← Back to Quantization
GPTQ-lite
QuantizationUsed in
106 PRs
Best BPB
0.0280
Avg BPB
1.0834
Submissions
PR #64by yesbhautik
1.1250PR #175by anthony-maio
1.1229PR #344by aryanbhosale
1.1330PR #379by dannywillowliu-uchi
1.1257PR #414by signalrush
1.1233PR #429by AbhisekBasu1
1.1231PR #445by newjordan
1.1236PR #456by Christopher-Lee-McClendon
1.1532PR #473by abaybektursun
1.1214PR #478by gowtham0992
1.1268PR #518by sofiabod
1.0622PR #531by pragnyanramtha
1.1324PR #534by rarce
1.1804PR #549by abaybektursunRECORD
1.1194PR #584by ssatia
1.1233PR #617by ryanadamsai
1.1228PR #625by Joeavaib
1.1194PR #642by minh-stakc
0.8173PR #645by FlynnCruse
1.8990PR #657by anthony-maio
1.1234PR #659by deanbrr
1.0920PR #668by Christopher-Lee-McClendon
1.0920PR #682by gthgomez
1.1233PR #691by xexyz
1.0988PR #710by Dhruba531
1.1240PR #714by Upsalla
1.1187PR #720by agalimova
1.1078PR #726by DeepReinforce
1.1147PR #733by stukenov
1.0278PR #745by stukenov
1.0222PR #754by aryanbhosale
1.1253PR #758by hypery11
1.0465PR #762by robinojw
0.7139PR #763by hypery11
0.9917PR #768by mradassaad
1.1201PR #770by minh-stakc
0.6672PR #771by sunnypatneedi
1.0705PR #784by iverbovoy
1.2065PR #786by shinegami-2002
0.8128PR #794by jeremyschied
1.3346PR #795by hypery11
0.8881PR #797by armantsaturian
0.8960PR #805by zeytx
1.1807PR #816by jimliu741523
1.1194PR #827by Programmerryoki
1.3999PR #838by aryanbhosale
1.1215PR #857by aruniyer
1.1093PR #865by aryanbhosale
0.2841PR #870by simon-marcus
0.0935PR #882by IshiPareek
1.3762PR #884by BhatiaUday
1.1448PR #887by anthony-maio
0.9642PR #889by anthony-maio
0.9642PR #892by robbiebusinessacc
1.1428PR #893by aryanbhosale
0.1310PR #914by mkenney2
1.1873PR #915by anthony-maio
0.9642PR #924by THUQiXuan
0.0280PR #926by NandhuRajRK
0.8705PR #927by Tonyy1977
1.1696PR #932by anthony-maio
1.1580PR #953by dexhunter
1.0722PR #964by vivekvar-dl
1.3900PR #967by dexhunter
1.0450PR #974by anthony-maio
1.6542PR #979by 0xadvait
1.1387PR #995by dexhunter
1.0362PR #1005by OnlyJundong
1.0853PR #1026by danielxmed
1.0945PR #1033by Naazimsnh02
0.4311PR #1043by okezue
1.1261PR #1048by mrdavtan
1.1724PR #1051by tejas-goyal
1.2826PR #1057by Programmerryoki
1.2201PR #1062by yaowubarbara
1.4508PR #1069by manfromnowhere143
1.1190PR #1069by manfromnowhere143
1.1190PR #1070by manfromnowhere143
1.1190PR #1077by malc3om
1.1130PR #1084by AnubhavBharadwaaj
1.1185PR #1086by Omrigotlieb
1.1349PR #1087by Dhenenjay
1.1407PR #1094by michaelwinczuk
0.4027PR #1128by AnubhavBharadwaaj
1.1154PR #1150by sahiee-dev
1.1151PR #1202by VirajDeshwal
1.1412PR #1230by nestamidavaine
1.1163PR #1231by nestamidavaine
1.1163PR #1247by fahmitech
1.2208PR #1269by Jtss-ux
1.1194PR #1276by BiggerDABOSS
1.1100PR #1280by aamodbhatt
1.1156PR #1298by Omrigotlieb
1.1043PR #1311by htrung1105
1.1303PR #1389by Rome-1
1.7270PR #1406by aamodbhatt
1.0887PR #1407by OnlyJundong
1.0960PR #1424by OnlyJundong
