| experiment_id timestamp description status val_loss accuracy roc_auc steps duration_s notes |
| exp_baseline_001 2026-03-10T13:45:55+00:00 Baseline adult d192 6L 6H swiglu rmsnorm success 0.311278 0.8529 0.9088 20 145.4 |
| exp_002_periodic_emb 2026-03-10T13:51:02.651688+00:00 Periodic numeric embeddings k=8 success 0.304675 0.8618 0.9131 20 161.3 |
| exp_003_larger_periodic 2026-03-10T13:57:40.981126+00:00 Periodic + larger d256 8L 8H success 0.305433 0.8593 0.9132 20 276.7 |
| exp_004_periodic_highreg 2026-03-10T14:03:09.869482+00:00 Periodic + higher regularization dropout=0.15 ftd=0.1 wd=3e-4 success 0.306192 0.8606 0.9117 20 279.1 |
| exp_005_periodic_cosine_warmup 2026-03-10T14:08:26.905794+00:00 Periodic + cosine warmup lr=5e-4 30ep success 0.304791 0.8587 0.9126 30 200.5 |
| exp_006_periodic16_low_lr 2026-03-10T14:16:23.325962+00:00 Periodic k=16 lr=1e-4 30ep success 0.304475 0.8612 0.9133 30 440.6 |
| exp_008_periodic16_wide_ff 2026-03-10T14:33:06.140263+00:00 Periodic k=16 + wider FFN d_ff=768 lr=1e-4 success 0.305985 0.8618 0.9136 25 334.4 |
| exp_009_periodic16_gradaccum 2026-03-10T14:40:43.897432+00:00 Periodic k=16 lr=1.5e-4 grad_accum=2 success 0.308625 0.8572 0.9105 30 380.8 |
| exp_010_periodic16_batch128 2026-03-10T14:50:01.858811+00:00 Periodic k=16 lr=2e-4 batch=128 success 0.304936 0.8578 0.9126 25 514.3 |
| exp_011_attn_pool_periodic16 2026-03-10T14:56:43.730505+00:00 Attention pooling + periodic k=16 lr=1e-4 success 0.306606 0.8587 0.9119 30 321.5 |
| exp_012_periodic16_labelsmooth 2026-03-10T15:03:21.525227+00:00 Periodic k=16 lr=1e-4 label_smooth=0.1 success 0.316696 0.8593 0.9126 30 277.0 |
| exp_013_periodic16_seed7 2026-03-10T15:06:35.251092+00:00 Periodic k=16 lr=1e-4 seed=7 success 0.312954 0.8560 0.9084 30 140.6 |
| exp_014_deep8_periodic16 2026-03-10T15:13:14.893440+00:00 Deep 8L + periodic k=16 lr=1e-4 success 0.305749 0.8578 0.9128 30 396.7 |
| exp_015_periodic16_lr2e4 2026-03-10T15:21:48.683501+00:00 Periodic k=16 lr=2e-4 30ep success 0.306264 0.8612 0.9123 30 423.3 |
| exp_016_periodic16_swa 2026-03-10T15:30:50.978920+00:00 Periodic k=16 + SWA avg last 10 epochs success 0.304207 0.8630 0.9138 30 485.3 |
| exp_019_periodic32_swa 2026-03-14T16:01:08.433067+00:00 Periodic k=32 + SWA from epoch 25, lr=1e-4, 40 epochs crash 0.0 Pretrained text encoding requires `transformers`. Install it with `pip install transformers` or switch model.text_encoder back to `custom`.
|
| exp_020_deep10_periodic16_swa 2026-03-14T16:01:08.762377+00:00 Deep 10L + periodic k=16 lr=1e-4 SWA from 20, 35ep crash 0.0 Pretrained text encoding requires `transformers`. Install it with `pip install transformers` or switch model.text_encoder back to `custom`.
|
| exp_021_wide256_periodic16_swa 2026-03-14T16:01:09.106544+00:00 Wide d256 8H + periodic k=16 lr=8e-5 SWA from 25, 40ep crash 0.0 Pretrained text encoding requires `transformers`. Install it with `pip install transformers` or switch model.text_encoder back to `custom`.
