{"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/deep-compression-autoencoder-for-efficient","title":"Deep Compression Autoencoder for Efficient High-Resolution Diffusion Models","arxiv_id":"2410.10733","date":"2024-10-14","proceeding":null,"authors":["Junyu Chen","Han Cai","Junsong Chen","Enze Xie","Shang Yang","Haotian Tang","Muyang Li","Yao Lu","Song Han"],"abstract":"We present Deep Compression Autoencoder (DC-AE), a new family of autoencoder models for accelerating high-resolution diffusion models. Existing autoencoder models have demonstrated impressive results at a moderate spatial compression ratio (e.g., 8x), but fail to maintain satisfactory reconstruction accuracy for high spatial compression ratios (e.g., 64x). We address this challenge by introducing two key techniques: (1) Residual Autoencoding, where we design our models to learn residuals based on the space-to-channel transformed features to alleviate the optimization difficulty of high spatial-compression autoencoders; (2) Decoupled High-Resolution Adaptation, an efficient decoupled three-phases training strategy for mitigating the generalization penalty of high spatial-compression autoencoders. With these designs, we improve the autoencoder's spatial compression ratio up to 128 while maintaining the reconstruction quality. Applying our DC-AE to latent diffusion models, we achieve significant speedup without accuracy drop. For example, on ImageNet 512x512, our DC-AE provides 19.1x inference speedup and 17.9x training speedup on H100 GPU for UViT-H while achieving a better FID, compared with the widely used SD-VAE-f8 autoencoder. Our code is available at https://github.com/mit-han-lab/efficientvit.","url_abs":"https://arxiv.org/abs/2410.10733v8","url_pdf":"https://arxiv.org/pdf/2410.10733v8.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":"deep-compression-autoencoder-for-efficient","repo_url":"https://github.com/mit-han-lab/efficientvit","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":null,"task_name":"GPU"},{"task_slug":"image-generation","task_name":"Image Generation"}],"methods":[{"method_slug":"diffusion","method_name":"Diffusion"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-generation-on-imagenet-512x512","task":"Image Generation","dataset":"ImageNet 512x512","model":"DC-AE-f32 + USiT-2B","rank_in_archive_order":16,"of":52,"metrics":{"FID":"1.72"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2410.10733","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2410.10733"}},"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/mit-han-lab/efficientvit","reach":null}],"summary":{"ran":7,"unverified":23},"by_repo_kind":{"official":{"samples":30,"ran":7,"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":"5a6b494abcac2208","entry":"ChannelDuplicatingPixelUnshuffleUpSampleLayer","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"deterministic","behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"5a6b494abcac2208"}},{"code_sha256_prefix":"a7400f7dc57e2876","entry":"DCAEConfig","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"a7400f7dc57e2876"}},{"code_sha256_prefix":"64d381f16614334c","entry":"DecoderConfig","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"64d381f16614334c"}},{"code_sha256_prefix":"7c0f5cb15a33304c","entry":"EncoderConfig","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"7c0f5cb15a33304c"}},{"code_sha256_prefix":"7d8080781cf3019c","entry":"IdentityLayer","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"7d8080781cf3019c"}},{"code_sha256_prefix":"a8bff7f7a6484f90","entry":"OpSequential","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"deterministic","behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"a8bff7f7a6484f90"}},{"code_sha256_prefix":"447757507c43986b","entry":"PixelUnshuffleChannelAveragingDownSampleLayer","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"447757507c43986b"}},{"code_sha256_prefix":"ee46943abe5b1d38","entry":"ConvLayer","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ee46943abe5b1d38"}},{"code_sha256_prefix":"a50b4796b9e7f993","entry":"ConvPixelShuffleUpSampleLayer","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"a50b4796b9e7f993"}},{"code_sha256_prefix":"c70aaa792533f82b","entry":"ConvPixelUnshuffleDownSampleLayer","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"c70aaa792533f82b"}},{"code_sha256_prefix":"7b99877bfad9d9db","entry":"DCAE","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"7b99877bfad9d9db"}},{"code_sha256_prefix":"516380f38de4bf18","entry":"Decoder","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"516380f38de4bf18"}},{"code_sha256_prefix":"644cd11053cbe8fa","entry":"EfficientViTBlock","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"644cd11053cbe8fa"}},{"code_sha256_prefix":"e8f172d1d06ebffc","entry":"Encoder","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"e8f172d1d06ebffc"}},{"code_sha256_prefix":"50c192aef38f49ce","entry":"GLUMBConv","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"50c192aef38f49ce"}},{"code_sha256_prefix":"e3551b0a1b8b4a44","entry":"InterpolateConvUpSampleLayer","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"e3551b0a1b8b4a44"}},{"code_sha256_prefix":"eca2ca74ae04da15","entry":"LiteMLA","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"eca2ca74ae04da15"}},{"code_sha256_prefix":"63542ac85cdf8517","entry":"MBConv","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"63542ac85cdf8517"}},{"code_sha256_prefix":"b8654d21821ceafc","entry":"ResBlock","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"b8654d21821ceafc"}},{"code_sha256_prefix":"9d531d6342d2cc9c","entry":"ResidualBlock","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"9d531d6342d2cc9c"}},{"code_sha256_prefix":"63bb3de89867ea9d","entry":"build_act","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"63bb3de89867ea9d"}},{"code_sha256_prefix":"a0bb0bb02b4fd978","entry":"build_block","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"a0bb0bb02b4fd978"}},{"code_sha256_prefix":"29728624d884920a","entry":"build_decoder_project_in_block","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"29728624d884920a"}},{"code_sha256_prefix":"498fce97a28da008","entry":"build_decoder_project_out_block","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"498fce97a28da008"}},{"code_sha256_prefix":"0d1e3498712a547c","entry":"build_downsample_block","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"0d1e3498712a547c"}},{"code_sha256_prefix":"1598ee0cf2028172","entry":"build_encoder_project_in_block","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"1598ee0cf2028172"}},{"code_sha256_prefix":"b8d5cdc9329421af","entry":"build_encoder_project_out_block","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"b8d5cdc9329421af"}},{"code_sha256_prefix":"71068d4faddfa7bb","entry":"build_norm","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"71068d4faddfa7bb"}},{"code_sha256_prefix":"c77ad3ad88af1f35","entry":"build_stage_main","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"c77ad3ad88af1f35"}},{"code_sha256_prefix":"885f68d08c4d49f1","entry":"build_upsample_block","repo":"mit-han-lab/efficientvit","repo_kind":"official","path":"efficientvit/models/efficientvit/dc_ae.py","file_url":"https://github.com/mit-han-lab/efficientvit/blob/HEAD/efficientvit/models/efficientvit/dc_ae.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"885f68d08c4d49f1"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}