Training in progress, step 200

all_results.json

@@ -1,20 +1,20 @@
 {
     "epoch": 3.0,
-    "eval_average_metrics": 90.63138686131387,
-    "eval_em": 90.36496350364963,
-    "eval_f1": 90.89781021897811,
-    "eval_loss": 0.2141435146331787,
-    "eval_runtime": 15.7609,
-    "eval_samples_per_second": 86.924,
-    "test_average_metrics": 74.47490712152019,
-    "test_em": 73.99182920400632,
-    "test_f1": 74.95798503903406,
-    "test_loss": 0.48021578788757324,
-    "test_runtime": 161.768,
-    "test_samples_per_second": 93.813,
-    "train_loss": 0.3297047741756515,
-    "train_runtime": 5659.5381,
+    "eval_average_metrics": 91.36131386861314,
+    "eval_em": 91.0948905109489,
+    "eval_f1": 91.62773722627739,
+    "eval_loss": 0.21504777669906616,
+    "eval_runtime": 32.7505,
+    "eval_samples_per_second": 41.831,
+    "test_average_metrics": 75.00792291086177,
+    "test_em": 74.50579862941487,
+    "test_f1": 75.51004719230866,
+    "test_loss": 0.48300713300704956,
+    "test_runtime": 342.8147,
+    "test_samples_per_second": 44.269,
+    "train_loss": 0.3175360437943145,
+    "train_runtime": 17672.1841,
     "train_samples": 137484,
-    "train_samples_per_second": 72.877,
-    "train_steps_per_second": 4.555
+    "train_samples_per_second": 23.339,
+    "train_steps_per_second": 1.459
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 3.0,
-    "eval_average_metrics": 90.63138686131387,
-    "eval_em": 90.36496350364963,
-    "eval_f1": 90.89781021897811,
-    "eval_loss": 0.2141435146331787,
-    "eval_runtime": 15.7609,
-    "eval_samples_per_second": 86.924
+    "eval_average_metrics": 91.36131386861314,
+    "eval_em": 91.0948905109489,
+    "eval_f1": 91.62773722627739,
+    "eval_loss": 0.21504777669906616,
+    "eval_runtime": 32.7505,
+    "eval_samples_per_second": 41.831
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e8ace9158b4ccbc5041040d01c8ebdd84c66838368f240b1990392deb3dd5139
-size 7551621
+oid sha256:73bb270c9eb59d9f8a29f0bc108b1a26a26964434e7c8e43708ee7b37a511790
+size 2631685

runs/Feb01_01-41-55_node1/events.out.tfevents.1643651046.node1

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8b890bc2228a1105ec12abd606bd641c665556eb421c44827cc0fb330c331aae
-size 58695
+oid sha256:f6c57d4a2a5d00b66d8713b9942e1625bfc930480ae2c78809d0534326af5a44
+size 59424

runs/Feb01_01-41-55_node1/events.out.tfevents.1643668752.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6ee097214c3435788d0149fdfe0d0613c4683aa80f0538d0be9deaedca10c049
+size 778

runs/Feb02_15-54-46_node1/1643788552.5548458/events.out.tfevents.1643788552.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:427de076ab994390934567ade01150a3a1e5ba6cd2847dd5851a17cd3d1506af
+size 5047

runs/Feb02_15-54-46_node1/events.out.tfevents.1643788552.node1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:84ff116273de9d30f8e85b06aaf6d6da8f411a96c210e7073f295f39f89a946b
+size 4335

test_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 3.0,
-    "test_average_metrics": 74.47490712152019,
-    "test_em": 73.99182920400632,
-    "test_f1": 74.95798503903406,
-    "test_loss": 0.48021578788757324,
-    "test_runtime": 161.768,
-    "test_samples_per_second": 93.813
+    "test_average_metrics": 75.00792291086177,
+    "test_em": 74.50579862941487,
+    "test_f1": 75.51004719230866,
+    "test_loss": 0.48300713300704956,
+    "test_runtime": 342.8147,
+    "test_samples_per_second": 44.269
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 3.0,
-    "train_loss": 0.3297047741756515,
-    "train_runtime": 5659.5381,
+    "train_loss": 0.3175360437943145,
+    "train_runtime": 17672.1841,
     "train_samples": 137484,
-    "train_samples_per_second": 72.877,
-    "train_steps_per_second": 4.555
+    "train_samples_per_second": 23.339,
+    "train_steps_per_second": 1.459
 }