|
| exp_022_cosine_warmup_swa 2026-03-14T16:01:09.460270+00:00 Periodic k=16 cosine_warmup lr=2e-4 SWA from 25, 40ep crash 0.0 Pretrained text encoding requires `transformers`. Install it with `pip install transformers` or switch model.text_encoder back to `custom`.
|
| exp_023_dropout02_periodic16_swa 2026-03-14T16:01:09.797021+00:00 Periodic k=16 dropout=0.2 ftd=0.1 lr=1e-4 SWA from 20, 35ep crash 0.0 Pretrained text encoding requires `transformers`. Install it with `pip install transformers` or switch model.text_encoder back to `custom`.
|
| exp_024_lr5e5_periodic16_swa 2026-03-14T16:01:10.108609+00:00 Periodic k=16 lr=5e-5 SWA from 30, 50ep crash 0.0 Pretrained text encoding requires `transformers`. Install it with `pip install transformers` or switch model.text_encoder back to `custom`.
|
| exp_025_wd3e4_periodic16_swa 2026-03-14T16:01:10.418949+00:00 Periodic k=16 lr=1e-4 wd=3e-4 SWA from 20, 35ep crash 0.0 Pretrained text encoding requires `transformers`. Install it with `pip install transformers` or switch model.text_encoder back to `custom`.
|
| exp_026_batch512_periodic16_swa 2026-03-14T16:01:10.739936+00:00 Periodic k=16 batch=512 lr=2e-4 SWA from 25, 40ep crash 0.0 Pretrained text encoding requires `transformers`. Install it with `pip install transformers` or switch model.text_encoder back to `custom`.
|
| exp_027_ff768_periodic16_swa 2026-03-14T16:01:11.047755+00:00 Periodic k=16 d_ff=768 lr=1e-4 SWA from 20, 35ep crash 0.0 Pretrained text encoding requires `transformers`. Install it with `pip install transformers` or switch model.text_encoder back to `custom`.
|
| exp_028_meanpool_periodic16_swa 2026-03-14T16:01:11.371026+00:00 Mean pooling + periodic k=16 lr=1e-4 SWA from 20, 35ep crash 0.0 Pretrained text encoding requires `transformers`. Install it with `pip install transformers` or switch model.text_encoder back to `custom`.
|
| exp_029_cosine_periodic16_swa 2026-03-14T16:01:11.687691+00:00 Periodic k=16 cosine lr=1.5e-4 SWA from 25, 40ep crash 0.0 Pretrained text encoding requires `transformers`. Install it with `pip install transformers` or switch model.text_encoder back to `custom`.
|
| exp_030_gradaccum4_periodic16_swa 2026-03-14T16:01:11.999051+00:00 Periodic k=16 grad_accum=4 lr=3e-4 SWA from 20, 35ep crash 0.0 Pretrained text encoding requires `transformers`. Install it with `pip install transformers` or switch model.text_encoder back to `custom`.
