{"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/quantize","entry":"quantize","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":8,"n_papers":11,"n_buckets":3,"n_distinct_outputs":6,"n_class_bearing":0,"not_compared":{"not_run_on_shared_input":{"n":1,"by_error":{"TypeError":1}},"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/quantize","buckets":[{"bucket":[[2,"float","float32"]],"n_implementations_compared":5,"n_papers":5,"n_not_digested":0,"not_digested_by_type":{},"clusters":[{"output_sha":"0ca6e75121f5f59d","size":2,"n_papers":2,"shape":[4,8],"dtype":"float32","type":"Tensor","finite":true,"max_difference_among_recorded_values":0.0,"members_with_recorded_values":2,"torch_reference_conventions_with_this_digest":[],"values":[1.796110987663269,-1.5300524234771729,0.4811161756515503,-0.6946439743041992,-0.8957607746124268,-1.0349955558776855,-0.6637029051780701,0.5739392042160034,0.5120571851730347,0.9607025384902954,-0.40070390701293945,-1.5609934329986572,-0.2924102544784546,2.322108745574951,-0.5089975595474243,0.542998194694519,0.06341183185577393,-1.6228755712509155,0.43470442295074463,-0.40070390701293945,-0.02941131591796875,-1.1742303371429443,-0.5399386882781982,0.9761730432510376,-0.8029376268386841,0.43470442295074463,0.9452320337295532,1.0225845575332642,-1.220641851425171,1.2701131105422974,-1.3289356231689453,-0.6018208265304565],"values_recorded":32,"members":[{"code_sha256_prefix":"7d546c0310d1c591","path":"models.py","papers":["2001.00705"],"paper_pages":[{"arxiv_id":"2001.00705","page":"/paper/fractional-skipping-towards-finer-grained"}],"arg_sig_recorded":[["x",2,"float","float32"]],"scalar_args":{"num_bits":"8","min_value":"None","max_value":"None","num_chunks":"None","stochastic":"False","inplace":"False"},"class_bearing":false},{"code_sha256_prefix":"ce590a0abac42625","path":"utils/quantize.py","papers":["2002.00104"],"paper_pages":[{"arxiv_id":"2002.00104","page":"/paper/near-lossless-post-training-quantization-of"}],"arg_sig_recorded":[["x",2,"float","float32"]],"scalar_args":{"num_bits":"8","min_value":"None","max_value":"None","inplace":"False","symmetric":"False","num_chunks":"None"},"class_bearing":false}]},{"output_sha":"04bac2a1047c23c3","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":[1.7988852262496948,-1.265554428100586,0.06453067064285278,-1.05885648727417,-1.0954266786575317,-0.8963925242424011,-0.6726675033569336,0.9693339467048645,0.47450393438339233,1.09145188331604,-0.5483061075210571,-1.6666308641433716,-0.7662093639373779,1.9909677505493164,-0.720895528793335,0.5690503716468811,0.2640630304813385,-1.3228641748428345,0.0931408703327179,-0.623917281627655,0.15901432931423187,-0.7655884623527527,-0.6397683620452881,1.3578300476074219,-0.8762044906616211,0.48618146777153015,1.3977158069610596,0.5536831021308899,-1.5405547618865967,1.1476353406906128,-1.5171217918395996,-0.16210871934890747],"values_recorded":32,"members":[{"code_sha256_prefix":"771654b864892612","path":"models.py","papers":["2010.10258"],"paper_pages":[{"arxiv_id":"2010.10258","page":"/paper/hierarchical-autoregressive-modeling-for-1"}],"arg_sig_recorded":[["x",2,"float","float32"]],"scalar_args":{"mode":"'noise'","offset":"None","scale":"1"},"class_bearing":false}]},{"output_sha":"bd37723d45f908a5","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":[1.7999999523162842,-1.5333333015441895,0.4745098352432251,-0.6941176652908325,-0.9058823585510254,-1.0235294103622437,-0.6705882549285889,0.5686274766921997,0.5137255191802979,0.9529411792755127,-0.40392154455184937,-1.5647058486938477,-0.29411