{"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/moe-infinity-activation-aware-expert","title":"MoE-Infinity: Efficient MoE Inference on Personal Machines with Sparsity-Aware Expert Cache","arxiv_id":"2401.14361","date":"2024-01-25","proceeding":null,"authors":["Leyang Xue","Yao Fu","Zhan Lu","Luo Mai","Mahesh Marina"],"abstract":"This paper presents MoE-Infinity, an efficient MoE inference system designed for personal machines with limited GPU memory capacity. The key idea for MoE-Infinity is that on personal machines, which are often single-user environments, MoE-based LLMs typically operate with a batch size of one. In this setting, MoE models exhibit a high degree of activation sparsity, meaning a small number of experts are frequently reused in generating tokens during the decode phase. Leveraging this idea, we design a sparsity-aware expert cache, which can trace the sparse activation of experts during inference and carefully select the trace that represents the sparsity pattern. By analyzing these selected traces, MoE-Infinity guides the replacement and prefetching of the expert cache, providing 3.1-16.7x per-token latency improvements over numerous state-of-the-art systems, including vLLM, Ollama, DeepSpeed and BrainStorm across various MoE models (DeepSeek and Mixtral) when handling different LLM tasks. MoE-Infinity's source code is publicly available at https://github.com/EfficientMoE/MoE-Infinity","url_abs":"https://arxiv.org/abs/2401.14361v3","url_pdf":"https://arxiv.org/pdf/2401.14361v3.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":"moe-infinity-activation-aware-expert","repo_url":"https://github.com/efficientmoe/moe-infinity","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"moe-infinity-activation-aware-expert","repo_url":"https://github.com/torchmoe/moe-infinity","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":null,"task_name":"GPU"},{"task_slug":"model","task_name":"model"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2401.14361","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2401.14361"}},"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/efficientmoe/moe-infinity","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/torchmoe/moe-infinity","reach":{"status":"ok","spdx":"Apache-2.0"}}],"summary":{"ran_draft_wrong":2,"ran":1},"by_repo_kind":{"official":{"samples":3,"ran":3,"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":0,"samples":[{"code_sha256_prefix":"3a5205d2f63607bc","entry":"capture_result","repo":"efficientmoe/moe-infinity","repo_kind":"official","path":"benchmarks/serving/validate_batched_dispatch.py","file_url":"https://github.com/efficientmoe/moe-infinity/blob/HEAD/benchmarks/serving/validate_batched_dispatch.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":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"3a5205d2f63607bc"}},{"code_sha256_prefix":"d546dd12f4a824c5","entry":"extract_context_feature","repo":"torchmoe/moe-infinity","repo_kind":"official","path":"moe_infinity/entrypoints/big_modeling.py","file_url":"https://github.com/torchmoe/moe-infinity/blob/HEAD/moe_infinity/entrypoints/big_modeling.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"d546dd12f4a824c5"}},{"code_sha256_prefix":"dbefd08c6aab8151","entry":"recommendation","repo":"efficientmoe/moe-infinity","repo_kind":"official","path":"benchmarks/serving/validate_batched_dispatch.py","file_url":"https://github.com/efficientmoe/moe-infinity/blob/HEAD/benchmarks/serving/validate_batched_dispatch.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":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"dbefd08c6aab8151"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}