|
| exp_019_periodic32_swa 2026-03-14T16:05:31.971057+00:00 Periodic k=32 + SWA from epoch 25, lr=1e-4, 40 epochs success 0.303895 0.8606 0.9162 40 208.7
|
| exp_020_deep10_periodic16_swa 2026-03-14T16:09:08.555485+00:00 Deep 10L + periodic k=16 lr=1e-4 SWA from 20, 35ep success 0.304103 0.8595 0.9163 35 216.2
|
| exp_021_wide256_periodic16_swa 2026-03-14T16:12:17.664351+00:00 Wide d256 8H + periodic k=16 lr=8e-5 SWA from 25, 40ep success 0.302079 0.8606 0.9161 40 188.7
|
| exp_022_cosine_warmup_swa 2026-03-14T16:15:36.464129+00:00 Periodic k=16 cosine_warmup lr=2e-4 SWA from 25, 40ep success 0.304447 0.8604 0.9164 40 198.4
|
| exp_023_dropout02_periodic16_swa 2026-03-14T16:18:30.760042+00:00 Periodic k=16 dropout=0.2 ftd=0.1 lr=1e-4 SWA from 20, 35ep success 0.306209 0.8565 0.9143 35 173.9
|
| exp_024_lr5e5_periodic16_swa 2026-03-14T16:22:56.113615+00:00 Periodic k=16 lr=5e-5 SWA from 30, 50ep success 0.303644 0.8587 0.9153 50 265.0
|
| exp_025_wd3e4_periodic16_swa 2026-03-14T16:27:08.847681+00:00 Periodic k=16 lr=1e-4 wd=3e-4 SWA from 20, 35ep success 0.302285 0.8589 0.9164 35 252.4
|
| exp_026_batch512_periodic16_swa 2026-03-14T16:29:56.079490+00:00 Periodic k=16 batch=512 lr=2e-4 SWA from 25, 40ep success 0.301533 0.8595 0.9168 40 166.9
|
| exp_027_ff768_periodic16_swa 2026-03-14T16:34:24.136234+00:00 Periodic k=16 d_ff=768 lr=1e-4 SWA from 20, 35ep success 0.302649 0.8628 0.9164 35 267.7
|
| exp_028_meanpool_periodic16_swa 2026-03-14T16:38:37.475591+00:00 Mean pooling + periodic k=16 lr=1e-4 SWA from 20, 35ep success 0.303917 0.8581 0.9147 35 252.9
|
| exp_029_cosine_periodic16_swa 2026-03-14T16:43:25.450043+00:00 Periodic k=16 cosine lr=1.5e-4 SWA from 25, 40ep success 0.302451 0.8597 0.9169 40 287.6
|
| exp_030_gradaccum4_periodic16_swa 2026-03-14T16:47:14.634409+00:00 Periodic k=16 grad_accum=4 lr=3e-4 SWA from 20, 35ep success 0.303882 0.8608 0.9153 35 228.8
|
| exp_031_batch512_wide256_swa 2026-03-14T17:27:20.991195+00:00 Batch=512 + wide d256 8H periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.304212 0.8612 0.9167 40 166.7
|
| exp_032_mixup_batch512_swa 2026-03-14T17:29:50.274673+00:00 Mixup alpha=0.2 + batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.330316 0.8518 0.8977 40 148.9
|
| exp_033_batch512_wd3e4_swa 2026-03-14T17:32:20.825229+00:00 Batch=512 wd=3e-4 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.301821 0.8608 0.9168 40 150.2
|
| exp_034_deep12_batch512_swa 2026-03-14T17:36:09.074101+00:00 Deep 12L batch=512 periodic k=16 lr=1.5e-4 SWA from 25, 40ep success 0.306275 0.8628 0.9153 40 227.9
|
| exp_035_periodic24_batch512_swa 2026-03-14T17:38:39.955795+00:00 Periodic k=24 batch=512 lr=2e-4 SWA from 25, 40ep success 0.304456 0.8600 0.9167 40 150.5
|
| exp_036_cosine_batch512_lr3e4_swa 2026-03-14T17:41:10.404141+00:00 Cosine warmup batch=512 lr=3e-4 periodic k=16 SWA from 25, 40ep success 0.303661 0.8583 0.9164 40 150.1
|
