Pivot / training /train_report.json
Q1z's picture
Pivot H200 full retrain
14bf8c2 verified
Raw History Blame Contribute Delete
4.1 kB
{
"history": [
{
"epoch": 0,
"phase": "ce",
"global_step": 2251,
"train_L_dec": 0.3132325249438689,
"train_L_cal": 0.022969920731528693,
"train_acc": 0.8643241892492226,
"val_acc": 0.9746167518329261,
"val_brier": 0.004308474582756061,
"val_ece": 0.00984656126080116,
"seconds": 1176.4715111255646,
"stability_spike_warnings_total": 2
},
{
"epoch": 1,
"phase": "ce",
"global_step": 4502,
"train_L_dec": 0.03158559516649023,
"train_L_cal": 0.0016624860179579265,
"train_acc": 0.9905389271434918,
"val_acc": 0.9796711841812931,
"val_brier": 0.0036710468228806767,
"val_ece": 0.010786583753486019,
"seconds": 1174.2428555488586,
"stability_spike_warnings_total": 5
},
{
"epoch": 2,
"phase": "pcgrad",
"global_step": 6753,
"train_L_dec": 0.021729152236390645,
"train_L_cal": 0.0011230140197688602,
"train_acc": 0.9936347734340293,
"val_acc": 0.9828371473005999,
"val_brier": 0.0029678672562138916,
"val_ece": 0.007705523323957138,
"seconds": 1947.3289678096771,
"stability_spike_warnings_total": 5
},
{
"epoch": 3,
"phase": "pcgrad",
"global_step": 9004,
"train_L_dec": 0.01443692476023041,
"train_L_cal": 0.0007580307265767108,
"train_acc": 0.995620002221235,
"val_acc": 0.9836702954898912,
"val_brier": 0.002991214513442398,
"val_ece": 0.008586534102594447,
"seconds": 1941.522153377533,
"stability_spike_warnings_total": 11
},
{
"epoch": 4,
"phase": "pcgrad",
"global_step": 11255,
"train_L_dec": 0.012373026743064744,
"train_L_cal": 0.00065262238645809,
"train_acc": 0.9961614282541092,
"val_acc": 0.9843368140413241,
"val_brier": 0.002747937278662336,
"val_ece": 0.009184632024925778,
"seconds": 1943.6136224269867,
"stability_spike_warnings_total": 18
},
{
"epoch": 5,
"phase": "pcgrad",
"global_step": 13506,
"train_L_dec": 0.010130272916737846,
"train_L_cal": 0.0004837604737681557,
"train_acc": 0.9971001450351374,
"val_acc": 0.9843368140413241,
"val_brier": 0.0026608713938844693,
"val_ece": 0.009472633415071256,
"seconds": 1943.7929818630219,
"stability_spike_warnings_total": 48
},
{
"epoch": 6,
"phase": "pcgrad",
"global_step": 15757,
"train_L_dec": 0.008186050639382882,
"train_L_cal": 0.0003799829597298574,
"train_acc": 0.9977093513993781,
"val_acc": 0.9847811597422795,
"val_brier": 0.0027190133560318773,
"val_ece": 0.009519949778864792,
"seconds": 1945.6040878295898,
"stability_spike_warnings_total": 83
},
{
"epoch": 7,
"phase": "pcgrad",
"global_step": 18008,
"train_L_dec": 0.009192034906286452,
"train_L_cal": 0.0004240516285463041,
"train_acc": 0.9975358174144825,
"val_acc": 0.9848922461675184,
"val_brier": 0.0026594605793753444,
"val_ece": 0.009249331458982588,
"seconds": 1939.7083559036255,
"stability_spike_warnings_total": 128
}
],
"best_acc": {
"acc": 0.9848922461675184,
"brier": 0.0026594605793753444,
"epoch": 7
},
"best_brier": {
"brier": 0.0026594605793753444,
"acc": 0.9848922461675184,
"epoch": 7
},
"split_sizes": {
"train": 144017,
"val": 18004,
"holdout": 18002
},
"pretrained_audit": {
"wrapper_class": "Lfm2BidirectionalForMaskedLM",
"encoder_class": "Lfm2BidirectionalModel",
"missing_keys": [],
"unexpected_keys": [],
"parameter_count": 354483968,
"fingerprints": {
"embed_tokens.weight": "b9d03e9055d7428a6467f81860d5ebb58286174da60f58364ecd66fdcfa58dd3",
"layers.0.feed_forward.w1.weight": "86a63d2c73d6334c58dba4e68282f97e70e9dc1342b3b06bb83b8451b86b4f69"
}
},
"total_params": 357631745,
"trainable_params": 357631745,
"stability_spike_warnings_total": 128
}