official","path":"src/extraction.py","file_url":"https://github.com/danielsc4/dynamic-activation-composition/blob/HEAD/src/extraction.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"59961b2712da82ad"}},{"code_sha256_prefix":"44925cf3831bfe79","entry":"rgetattr","repo":"danielsc4/dynamic-activation-composition","repo_kind":"official","path":"src/utils/model_utils.py","file_url":"https://github.com/danielsc4/dynamic-activation-composition/blob/HEAD/src/utils/model_utils.py","link_basis":"plan_row","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"44925cf3831bfe79"}},{"code_sha256_prefix":"f9de4cdd8a624801","entry":"rsetattr","repo":"danielsc4/dynamic-activation-composition","repo_kind":"official","path":"src/utils/model_utils.py","file_url":"https://github.com/danielsc4/dynamic-activation-composition/blob/HEAD/src/utils/model_utils.py","link_basis":"plan_row","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"f9de4cdd8a624801"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}