| exp_037_wide256_wd3e4_batch512_swa 2026-03-14T17:43:56.810438+00:00 Wide d256 wd=3e-4 batch=512 lr=1.5e-4 periodic k=16 SWA from 25, 40ep success 0.302834 0.8624 0.9168 40 166.0
|
| exp_038_batch1024_swa 2026-03-14T17:46:03.590420+00:00 Batch=1024 periodic k=16 lr=3e-4 SWA from 25, 40ep success 0.303393 0.8591 0.9158 40 126.4
|
| exp_039_d256_batch512_k32_swa 2026-03-14T17:48:50.541514+00:00 Wide d256 batch=512 periodic k=32 lr=1.5e-4 SWA from 25, 40ep success 0.303876 0.8618 0.9160 40 166.6
|
| exp_040_deep8_d256_batch512_swa 2026-03-14T17:52:11.356260+00:00 Deep 8L d256 8H batch=512 periodic k=16 lr=1e-4 SWA from 25, 40ep success 0.301844 0.8616 0.9166 40 200.4
|
| exp_041_dropout015_wd2e4_batch512_swa 2026-03-14T17:54:41.215364+00:00 Dropout=0.15 wd=2e-4 batch=512 lr=2e-4 periodic k=16 SWA from 25, 40ep success 0.302355 0.8610 0.9167 40 149.5
|
| exp_042_long60_batch512_swa 2026-03-14T17:58:25.273061+00:00 Long 60ep batch=512 lr=2e-4 periodic k=16 SWA from 35 success 0.311481 0.8583 0.9150 60 223.7
|
| exp_060_labelsmooth005_batch512_swa 2026-03-14T18:05:44.477272+00:00 Label smoothing=0.05 batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.304545 0.8595 0.9168 40 153.2
|
| exp_061_labelsmooth01_batch512_swa 2026-03-14T18:08:14.305752+00:00 Label smoothing=0.1 batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.314261 0.8597 0.9168 40 149.5
|
| exp_062_snapshot_5x8 2026-03-14T18:10:45.314643+00:00 Snapshot ensemble: 5 cycles × 8 epochs, average predictions success 0.303110 0.8606 0.9148 40 150.7
|
| exp_063_attnpool_batch512_swa 2026-03-14T18:13:17.418737+00:00 Attention pooling batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.305705 0.8616 0.9159 40 151.8
|
| exp_064_multiseed5_best 2026-03-14T18:25:48.318927+00:00 Multi-seed ensemble: 5 seeds of best config (batch=512 periodic k=16) success 0.301662 0.8600 0.9172 40 750.5
|
| exp_065_swa_wide_15_batch512 2026-03-14T18:28:17.965341+00:00 Wider SWA from epoch 15 batch=512 periodic k=16 lr=2e-4, 40ep success 0.301982 0.8600 0.9164 40 149.3
|
| exp_066_hash_batch512_swa 2026-03-14T18:30:48.959552+00:00 Hash schema encoder batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.303033 0.8626 0.9162 40 150.6
|
| exp_067_ftdrop01_batch512_swa 2026-03-14T18:33:18.838769+00:00 Feature token dropout=0.1 batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.301937 0.8583 0.9168 40 149.5
|
| exp_068_snapshot_8x5 2026-03-14T18:35:48.038535+00:00 Snapshot ensemble: 8 cycles × 5 epochs, average predictions success 0.303439 0.8610 0.9149 40 148.8
|
| exp_069_ls005_wd3e4_batch512_swa 2026-03-14T18:38:00.006065+00:00 Label smooth=0.05 wd=3e-4 batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.304394 0.8587 0.9169 40 131.6
|
| exp_050_transfer_d256_swa 2026-03-14T18:42:06.106374+00:00 Transfer d256 8L finetune-all lr=2e-4 SWA from 25 success 0.307096 0.8600 0.9136 40 125.8
|
| exp_051_transfer_freeze5_swa 2026-03-14T18:44:09.773753+00:00 Transfer freeze-5ep then unfreeze lr=3e-4 SWA from 25 success 0.308911 0.8608 0.9148 40 123.4
