← Back to Weight Averaging
EMA + Tight SWA
Weight AveragingUsed in
28 PRs
Best BPB
0.0280
Avg BPB
0.9975
Submissions
PR #625by Joeavaib
1.1194PR #634by raahilshah
1.1171PR #653by demirelo
1.1552PR #657by anthony-maio
1.1234PR #794by jeremyschied
1.3346PR #872by gowtham0992
1.0467PR #884by BhatiaUday
1.1448PR #885by lolrazh
0.9958PR #889by anthony-maio
0.9642PR #890by sofiabod
0.4405PR #924by THUQiXuan
0.0280PR #925by THUQiXuan
0.0281PR #952by FlashyFlash3011
1.1144PR #1026by danielxmed
1.0945PR #1084by AnubhavBharadwaaj
1.1185PR #1128by AnubhavBharadwaaj
1.1154PR #1202by VirajDeshwal
1.1412PR #1263by xexyz
0.9354PR #1269by Jtss-ux
1.1194PR #1276by BiggerDABOSS
1.1100PR #1280by aamodbhatt
1.1156PR #1284by tyrel-beede
1.1207PR #1303by anthony-maio
0.9462PR #1392by Its-Just-Crump
1.1020PR #1407by OnlyJundong
1.0960PR #1424by OnlyJundong
1.0858PR #1446by LauraGomezjurado
1.0960PR #1696by kings-crown
1.1224Hyperparameters Across PRs
| pr_number | parameters |
|---|---|
| 625 | {"EMA_decay":0.997,"SWA_every":50} |
| 634 | {"EMA_decay":0.997,"SWA_frequency_steps":50,"SWA_scale_threshold":0.2} |
| 653 | {"ema_decay":0.997,"swa_every":50} |
| 657 | {"ema_decay":0.997,"swa_scale_max":0.2} |
| 794 | {"ema_decay":0.997,"swa_every":50} |
| 872 | {"ema_decay":0.997,"swa_interval":50} |
| 884 | {"decay":0.997,"swa_interval_steps":50} |
| 885 | {"ema_decay":0.997} |
| 889 | {"decay":0.997} |
| 890 | {"ema_decay":0.997} |
| 924 | {"ema_decay":0.997,"swa_interval":50} |
| 925 | {"ema_decay":0.997,"swa_every":50} |
| 952 | {"ema_decay":0.997,"swa_every":50} |
| 1026 | {"ema_decay":0.997,"swa_every":50} |
| 1084 | {"ema_decay":0.997,"swa_every":50} |
| 1128 | {"decay":0.997,"swa_every":50} |
| 1202 | {"ema_decay":0.997,"swa_every":50} |
| 1263 | {"decay":0.997,"swa_every":50,"swa_start_step":4600} |
| 1269 | {"ema_decay":0.997,"swa_every":50} |
| 1276 | {"ema_decay":0.997,"swa_every":50} |
| 1280 | {"ema_decay":0.997,"swa_every":50} |
| 1284 | {"ema_decay":0.997,"swa_interval":50} |
| 1303 | {"decay":0.997} |
| 1392 | {"ema_decay":0.997,"swa_every":50} |
| 1407 | {"ema_decay":0.997,"swa_every":50} |
| 1424 | {"ema_decay":0.997,"swa_every":50} |
| 1446 | {"decay":0.997,"swa_every":50} |
| 1696 | {"ema_decay":0.997,"swa_interval":50} |