{"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":"/census/trunc-normal","entry":"trunc_normal_","source":"Syntology differential census (groundwork/55, run v2_2026-09-22), per sample; not an archive number","census_date":"2026-09-22","battery_sha256":["b986f7e04d794a0d88ad4c5f32cf63ec3590b5deff0192150737bc5f1c0b4677"],"runner_sha256":["5a452d0e7c0da5b80771d1be2afe3572e253568cf5c5d59a0d08ebd663d00808"],"bucket_key":"positional (rank, kind, dtype) of each array argument; the argument name is not part of the key because the harness draws the shared array from (rank, kind) and casts it to the dtype, whatever the name","bucket_fields":["rank","kind","dtype"],"claim":"Implementations sharing this entry name were each run on one shared input fixed by the positional (rank, kind, dtype) of their array arguments (the bucket). A cluster is the set whose recorded output digest (sha256 of the output rounded to 6 decimals) is identical. Identical values to six decimals on the shared input are agreement on those inputs, not a statement about the whole domain and not a substitution claim.","n_implementations_compared":37,"n_papers":54,"n_buckets":1,"n_distinct_outputs":4,"n_class_bearing":0,"not_compared":{"not_run_on_shared_input":{"n":0,"by_error":{}},"no_array_argument_ran_on_own_fixture_arguments_only":{"n":0,"n_papers":0,"recorded_shared_digest_equals_own_fixture_digest":0},"output_not_digested_non_numeric":{"n":0,"by_type":{}}},"withdrawn_excluded":0,"code_page":"/code/trunc-normal","buckets":[{"bucket":[[2,"float","float32"]],"n_implementations_compared":37,"n_papers":54,"n_not_digested":0,"not_digested_by_type":{},"clusters":[{"output_sha":"42f0dd1794688eb5","size":34,"n_papers":50,"shape":[4,8],"dtype":"float32","type":"Tensor","finite":true,"max_difference_among_recorded_values":0.0,"members_with_recorded_values":34,"torch_reference_conventions_with_this_digest":[],"values":[-0.008956512436270714,0.6935495734214783,-1.2415472269058228,-1.041709542274475,-0.4783933460712433,0.3265029191970825,-0.023704487830400467,1.1670573949813843,-0.10636364668607712,0.3220333456993103,-0.3697916865348816,-0.23735886812210083,-1.7053985595703125,-0.9005043506622314,-0.5150284171104431,0.044329289346933365,0.49209433794021606,0.7938574552536011,-0.9289683699607849,-0.547040581703186,0.44916948676109314,1.260761022567749,-0.24873757362365723,1.067521095275879,-0.19403237104415894,0.12692409753799438,1.491908073425293,-1.5781161785125732,-0.8432185649871826,-0.30764320492744446,-0.4846338629722595,1.3553329706192017],"values_recorded":32,"members":[{"code_sha256_prefix":"c76d18678f19d332","path":"real/test_code/architecture/MST.py","papers":["2111.07910","2204.01696","2204.07908","2311.10959","2411.15648"],"paper_pages":[{"arxiv_id":"2111.07910","page":"/paper/mask-guided-spectral-wise-transformer-for"},{"arxiv_id":"2204.01696","page":"/paper/joint-hand-motion-and-interaction-hotspots"},{"arxiv_id":"2204.07908","page":"/paper/mst-multi-stage-spectral-wise-transformer-for"},{"arxiv_id":"2311.10959","page":"/paper/structure-aware-sparse-view-x-ray-3d"},{"arxiv_id":"2411.15648","page":"/paper/sample-and-parameter-efficient-auto"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"915b71e73a54fc97","path":"architectures/backbone/DINO_ViT.py","papers":["2301.12246","2410.13798","2506.03110","2602.01951"],"paper_pages":[{"arxiv_id":"2301.12246","page":"/paper/a-closer-look-at-few-shot-classification"},{"arxiv_id":"2410.13798","page":"/paper/learning-graph-quantized-tokenizers-for"},{"arxiv_id":"2506.03110","page":null},{"arxiv_id":"2602.01951","page":"/paper/arxiv-2602-01951"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"02566da69866c48c","path":"model/SSRT.py","papers":["2204.07683","2605.30107","2601.12079"],"paper_pages":[{"arxiv_id":"2204.07683","page":"/paper/safe-self-refinement-for-transformer-based"},{"arxiv_id":"2605.30107