|
| exp_052_transfer_lowlr_swa 2026-03-14T18:46:14.315505+00:00 Transfer high head-lr=5e-4 backbone-lr=5e-5 SWA from 25 success 0.309830 0.8600 0.9150 40 124.2
|
| exp_053_transfer_long60_swa 2026-03-14T18:49:21.306848+00:00 Transfer lr=1e-4 60ep SWA from 35 success 0.309144 0.8569 0.9126 60 186.7
|
| exp_054_transfer_cosine_swa 2026-03-14T18:51:25.694018+00:00 Transfer cosine-warmup lr=2e-4 SWA from 25 success 0.309487 0.8575 0.9124 40 124.1
|
| exp_070_combined_best_swa 2026-03-14T18:56:29.714152+00:00 Combined: wd=3e-4 ft_drop=0.1 batch=512 periodic k=16 lr=2e-4 SWA from 15, 40ep success 0.302563 0.8585 0.9164 40 121.3
|
| exp_071_ensemble10_best 2026-03-14T19:15:41.191695+00:00 10-seed ensemble: best config (batch=512 periodic k=16 lr=2e-4 SWA) success 0.301803 0.8610 0.9172 40 1151.1
|
| exp_080_eng_base_swa 2026-03-14T19:22:21.425228+00:00 Engineered features + batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep crash 0.0 string indices must be integers, not 'str'
|
| exp_081_eng_wide256_swa 2026-03-14T19:22:21.633397+00:00 Engineered + d256 8H batch=512 periodic k=16 lr=1.5e-4 SWA from 25, 40ep crash 0.0 string indices must be integers, not 'str'
|
| exp_082_eng_wd3e4_swa 2026-03-14T19:22:21.840543+00:00 Engineered + wd=3e-4 batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep crash 0.0 string indices must be integers, not 'str'
|
| exp_083_eng_deep8_swa 2026-03-14T19:22:22.046765+00:00 Engineered + 8L batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep crash 0.0 string indices must be integers, not 'str'
|
| exp_084_eng_ftdrop_swa 2026-03-14T19:22:22.254770+00:00 Engineered + ft_dropout=0.1 batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep crash 0.0 string indices must be integers, not 'str'
|
| exp_085_eng_swa15_swa 2026-03-14T19:22:22.463689+00:00 Engineered + SWA from 15 batch=512 periodic k=16 lr=2e-4, 40ep crash 0.0 string indices must be integers, not 'str'
|
| exp_086_eng_k32_swa 2026-03-14T19:22:22.671478+00:00 Engineered + periodic k=32 batch=512 lr=2e-4 SWA from 25, 40ep crash 0.0 string indices must be integers, not 'str'
|
| exp_087_eng_d256_8L_swa 2026-03-14T19:22:22.883648+00:00 Engineered + d256 8L 8H batch=512 periodic k=16 lr=1e-4 SWA from 25, 40ep crash 0.0 string indices must be integers, not 'str'
|
| exp_072_diverse_ensemble5 2026-03-14T19:25:23.981169+00:00 Diverse ensemble: 5 diff architectures from best configs success 0.301972 0.8612 0.9168 40 582.5
|
| exp_080_eng_base_swa 2026-03-14T19:26:35.727885+00:00 Engineered features + batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.300691 0.8602 0.9176 40 148.6
|
| exp_081_eng_wide256_swa 2026-03-14T19:29:05.724733+00:00 Engineered + d256 8H batch=512 periodic k=16 lr=1.5e-4 SWA from 25, 40ep success 0.301570 0.8614 0.9174 40 149.6
|
| exp_082_eng_wd3e4_swa 2026-03-14T19:31:21.803115+00:00 Engineered + wd=3e-4 batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.300717 0.8602 0.9176 40 135.7
|
