{"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/subzero-random-subspace-zeroth-order","title":"Zeroth-Order Fine-Tuning of LLMs in Random Subspaces","arxiv_id":"2410.08989","date":"2024-10-11","proceeding":null,"authors":["Ziming Yu","Pan Zhou","Sike Wang","Jia Li","Hua Huang"],"abstract":"Fine-tuning Large Language Models (LLMs) has proven effective for a variety of downstream tasks. However, as LLMs grow in size, the memory demands for backpropagation become increasingly prohibitive. Zeroth-order (ZO) optimization methods offer a memory-efficient alternative by using forward passes to estimate gradients, but the variance of gradient estimates typically scales linearly with the model's parameter dimension$\\unicode{x2013}$a significant issue for LLMs. In this paper, we propose the random Subspace Zeroth-order (SubZero) optimization to address the challenges posed by LLMs' high dimensionality. We introduce a low-rank perturbation tailored for LLMs that significantly reduces memory consumption while improving training performance. Additionally, we prove that our gradient estimation closely approximates the backpropagation gradient, exhibits lower variance than traditional ZO methods, and ensures convergence when combined with SGD. Experimental results show that SubZero enhances fine-tuning performance and achieves faster convergence compared to standard ZO approaches like MeZO across various language modeling tasks.","url_abs":"https://arxiv.org/abs/2410.08989v2","url_pdf":"https://arxiv.org/pdf/2410.08989v2.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":"subzero-random-subspace-zeroth-order","repo_url":"https://github.com/zimingyy/subzero","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"GPL-3.0"}}],"tasks":[{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"}],"methods":[{"method_slug":"sgd","method_name":"SGD"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2410.08989","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2410.08989"}},"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/zimingyy/subzero","reach":{"status":"ok","spdx":"GPL-3.0"}}],"summary":{"ran_fixture":2,"ran":4,"ran_draft_wrong":3,"unverified":6},"by_repo_kind":{"official":{"samples":15,"ran":9,"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":15,"samples":[{"code_sha256_prefix":"30d7eec482ebf6b1","entry":"repeat_kv","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/modeling_llama.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/modeling_llama.py","link_basis":"harvester_set","language":"python","status":"ran_fixture","verification_level":2,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"30d7eec482ebf6b1"}},{"code_sha256_prefix":"d61c483a3c2b3156","entry":"apply_rotary_pos_emb","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/modeling_llama.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/modeling_llama.py","link_basis":"harvester_set","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"d61c483a3c2b3156"}},{"code_sha256_prefix":"f53e3216ab15d617","entry":"encode_prompt","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/utils.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"f53e3216ab15d617"}},{"code_sha256_prefix":"3bfc0eb3499ede63","entry":"fast_svd_method_v1","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/trainer.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/trainer.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"3bfc0eb3499ede63"}},{"code_sha256_prefix":"b5adfd6ac99a1b18","entry":"find_module","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/lora.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/lora.py","link_basis":"plan_row","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"b5adfd6ac99a1b18"}},{"code_sha256_prefix":"6a8d363117a347bd","entry":"forward_wrap_with_option_len","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/utils.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"6a8d363117a347bd"}},{"code_sha256_prefix":"e7e75981cb464788","entry":"normalize_answer","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/metrics.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/metrics.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":true,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"e7e75981cb464788"}},{"code_sha256_prefix":"1840e2c8983b56a3","entry":"reshape_matrix","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/trainer.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/trainer.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"1840e2c8983b56a3"}},{"code_sha256_prefix":"b99eea6376d1e212","entry":"rotate_half","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/modeling_llama.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/modeling_llama.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"b99eea6376d1e212"}},{"code_sha256_prefix":"bdb460cb247501dc","entry":"attn_forward_hook","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/prefix_tuning.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/prefix_tuning.py","link_basis":"plan_row","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"bdb460cb247501dc"}},{"code_sha256_prefix":"84c658f2c9252878","entry":"calculate_metric","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/metrics.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/metrics.py","link_basis":"plan_row","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"84c658f2c9252878"}},{"code_sha256_prefix":"b989d4bce26f77ce","entry":"f1","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/metrics.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/metrics.py","link_basis":"plan_row","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"b989d4bce26f77ce"}},{"code_sha256_prefix":"bd2c31732da89465","entry":"fast_svd_method_v2","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/trainer.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/trainer.py","link_basis":"plan_row","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"bd2c31732da89465"}},{"code_sha256_prefix":"f7c384f1e681ea7a","entry":"get_task","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/tasks.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/tasks.py","link_basis":"plan_row","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"f7c384f1e681ea7a"}},{"code_sha256_prefix":"84920a8095dca4fa","entry":"prepare_inputs_for_generation","repo":"zimingyy/subzero","repo_kind":"official","path":"large_models/prefix_tuning.py","file_url":"https://github.com/zimingyy/subzero/blob/HEAD/large_models/prefix_tuning.py","link_basis":"plan_row","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"GPL-3.0","inline_ok":false,"mcp_get_code":{"code_sha256":"84920a8095dca4fa"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}