","page":"/paper/arxiv-2605-30107"},{"arxiv_id":"2601.12079","page":"/paper/arxiv-2601-12079"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"056f371660385ea5","path":"corun_colabator/archs/corun_arch.py","papers":["2406.07966","2411.11409"],"paper_pages":[{"arxiv_id":"2406.07966","page":"/paper/real-world-image-dehazing-with-coherence"},{"arxiv_id":"2411.11409","page":"/paper/ikea-manuals-at-work-4d-grounding-of-assembly"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"359975e35f7d1d79","path":"models/FM/EEGPT/Model_EEGPT.py","papers":["2601.17883","2609.15687"],"paper_pages":[{"arxiv_id":"2601.17883","page":"/paper/arxiv-2601-17883"},{"arxiv_id":"2609.15687","page":"/paper/arxiv-2609-15687"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"605878f6dcef72e9","path":"fastreid/modeling/backbones/vision_transformer.py","papers":["2105.14432","2402.00282"],"paper_pages":[{"arxiv_id":"2105.14432","page":"/paper/transformer-based-deep-image-matching-for"},{"arxiv_id":"2402.00282","page":"/paper/pam-prompting-audio-language-models-for-audio"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"c3487124f762d881","path":"easy_ViTPose/vit_models/model.py","papers":["2204.12484","2411.16289"],"paper_pages":[{"arxiv_id":"2204.12484","page":"/paper/vitpose-simple-vision-transformer-baselines"},{"arxiv_id":"2411.16289","page":"/paper/utilizing-uncertainty-in-2d-pose-detectors"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"d8578a1e5c318cf9","path":"src/model/model_action.py","papers":["2309.17342","2308.09951"],"paper_pages":[{"arxiv_id":"2309.17342","page":"/paper/towards-free-data-selection-with-general-1"},{"arxiv_id":"2308.09951","page":"/paper/semantics-meets-temporal-correspondence-self"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"e6ea9664b9a77523","path":"model/impl/actionformer.py","papers":["2504.08222","2403.11882"],"paper_pages":[{"arxiv_id":"2504.08222","page":"/paper/f-3-set-towards-analyzing-fast-frequent-and"},{"arxiv_id":"2403.11882","page":"/paper/regennet-towards-human-action-reaction"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"fabe0899316ffaa5","path":"models/cha_mae_vit.py","papers":["2503.19331","2601.21159"],"paper_pages":[{"arxiv_id":"2503.19331","page":"/paper/cha-maevit-unifying-channel-aware-masked"},{"arxiv_id":"2601.21159","page":"/paper/arxiv-2601-21159"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"04c04dfd96b26f9a","path":"models/early_exit.py","papers":["2410.09633"],"paper_pages":[{"arxiv_id":"2410.09633","page":"/paper/duodiff-accelerating-diffusion-models-with-a"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"0af416f19e3851c1","path":"basicsr/archs/swinir_arch.py","papers":["2108.10257"],"paper_pages":[{"arxiv_id":"2108.10257","page":"/paper/swinir-image-restoration-using-swin"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"28455f5511c6676f","path":"models/jepa.py","papers":["2410.06940"],"paper_pages":[{"arxiv_id":"2410.06940","page":"/paper/representation-alignment-for-generation"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"462cf9fa4898bc51","path":"easycv/models/backbones/vision_transformer.py","papers":["2012.12877"],"paper_pages":[{"arxiv_id":"2012.12877","page":"/paper/training-data-efficient-image-transformers"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"4e02bf1d1efcb33d","path":"src/models/predictor.py","papers":["2602.02603"],"paper_pages":[{"arxiv_id":"2602.02603","page":"/paper/arxiv-2602-02603"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"505222cf2cc4240f","path":"real/test_code/architecture/DAUHST.py","papers":["2205.10102"],"paper_pages":[{"arxiv_id":"2205.10102","page":"/paper/degradation-aware-unfolding-half-shuffle"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"6586759c888019fc","path":"split_model.py","papers":["2311.07788"],"paper_pages":[{"arxiv_id":"2311.07788","page":"/paper/cslp-ae-a-contrastive-split-latent-1"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"8347c57b16dab539","path":"models/network_vrt.py","papers":["2201.12288"],"paper_pages":[{"arxiv_id":"2201.12288","page":"/paper/vrt-a-video-restoration-transformer"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"8c3374496d26feda","path":"modeling/fusion_part/CRM.py","papers":["aaai_28388"],"paper_pages":[{"arxiv_id":"aaai_28388","page":null}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"95f8054d3a8863e8","path":"foundation/model.py","papers":["2407.08647"],"paper_pages":[{"arxiv_id":"2407.08647","page":"/paper/from-real-to-cloned-singer-identification"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"ab8464b48ee9d702","path":"vissl/models/trunks/beit_transformer.py","papers":["2106.08254"],"paper_pages":[{"arxiv_id":"2106.08254","page":"/paper/beit-bert-pre-training-of-image-transformers"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"ada91a01b9efbb1d","path":"towhee/models/swin_transformer/model.py","papers":["2103.14030"],"paper_pages":[{"arxiv_id":"2103.14030","page":"/paper/swin-transformer-hierarchical-vision"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"af8aefa30bd76920","path":"src/models/reefl_vit.py","papers":["2405.14791"],"paper_pages":[{"arxiv_id":"2405.14791","page":"/paper/recurrent-early-exits-for-federated-learning"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"b425e24882305880","path":"models/sam_withToken.py","papers":["2606.27655"],"paper_pages":[{"arxiv_id":"2606.27655","page":"/paper/arxiv-2606-27655"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"ba6277edefa610aa","path":"model/backbones/vit_pytorch.py","papers":["2308.03322"],"paper_pages":[{"arxiv_id":"2308.03322","page":"/paper/part-aware-transformer-for-generalizable"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"be2feea89481c009","path":"models/dichavit.py","papers":["2405.16419"],"paper_pages":[{"arxiv_id":"2405.16419","page":"/paper/enhancing-feature-diversity-boosts-channel"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"c92dcfe50ad922d8","path":"timm/models/bixt.py","papers":["2402.12138"],"paper_pages":[{"arxiv_id":"2402.12138","page":"/paper/perceiving-longer-sequences-with-bi"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"ca3425736d00ed34","path":"models/vision_transformer_moe.py","papers":["2210.14793"],"paper_pages":[{"arxiv_id":"2210.14793","page":"/paper/m-3-vit-mixture-of-experts-vision-transformer"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"cd079d4c82711147","path":"dic_models.py","papers":["2501.00603"],"paper_pages":[{"arxiv_id":"2501.00603","page":"/paper/dic-rethinking-conv3x3-designs-in-diffusion"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"d347dc5f198621fd","path":"pypots/nn/modules/tslanet/backbone.py","papers":["2404.08472"],"paper_pages":[{"arxiv_id":"2404.08472","page":"/paper/tslanet-rethinking-transformers-for-time"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"ef92035251726e40","path":"joint_evol_opt.py","papers":["2308.10814"],"paper_pages":[{"arxiv_id":"2308.10814","page":"/paper/jumping-through-local-minima-quantization-in"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"f3e47c6e3250d13b","path":"towhee/models/uniformer/uniformer.py","papers":["2201.04676"],"paper_pages":[{"arxiv_id":"2201.04676","page":"/paper/uniformer-unified-transformer-for-efficient-1"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"f836ffb10ee7ef21","path":"codes/models/decoder_affordance.py","papers":["2505.24103"],"paper_pages":[{"arxiv_id":"2505.24103","page":"/paper/weakly-supervised-affordance-grounding-guided"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false},{"code_sha256_prefix":"fd05acd780fc4073","path":"models.py","papers":["2506.00129"],"paper_pages":[{"arxiv_id":"2506.00129","page":"/paper/geo-sign-hyperbolic-contrastive-1"