| exp_083_eng_deep8_swa 2026-03-14T19:33:51.066281+00:00 Engineered + 8L batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.303296 0.8649 0.9173 40 148.9
|
| exp_084_eng_ftdrop_swa 2026-03-14T19:36:09.780486+00:00 Engineered + ft_dropout=0.1 batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.300232 0.8604 0.9176 40 138.2
|
| exp_085_eng_swa15_swa 2026-03-14T19:38:27.700415+00:00 Engineered + SWA from 15 batch=512 periodic k=16 lr=2e-4, 40ep success 0.302459 0.8612 0.9163 40 137.6
|
| exp_086_eng_k32_swa 2026-03-14T19:40:44.953589+00:00 Engineered + periodic k=32 batch=512 lr=2e-4 SWA from 25, 40ep success 0.304065 0.8608 0.9163 40 136.8
|
| exp_087_eng_d256_8L_swa 2026-03-14T19:43:41.884461+00:00 Engineered + d256 8L 8H batch=512 periodic k=16 lr=1e-4 SWA from 25, 40ep success 0.300632 0.8616 0.9173 40 176.6
|
| exp_073_ensemble10_wd3e4 2026-03-14T19:46:02.585348+00:00 10-seed ensemble: batch=512 wd=3e-4 periodic k=16 lr=2e-4 SWA success 0.301763 0.8606 0.9172 40 1238.3
|
| exp_090_eng_ftdrop_wd3e4_swa 2026-03-14T19:48:39.761154+00:00 Eng + ft_dropout=0.1 wd=3e-4 batch=512 SWA from 25, 40ep success 0.300254 0.8606 0.9176 40 139.8
|
| exp_091_eng_ftdrop015_swa 2026-03-14T19:51:01.441241+00:00 Eng + ft_dropout=0.15 batch=512 SWA from 25, 40ep success 0.301554 0.8604 0.9168 40 141.3
|
| exp_092_eng_ftdrop01_drop015_swa 2026-03-14T19:53:21.152582+00:00 Eng + ft_dropout=0.1 dropout=0.15 batch=512 SWA from 25, 40ep success 0.302337 0.8614 0.9170 40 139.4
|
| exp_093_eng_ftdrop01_d256_swa 2026-03-14T19:55:57.350775+00:00 Eng + ft_dropout=0.1 d256 8H batch=512 lr=1.5e-4 SWA from 25, 40ep success 0.300951 0.8634 0.9172 40 155.8
|
| exp_074_mega_diverse_10 2026-03-14T20:07:32.922592+00:00 Mega diverse: 10 different arch+seed combos success 0.301863 0.8612 0.9171 40 1290.1
|
| exp_094_eng_ensemble10_ftdrop 2026-03-14T20:16:59.679153+00:00 10-seed ensemble: eng + ft_dropout=0.1 batch=512 SWA success 0.300479 0.8608 0.9176 40 1261.9
|
| exp_095_eng_ftdrop01_cosine_swa 2026-03-14T20:19:02.086332+00:00 Eng + ft_dropout=0.1 cosine_warmup lr=2e-4 batch=512 SWA from 25, 40ep success 0.303769 0.8585 0.9154 40 122.2
|
| exp_096_eng_ftdrop01_swa20 2026-03-14T20:20:42.406487+00:00 Eng + ft_dropout=0.1 batch=512 SWA from 20, 40ep success 0.301320 0.8608 0.9170 40 99.9
|
| exp_097_eng_ftdrop01_swa10 2026-03-14T20:22:17.455572+00:00 Eng + ft_dropout=0.1 batch=512 SWA from 10, 40ep (30 checkpoints) success 0.302908 0.8595 0.9159 40 94.7
|
| exp_100_v2_base_swa 2026-03-14T20:35:41.369806+00:00 v2 data + ft_dropout=0.1 batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.303195 0.8614 0.9161 40 156.7
|
| exp_101_v2_wd3e4_swa 2026-03-14T20:38:16.042863+00:00 v2 + wd=3e-4 batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.303151 0.8612 0.9162 40 154.3
|
| exp_102_v2_d256_swa 2026-03-14T20:41:44.152904+00:00 v2 + d256 8H batch=512 periodic k=16 lr=1.5e-4 SWA from 25, 40ep success 0.301656 0.8597 0.9164 40 207.6
|