762952804565,2.3176469802856445,-0.5215686559677124,0.545098066329956,0.058823585510253906,-1.619607925415039,0.4274510145187378,-0.4117646813392639,-0.027450978755950928,-1.1803921461105347,-0.5372549295425415,0.9764705896377563,-0.7960784435272217,0.4274510145187378,0.9372549057006836,1.015686273574829,-1.2196078300476074,1.2666666507720947,-1.3215686082839966,-0.6000000238418579],"values_recorded":32,"members":[{"code_sha256_prefix":"34dad53061b916d4","path":"rivagan/rivagan.py","papers":["1909.01285"],"paper_pages":[{"arxiv_id":"1909.01285","page":"/paper/robust-invisible-video-watermarking-with"}],"arg_sig_recorded":[["frames",2,"float","float32"]],"scalar_args":{},"class_bearing":false}]},{"output_sha":"bf082cc7cc934b0c","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":[1.0,-1.0,1.0,-1.0,-1.0,-1.0,-1.0,1.0,1.0,1.0,-1.0,-1.0,0.0,1.0,-1.0,1.0,0.0,-1.0,1.0,-1.0,0.0,-1.0,-1.0,1.0,-1.0,1.0,1.0,1.0,-1.0,1.0,-1.0,-1.0],"values_recorded":32,"members":[{"code_sha256_prefix":"d68f2c13b2387a62","path":"quantification.py","papers":["1612.01064"],"paper_pages":[{"arxiv_id":"1612.01064","page":"/paper/trained-ternary-quantization"}],"arg_sig_recorded":[["tensor_data",2,"float","float32"]],"scalar_args":{"w_p":"1.0","w_n":"1.0","threshold":"0.15"},"class_bearing":false}]}]},{"bucket":[[2,"float","float32"],[0,"float","float32"],[0,"float","float32"]],"n_implementations_compared":2,"n_papers":4,"n_not_digested":0,"not_digested_by_type":{},"clusters":[{"output_sha":"34e51086a3d2c224","size":2,"n_papers":4,"shape":[4,8],"dtype":"float32","type":"Tensor","finite":true,"max_difference_among_recorded_values":0.0,"members_with_recorded_values":2,"torch_reference_conventions_with_this_digest":[],"values":[-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589,-2.027277708053589],"values_recorded":32,"members":[{"code_sha256_prefix":"41c57e55d21c4d96","path":"quant.py","papers":["2407.10032","2605.04084","2601.15287"],"paper_pages":[{"arxiv_id":"2407.10032","page":"/paper/leanquant-accurate-large-language-model"},{"arxiv_id":"2605.04084","page":"/paper/arxiv-2605-04084"},{"arxiv_id":"2601.15287","page":"/paper/arxiv-2601-15287"}],"arg_sig_recorded":[["x",2,"float","float32"],["scale",0,"float","float32"],["zero",0,"float","float32"]],"scalar_args":{"maxq":"255"},"class_bearing":false},{"code_sha256_prefix":"8f8eedd720c09255","path":"deltazip/core/quant.py","papers":["2312.05215"],"paper_pages":[{"arxiv_id":"2312.05215","page":"/paper/deltazip-multi-tenant-language-model-serving"}],"arg_sig_recorded":[["x",2,"float","float32"],["scale",0,"float","float32"],["zero",0,"float","float32"]],"scalar_args":{"maxq":"255"},"class_bearing":false}]}]},{"bucket":[[2,"float","float32"],[0,"float","float32"],[0,"float","float32"],[0,"float","float32"],[0,"float","float32"]],"n_implementations_compared":1,"n_papers":2,"n_not_digested":0,"not_digested_by_type":{},"clusters":[{"output_sha":"98c3b1dee024d132","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":[-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0,-0.0],"values_recorded":32,"members":[{"code_sha256_prefix":"a166c4b79b21b60a","path":"qeft/quant.py","papers":["2306.00978","2306.02272"],"paper_pages":[{"arxiv_id":"2306.00978","page":"/paper/awq-activation-aware-weight-quantization-for"},{"arxiv_id":"2306.02272","page":"/paper/owq-lessons-learned-from-activation-outliers"}],"arg_sig_recorded":[["x",2,"float","float32"],["scale",0,"float","float32"],["zero",0,"float","float32"],["minq",0,"float","float32"],["maxq",0,"float","float32"]],"scalar_args":{},"class_bearing":false}]}]}]}