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false}]},{"output_sha":"8983255eae79bfda","size":1,"n_papers":2,"shape":[4,8],"dtype":"float32","type":"Tensor","finite":true,"max_difference_among_recorded_values":null,"members_with_recorded_values":1,"torch_reference_conventions_with_this_digest":[],"values":[-1.1258398294448853,-1.152360200881958,-0.2505785822868347,-0.4338788390159607,0.8487103581428528,0.6920092105865479,-0.31601276993751526,-0.1152195930480957,0.32227492332458496,-1.2633347511291504,0.34998318552970886,0.30813390016555786,0.11984150856733322,1.237657904624939,1.1167771816253662,-0.2472776472568512,-1.3526537418365479,-1.695931315422058,0.5666505098342896,0.7935084104537964,0.5988394618034363,-1.5550950765609741,-0.34136030077934265,1.85300612449646,0.7501894235610962,-0.5854971408843994,-0.1733970195055008,0.18347792327404022,1.3893661499023438,1.5863343477249146,0.946298360824585,-0.8436768054962158],"values_recorded":32,"members":[{"code_sha256_prefix":"5cb5985f521b3573","path":"modules/fusion_module.py","papers":["2203.14104","2312.08010"],"paper_pages":[{"arxiv_id":"2203.14104","page":"/paper/bridge-prompt-towards-ordinal-action"},{"arxiv_id":"2312.08010","page":"/paper/ez-clip-efficient-zeroshot-video-action"}],"arg_sig_recorded":[["x",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0"},"class_bearing":false}]},{"output_sha":"10d0186bfd2086a8","size":1,"n_papers":1,"shape":[4,8],"dtype":"float32","type":"Tensor","finite":true,"max_difference_among_recorded_values":null,"members_with_recorded_values":1,"torch_reference_conventions_with_this_digest":[],"values":[-0.008956528268754482,0.6935495734214783,-1.2415472269058228,-1.041709542274475,-0.47839340567588806,0.3265029788017273,-0.023704543709754944,1.1670573949813843,-0.10636366903781891,0.3220333158969879,-0.3697916865348816,-0.2373589277267456,-1.7053985595703125,-0.900504469871521,-0.5150284171104431,0.044329285621643066,0.49209433794021606,0.7938574552536011,-0.9289685487747192,-0.5470406413078308,0.4491695463657379,1.260761022567749,-0.24873758852481842,1.0675209760665894,-0.19403238594532013,0.1269240379333496,1.491908073425293,-1.5781161785125732,-0.8432185649871826,-0.30764323472976685,-0.4846338629722595,1.3553330898284912],"values_recorded":32,"members":[{"code_sha256_prefix":"92221940e53dde88","path":"pbb/models.py","papers":["2007.12911"],"paper_pages":[{"arxiv_id":"2007.12911","page":"/paper/tighter-risk-certificates-for-neural-networks"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0.0","std":"1.0","a":"-2.0","b":"2.0"},"class_bearing":false}]},{"output_sha":"8be0b5c837c2bd07","size":1,"n_papers":1,"shape":[4,8],"dtype":"float32","type":"Tensor","finite":true,"max_difference_among_recorded_values":null,"members_with_recorded_values":1,"torch_reference_conventions_with_this_digest":[],"values":[-0.011258398182690144,0.008487103506922722,0.003222749102860689,0.0011984150623902678,-0.013526537455618382,0.005988394375890493,0.0075018941424787045,0.013893661089241505,-0.006135830655694008,0.004396958742290735,-0.0010230964981019497,0.005228604190051556,-0.006730899214744568,-0.002303354674950242,-0.00446217181161046,-0.015311842784285545,-0.005692480131983757,-0.014781743288040161,-0.01629325933754444,-0.010669799521565437,-0.0009334822534583509,0.00841894093900919,-0.002460009418427944,-0.010449785739183426,0.016458669677376747,-0.0169647466391325,0.002469263970851898,0.004569407552480698,-0.0021883368026465178,0.00962518248707056,-0.0062269847840070724,0.0012390248011797667],"values_recorded":32,"members":[{"code_sha256_prefix":"49e6743e36a2c42f","path":"models/modeling/PSVMAModel/PSVMANet.py","papers":["2303.15322"],"paper_pages":[{"arxiv_id":"2303.15322","page":"/paper/progressive-semantic-visual-mutual-adaption"}],"arg_sig_recorded":[["tensor",2,"float","float32"]],"scalar_args":{"mean":"0","std":"0.01"},"class_bearing":false}]}]}]}