| exp_103_v2_deep8_swa 2026-03-14T20:44:37.034341+00:00 v2 + 8L batch=512 periodic k=16 lr=2e-4 SWA from 25, 40ep success 0.304095 0.8626 0.9168 40 172.4
|
| exp_104_v2_d256_8L_swa 2026-03-14T20:50:02.803978+00:00 v2 + d256 8L 8H batch=512 periodic k=16 lr=1e-4 SWA from 25, 40ep success 0.301637 0.8581 0.9167 40 325.4
|
| exp_110_eng_80ep_swa50 2026-03-14T22:17:44.955848+00:00 Eng + 80ep lr=1.5e-4 SWA from 50 success 0.302702 0.8612 0.9176 80 852.9
|
| exp_111_eng_gradaccum4 2026-03-14T22:24:41.740748+00:00 Eng + grad_accum=4 eff_batch=2048 lr=4e-4 SWA success 0.302784 0.8595 0.9154 40 415.7
|
| exp_112_eng_ls002 2026-03-14T22:31:46.148557+00:00 Eng + label_smoothing=0.02 SWA success 0.299319 0.8597 0.9176 40 422.9
|
| exp_114_eng_cosine60 2026-03-14T22:42:54.945341+00:00 Eng + cosine_warmup 60ep warmup=3 SWA from 40 success 0.299142 0.8612 0.9183 60 665.2
|
| exp_115_eng_deep12_d128 2026-03-14T22:52:33.629255+00:00 Eng + 12L d128 4H d_ff=256 SWA success 0.304036 0.8579 0.9158 40 576.5
|
| exp_116_eng_wide384_4L 2026-03-14T22:59:02.648818+00:00 Eng + d384 4L 12H d_ff=768 lr=1e-4 SWA success 0.305207 0.8571 0.9146 40 387.0
|
| exp_118_eng_lowlr60 2026-03-14T23:10:12.574753+00:00 Eng + lr=1e-4 60ep SWA from 40 success 0.300814 0.8593 0.9174 60 669.6
|
| exp_119_eng_ftd15_cosine 2026-03-14T23:19:01.680692+00:00 Eng + ft_dropout=0.15 cosine_warmup 50ep SWA from 30 success 0.300477 0.8583 0.9172 50 528.7
|
| exp_120_ls02_cosine60 2026-03-14T23:31:23.260883+00:00 Eng + ls=0.02 + cosine_warmup 60ep SWA from 40 success 0.298519 0.8606 0.9185 60 217.2
|
| exp_121_ls02_cosine60_swa35 2026-03-14T23:34:27.116187+00:00 Eng + ls=0.02 + cosine_warmup 60ep SWA from 35 success 0.298557 0.8606 0.9185 60 183.5
|
| exp_122_ls02_cosine80 2026-03-14T23:38:36.550798+00:00 Eng + ls=0.02 + cosine_warmup 80ep SWA from 55 success 0.298173 0.8614 0.9190 80 249.0
|
| exp_123_ls01_cosine60 2026-03-14T23:41:29.846536+00:00 Eng + ls=0.01 + cosine_warmup 60ep SWA from 40 success 0.297603 0.8624 0.9188 60 172.9
|
| exp_124_ls02_cosine_lowlr 2026-03-14T23:44:34.410626+00:00 Eng + ls=0.02 + cosine_warmup lr=1.5e-4 60ep SWA from 40 success 0.300306 0.8600 0.9175 60 184.2
|
| exp_125_ls02_cosine_wd3e4 2026-03-14T23:47:38.332509+00:00 Eng + ls=0.02 + cosine_warmup wd=3e-4 60ep SWA from 40 success 0.298560 0.8606 0.9185 60 183.5
|
| exp_126_ls03_cosine60 2026-03-14T23:50:44.768205+00:00 Eng + ls=0.03 + cosine_warmup 60ep SWA from 40 success 0.300338 0.8626 0.9183 60 186.1
|
| exp_127_ls02_cosine_ftd05 2026-03-14T23:53:48.987813+00:00 Eng + ls=0.02 + cosine_warmup ft_dropout=0.05 60ep SWA from 40 success 0.297698 0.8622 0.9190 60 183.8
|
| exp_130_ls005_cosine60 2026-03-14T23:59:09.563525+00:00 Eng + ls=0.005 + cosine_warmup 60ep SWA from 40 success 0.298151 0.8618 0.9185 60 189.2
|
| exp_131_ls01_cosine_ftd05 2026-03-15T00:02:05.984044+00:00 Eng + ls=0.01 + cosine_warmup + ft_dropout=0.05 60ep SWA40 success 0.298119 0.8614 0.9186 60 176.1
|
