restage best simple adapt wce model

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e524d8c6de17a12223d54579bc7603576fdd2251c1ccdaa256df78f559b534a0
+oid sha256:623355e75f46333ac7063cac2faa61ee6310cd11f3ce678bf973bc7ad41d1559
 size 1845323269

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b20143719beb78224c536731c8f9a54a1fe8212f89dd7b15cd99f45d03524db2
+oid sha256:ee38a6de8dd03283c1ba66832e911e67047f19a08be134ef62922fdc507bd8bb
 size 922885701

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c8ef47105d233625a0f710b074abad1745797dfbec46b3eb1333b79ae26efa82
+oid sha256:6ac50aa0bc2681fcf0be968dbafd36f389455d88752c8cf6a8c34cc5388bfe22
 size 14575

scaler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:48343ace91e4b0aed71d9de2ef599675c24fab85baa5168e44ded595d08da559
+oid sha256:8a4375b5fb7fce1ea4586fbea8f86d90114295850e28fbc15b989d4a546bad7f
 size 557

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b2d280d247d94ac98ee856efffc708e91babbdb3306480c0c30168fc32146e59
+oid sha256:d841877dae97f3484921bd550ae3afec1339fe3ea6e45d722c6eee90ccd95ea9
 size 627

trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 43.5904,
-  "best_model_checkpoint": "opus_big_adapt_wce_gloss_train-sampled_bands_4_ubweight_1.5/checkpoint-80000",
+  "best_metric": 43.8722,
+  "best_model_checkpoint": "opus_big_adapt_wce_gloss_train-sampled_bands_5_ubweight_1.75/checkpoint-80000",
   "epoch": 1.9716574245224892,
   "global_step": 80000,
   "is_hyper_param_search": false,
@@ -10,161 +10,161 @@
     {
       "epoch": 0.2,
       "learning_rate": 1.9753635243376465e-05,
-      "loss": 0.176,
+      "loss": 0.1776,
       "step": 8000
     },
     {
       "epoch": 0.2,
-      "eval_bleu": 42.0449,
-      "eval_gen_len": 33.5264,
-      "eval_loss": 0.09838402271270752,
-      "eval_runtime": 157.9869,
-      "eval_samples_per_second": 6.602,
-      "eval_steps_per_second": 0.418,
+      "eval_bleu": 42.0642,
+      "eval_gen_len": 33.7881,
+      "eval_loss": 0.09869997948408127,
+      "eval_runtime": 84.9196,
+      "eval_samples_per_second": 12.282,
+      "eval_steps_per_second": 0.777,
       "step": 8000
     },
     {
       "epoch": 0.39,
-      "learning_rate": 1.9507270486752928e-05,
-      "loss": 0.1453,
+      "learning_rate": 1.950723967960567e-05,
+      "loss": 0.1467,
       "step": 16000
     },
     {
       "epoch": 0.39,
-      "eval_bleu": 42.2559,
-      "eval_gen_len": 34.0422,
-      "eval_loss": 0.09732737392187119,
-      "eval_runtime": 127.5441,
-      "eval_samples_per_second": 8.178,
-      "eval_steps_per_second": 0.517,
+      "eval_bleu": 41.7526,
+      "eval_gen_len": 34.7795,
+      "eval_loss": 0.09779118001461029,
+      "eval_runtime": 93.0132,
+      "eval_samples_per_second": 11.213,
+      "eval_steps_per_second": 0.71,
       "step": 16000
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.926090573012939e-05,
-      "loss": 0.1414,
+      "loss": 0.1426,
       "step": 24000
     },
     {
       "epoch": 0.59,
-      "eval_bleu": 42.404,
-      "eval_gen_len": 33.6731,
-      "eval_loss": 0.0977756679058075,
-      "eval_runtime": 115.3689,
-      "eval_samples_per_second": 9.041,
-      "eval_steps_per_second": 0.572,
+      "eval_bleu": 42.0584,
+      "eval_gen_len": 33.464,
+      "eval_loss": 0.09806442260742188,
+      "eval_runtime": 80.6234,
+      "eval_samples_per_second": 12.937,
+      "eval_steps_per_second": 0.819,
       "step": 24000
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.90145101663586e-05,
-      "loss": 0.1375,
+      "learning_rate": 1.9014540973505854e-05,
+      "loss": 0.1387,
       "step": 32000
     },
     {
       "epoch": 0.79,
-      "eval_bleu": 42.4354,
-      "eval_gen_len": 34.1505,
-      "eval_loss": 0.09676187485456467,
-      "eval_runtime": 123.6757,
-      "eval_samples_per_second": 8.433,
-      "eval_steps_per_second": 0.534,
+      "eval_bleu": 42.6266,
+      "eval_gen_len": 33.5254,
+      "eval_loss": 0.09653611481189728,
+      "eval_runtime": 81.9234,
+      "eval_samples_per_second": 12.731,
+      "eval_steps_per_second": 0.806,
       "step": 32000