1.0858PR #1444by hypnoastic
1.3081PR #1573by shivangbaveja
1.1464PR #1574by KRGulaj
1.3587PR #1579by Tonyy1977
1.1372PR #1582by He-Wenhao
1.3428PR #1630by KevinChunye
1.1412PR #1709by Bananakin1
1.1470PR #1748by elad-simbalista
1.2098Hyperparameters Across PRs
| pr_number | bits | scope |
|---|---|---|
| 64 | 6 | mlp, attn, tok_emb |
| 175 | 6 | all |
| 344 | — | per-row weights |
| 379 | 6 | all weights |
| 414 | 6 | MLP and attention weights |
| 429 | — | all |
| 445 | 6 | all |
| 456 | — | 75% of layers |
| 473 | 6 | model weights |
| 478 | 6 | all large weights |
| 518 | 6 | all |
| 531 | — | attention layers int6, MLP int5, rest int8 or pass-through |
| 534 | — | all |
| 549 | 6 | all |
| 584 | — | model weights |
| 617 | — | all |
| 625 | 6 | — |
| 642 | 6 | all |
| 645 | — | — |
| 657 | 6 | all |
| 659 | 6 | all |
| 668 | 6 | all weights including embeddings |
| 682 | 6 | large 2-D tensors / model weights |
| 691 | 6 | all |
| 710 | 6 | MLP and attention weights |
| 714 | 6 | all |
| 720 | 6 | model weights |
| 726 | 6 | model weights |
| 733 | 6 | all |
| 745 | 6 | all |
| 754 | 6 | per-row weights |
| 758 | 6 | all |
| 762 | 6 | all |
| 763 | 6 | all |
| 768 | 6 | model weights |
| 770 | — | all |
| 771 | 6 | all |
| 784 | 8 | all |
| 786 | 6 | model |
| 794 | 6 | model weights |
| 795 | 6 | all |
| 797 | 6 | all |
| 805 | 6 | per-row weights |
| 816 | 6 | all |
| 827 | 6 | all weights |
| 838 | 6 | all weights with FP16 embedding passthrough |
| 857 | — | all |
| 865 | 6 | model |
| 870 | 6 | all |
| 882 | — | all |
| 884 | 6 | model weights |
| 887 | 6 | all |
| 889 | 6 | model |
| 892 | — | block weights |
| 893 | 6 | model weights |
| 914 | 6 | all |
| 915 | 6 | model weights |
| 924 | 6 | base model |
| 926 | 6 | all |
| 927 | 8 | final artifact |
| 932 | 6 | model |
| 953 | 5 | all |
| 964 | 6 | all |
| 967 | 5 | base model |
| 974 | 6 | block weights |
| 979 | 6 | attn/MLP weights |
| 995 | 5 | model |
| 1005 | 6 | all |
| 1026 | 6 | model |
| 1033 | 6 | all |
| 1043 | 6 | all |
| 1048 | 6 | well-conditioned weights |
| 1051 | 6 | model weights |
| 1057 | 6 | all |
| 1062 | 6 | model weights |
| 1069 | 6 | weights |
| 1069 | 8 | weights |
| 1070 | 6 | MLP+attn |
| 1077 | 6 | per-row |
| 1084 | 6 | model |
| 1086 | — | per-row |
| 1087 | — | all |
| 1094 | 6 | all |
| 1128 | 6 | all |
| 1150 | 6 | all |
| 1202 | 6 | weights |
| 1230 | 6 | export |
| 1231 | 6 | all |
| 1247 | 6 | MLP + attention weights |
| 1269 | 6 | all |
| 1276 | 6 | all |
| 1280 | 6 | all |
| 1298 | 6 | all |
| 1311 | 6 | all large weight matrices |
| 1389 | 6 | all int6 tensors |
| 1406 | 6 | all |
| 1407 | 6 | model + code |
| 1424 | 6 | model weights |
| 1444 | 6 | all |
| 1573 | 5 | MLP and attention weights |
| 1574 | 6 | all weights; embeddings int8 |
| 1579 | 6 | all |
| 1582 | 8 | per-row weights |
| 1630 | 6 | all |
| 1709 | 6 | all |
| 1748 | 8 | per-row |