| exp_132_ls01_cosine80 2026-03-15T00:06:09.830132+00:00 Eng + ls=0.01 + cosine_warmup 80ep SWA from 55 success 0.299673 0.8612 0.9183 80 243.5
|
| exp_133_ls01_cosine_ftd15 2026-03-15T00:09:09.092132+00:00 Eng + ls=0.01 + cosine_warmup + ft_dropout=0.15 60ep SWA40 success 0.299300 0.8602 0.9180 60 178.9
|
| exp_134_ls01_cosine_warm5 2026-03-15T00:12:14.662779+00:00 Eng + ls=0.01 + cosine_warmup warmup=5 60ep SWA40 success 0.297779 0.8610 0.9188 60 185.2
|
| exp_135_ls01_cosine_wd5e5 2026-03-15T00:15:17.884870+00:00 Eng + ls=0.01 + cosine_warmup wd=5e-5 60ep SWA40 success 0.297603 0.8624 0.9188 60 182.8
|
| exp_136_ls01_cosine_drop05 2026-03-15T00:18:20.601020+00:00 Eng + ls=0.01 + cosine_warmup dropout=0.05 60ep SWA40 success 0.298339 0.8602 0.9184 60 182.3
|
| exp_137_ls01_cosine_swa30 2026-03-15T00:21:18.996783+00:00 Eng + ls=0.01 + cosine_warmup 60ep SWA from 30 success 0.297611 0.8616 0.9187 60 178.0
|
| exp_140_seed123 2026-03-15T00:33:43.995185+00:00 Best config seed=123 success 0.301722 0.8614 0.9161 60 145.0
|
| exp_141_seed789 2026-03-15T00:36:04.803487+00:00 Best config seed=789 success 0.301797 0.8591 0.9158 60 140.4
|
| exp_142_periodic32 2026-03-15T00:38:42.216613+00:00 Best config + periodic k=32 success 0.301900 0.8616 0.9164 60 157.0
|
| exp_143_periodic8 2026-03-15T00:41:43.961206+00:00 Best config + periodic k=8 success 0.300685 0.8622 0.9174 60 181.4
|
| exp_144_mean_pool 2026-03-15T00:44:43.666667+00:00 Best config + mean pooling success 0.297888 0.8610 0.9182 60 179.3
|
| exp_145_wd5e5 2026-03-15T00:47:43.633939+00:00 Best config + wd=5e-5 success 0.297603 0.8624 0.9188 60 179.6
|
| exp_146_8layers 2026-03-15T00:51:05.528689+00:00 Best config 8L d192 success 0.298874 0.8647 0.9179 60 201.5
|
| exp_147_5seed_ensemble 2026-03-15T01:23:56.112844+00:00 5-seed ensemble ls=0.01 cosine 60ep SWA40 success 0.298922 0.8604 0.9177 60 1970.2
|
| exp_150_mean_ls01_cosine 2026-03-15T01:33:02.735342+00:00 Eng + mean pooling + ls=0.01 + cosine 60ep SWA40 success 0.297888 0.8610 0.9182 60 386.6
|
| exp_151_batch1024 2026-03-15T01:38:19.231464+00:00 Eng + batch=1024 lr=3e-4 ls=0.01 cosine 60ep SWA40 success 0.299009 0.8591 0.9176 60 316.1
|
| exp_152_batch256 2026-03-15T01:49:59.628155+00:00 Eng + batch=256 lr=1.5e-4 ls=0.01 cosine 60ep SWA40 success 0.298515 0.8610 0.9183 60 697.8
|
| exp_153_3heads 2026-03-15T01:57:04.274284+00:00 Eng + 3 heads ls=0.01 cosine 60ep SWA40 success 0.300333 0.8612 0.9173 60 418.9
|
| exp_154_12heads 2026-03-15T02:04:04.422270+00:00 Eng + 12 heads ls=0.01 cosine 60ep SWA40 success 0.299629 0.8614 0.9181 60 419.7
|
| exp_155_ff768 2026-03-15T02:10:57.628738+00:00 Eng + d_ff=768 ls=0.01 cosine 60ep SWA40 success 0.300390 0.8622 0.9174 60 412.8
|
| exp_156_ff192 2026-03-15T02:17:32.587917+00:00 Eng + d_ff=192 ls=0.01 cosine 60ep SWA40 success 0.299577 0.8624 0.9175 60 390.9
|
| exp_157_mean_b256_80ep 2026-03-15T02:33:13.468599+00:00 Eng + mean pool + batch=256 + ls=0.01 cosine 80ep SWA55 success 0.306390 0.8597 0.9141 80 940.5
|
|
|