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.876814540973506e-05,
-      "loss": 0.1361,
+      "loss": 0.1373,
       "step": 40000
     },
     {
       "epoch": 0.99,
-      "eval_bleu": 42.9091,
-      "eval_gen_len": 33.7948,
-      "eval_loss": 0.09602385014295578,
-      "eval_runtime": 121.7774,
-      "eval_samples_per_second": 8.565,
-      "eval_steps_per_second": 0.542,
+      "eval_bleu": 42.6264,
+      "eval_gen_len": 33.9195,
+      "eval_loss": 0.09597118198871613,
+      "eval_runtime": 81.6116,
+      "eval_samples_per_second": 12.78,
+      "eval_steps_per_second": 0.809,
       "step": 40000
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.852178065311152e-05,
-      "loss": 0.1212,
+      "loss": 0.1222,
       "step": 48000
     },
     {
       "epoch": 1.18,
-      "eval_bleu": 42.9528,
-      "eval_gen_len": 33.884,
-      "eval_loss": 0.09735256433486938,
-      "eval_runtime": 137.452,
-      "eval_samples_per_second": 7.588,
-      "eval_steps_per_second": 0.48,
+      "eval_bleu": 42.7749,
+      "eval_gen_len": 34.1208,
+      "eval_loss": 0.09734106063842773,
+      "eval_runtime": 94.7556,
+      "eval_samples_per_second": 11.007,
+      "eval_steps_per_second": 0.697,
       "step": 48000
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.8275415896487988e-05,
-      "loss": 0.1215,
+      "loss": 0.1226,
       "step": 56000
     },
     {
       "epoch": 1.38,
-      "eval_bleu": 42.644,
-      "eval_gen_len": 33.4746,
-      "eval_loss": 0.09731972962617874,
-      "eval_runtime": 116.3034,
-      "eval_samples_per_second": 8.968,
-      "eval_steps_per_second": 0.567,
+      "eval_bleu": 42.687,
+      "eval_gen_len": 34.1697,
+      "eval_loss": 0.09724980592727661,
+      "eval_runtime": 191.9924,
+      "eval_samples_per_second": 5.433,
+      "eval_steps_per_second": 0.344,
       "step": 56000
     },
     {
       "epoch": 1.58,
-      "learning_rate": 1.802908194701171e-05,
-      "loss": 0.1221,
+      "learning_rate": 1.802905113986445e-05,
+      "loss": 0.1233,
       "step": 64000
     },
     {
       "epoch": 1.58,
-      "eval_bleu": 42.5832,
-      "eval_gen_len": 34.5647,
-      "eval_loss": 0.09653764218091965,
-      "eval_runtime": 165.7514,
-      "eval_samples_per_second": 6.293,
-      "eval_steps_per_second": 0.398,
+      "eval_bleu": 42.9609,
+      "eval_gen_len": 34.3432,
+      "eval_loss": 0.09664924442768097,
+      "eval_runtime": 125.3853,
+      "eval_samples_per_second": 8.318,
+      "eval_steps_per_second": 0.526,
       "step": 64000
     },
     {
       "epoch": 1.77,
-      "learning_rate": 1.778268638324091e-05,
-      "loss": 0.1217,
+      "learning_rate": 1.7782717190388173e-05,
+      "loss": 0.1227,
       "step": 72000
     },
     {
       "epoch": 1.77,
-      "eval_bleu": 42.6081,
-      "eval_gen_len": 34.232,
-      "eval_loss": 0.09667492657899857,
-      "eval_runtime": 124.0305,
-      "eval_samples_per_second": 8.409,
-      "eval_steps_per_second": 0.532,
+      "eval_bleu": 42.7711,
+      "eval_gen_len": 34.1985,
+      "eval_loss": 0.09645862132310867,
+      "eval_runtime": 85.6207,
+      "eval_samples_per_second": 12.182,
+      "eval_steps_per_second": 0.771,
       "step": 72000
     },
     {
       "epoch": 1.97,
-      "learning_rate": 1.7536352433764633e-05,
-      "loss": 0.1213,
+      "learning_rate": 1.7536321626617377e-05,
+      "loss": 0.1224,
       "step": 80000
     },
     {
       "epoch": 1.97,
-      "eval_bleu": 43.5904,
-      "eval_gen_len": 33.9358,
-      "eval_loss": 0.0963975265622139,
-      "eval_runtime": 128.1155,
-      "eval_samples_per_second": 8.141,
-      "eval_steps_per_second": 0.515,
+      "eval_bleu": 43.8722,
+      "eval_gen_len": 33.5513,
+      "eval_loss": 0.09601087123155594,
+      "eval_runtime": 85.5385,
+      "eval_samples_per_second": 12.193,
+      "eval_steps_per_second": 0.772,
       "step": 80000
     }
   ],

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b8dc89fcc09c93904c5d19ae7f0a0e37bb91f96030931841ba0eb8773e6739c6
+oid sha256:8194c675c873b06e95939930fd626a08ef38cefda0506613da118f3fcdb61963
 size 3835