trainer_state.json

@@ -1,841 +1,1611 @@
 {
-  "best_metric": 90.63138686131387,
-  "best_model_checkpoint": "outputs/bitfit/t5-base/superglue-record/checkpoint-21500",
+  "best_metric": 91.36131386861314,
+  "best_model_checkpoint": "outputs/bitfit/t5-base/superglue-record/checkpoint-1200",
   "epoch": 3.0,
   "global_step": 25779,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
+    {
+      "epoch": 0.02,
+      "eval_average_metrics": 89.01338199513381,
+      "eval_em": 88.75912408759125,
+      "eval_f1": 89.2676399026764,
+      "eval_loss": 0.2185831516981125,
+      "eval_runtime": 34.599,
+      "eval_samples_per_second": 39.597,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "eval_average_metrics": 89.71289537712894,
+      "eval_em": 89.48905109489051,
+      "eval_f1": 89.93673965936739,
+      "eval_loss": 0.22008372843265533,
+      "eval_runtime": 36.1565,
+      "eval_samples_per_second": 37.891,
+      "step": 400
+    },
     {
       "epoch": 0.06,
       "learning_rate": 0.0002941813103689049,
-      "loss": 0.3506,
+      "loss": 0.3381,
       "step": 500
     },
     {
-      "epoch": 0.06,
-      "eval_average_metrics": 90.1404240528328,
-      "eval_em": 89.85401459854015,
-      "eval_f1": 90.42683350712547,
-      "eval_loss": 0.23184123635292053,
-      "eval_runtime": 15.5789,
-      "eval_samples_per_second": 87.939,
-      "step": 500
+      "epoch": 0.07,
+      "eval_average_metrics": 89.17518248175182,
+      "eval_em": 88.9051094890511,
+      "eval_f1": 89.44525547445255,
+      "eval_loss": 0.22487574815750122,
+      "eval_runtime": 38.7912,
+      "eval_samples_per_second": 35.317,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "eval_average_metrics": 90.32725060827251,
+      "eval_em": 89.92700729927007,
+      "eval_f1": 90.72749391727494,
+      "eval_loss": 0.22092510759830475,
+      "eval_runtime": 38.8963,
+      "eval_samples_per_second": 35.222,
+      "step": 800
     },
     {
       "epoch": 0.12,
       "learning_rate": 0.0002883626207378098,
-      "loss": 0.3505,
+      "loss": 0.3288,
       "step": 1000
     },
     {
       "epoch": 0.12,
-      "eval_average_metrics": 89.19586374695865,
-      "eval_em": 88.75912408759125,
-      "eval_f1": 89.63260340632604,
-      "eval_loss": 0.2271701991558075,
-      "eval_runtime": 15.9901,
-      "eval_samples_per_second": 85.678,
+      "eval_average_metrics": 90.82603406326035,
+      "eval_em": 90.58394160583941,
+      "eval_f1": 91.06812652068128,
+      "eval_loss": 0.2161024510860443,
+      "eval_runtime": 38.0375,
+      "eval_samples_per_second": 36.017,
       "step": 1000
     },
+    {
+      "epoch": 0.14,
+      "eval_average_metrics": 91.36131386861314,
+      "eval_em": 91.0948905109489,
+      "eval_f1": 91.62773722627739,
+      "eval_loss": 0.21504777669906616,
+      "eval_runtime": 36.1024,
+      "eval_samples_per_second": 37.948,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "eval_average_metrics": 89.69829683698296,
+      "eval_em": 89.41605839416059,
+      "eval_f1": 89.98053527980535,
+      "eval_loss": 0.21785993874073029,
+      "eval_runtime": 38.4041,
+      "eval_samples_per_second": 35.673,
+      "step": 1400
+    },
     {
       "epoch": 0.17,
       "learning_rate": 0.00028254393110671474,
-      "loss": 0.3559,
+      "loss": 0.333,
       "step": 1500
     },
     {
-      "epoch": 0.17,
-      "eval_average_metrics": 87.68126520681264,
-      "eval_em": 87.37226277372262,
-      "eval_f1": 87.99026763990267,
-      "eval_loss": 0.23542903363704681,
-      "eval_runtime": 15.5654,
-      "eval_samples_per_second": 88.016,
-      "step": 1500
+      "epoch": 0.19,
+      "eval_average_metrics": 89.78345498783455,
+      "eval_em": 89.34306569343066,
+      "eval_f1": 90.22384428223845,
+      "eval_loss": 0.21883782744407654,
+      "eval_runtime": 35.4761,
+      "eval_samples_per_second": 38.618,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "eval_average_metrics": 89.54866180048661,
+      "eval_em": 89.41605839416059,
+      "eval_f1": 89.68126520681264,
+      "eval_loss": 0.22056403756141663,
+      "eval_runtime": 35.8766,
+      "eval_samples_per_second": 38.186,
+      "step": 1800
     },
     {
       "epoch": 0.23,
       "learning_rate": 0.0002767252414756197,
-      "loss": 0.3487,
+      "loss": 0.3283,
       "step": 2000
     },
     {
       "epoch": 0.23,
-      "eval_average_metrics": 88.3946819603754,
-      "eval_em": 87.95620437956204,
-      "eval_f1": 88.83315954118875,
-      "eval_loss": 0.23398332297801971,
-      "eval_runtime": 15.718,
-      "eval_samples_per_second": 87.161,
+      "eval_average_metrics": 88.84306569343067,
+      "eval_em": 88.39416058394161,
+      "eval_f1": 89.29197080291972,
+      "eval_loss": 0.22133098542690277,
+      "eval_runtime": 36.0963,
+      "eval_samples_per_second": 37.954,
       "step": 2000
     },
+    {
+      "epoch": 0.26,
+      "eval_average_metrics": 89.49391727493916,
+      "eval_em": 89.19708029197079,
+      "eval_f1": 89.79075425790754,
+      "eval_loss": 0.22017844021320343,
+      "eval_runtime": 36.292,
+      "eval_samples_per_second": 37.749,
+      "step": 2200
+    },
+    {
+      "epoch": 0.28,
+      "eval_average_metrics": 90.77737226277371,
+      "eval_em": 90.43795620437956,
+      "eval_f1": 91.11678832116786,
+      "eval_loss": 0.21572071313858032,
+      "eval_runtime": 38.375,
+      "eval_samples_per_second": 35.7,
+      "step": 2400
+    },
     {
       "epoch": 0.29,
       "learning_rate": 0.0002709065518445246,
-      "loss": 0.3461,
+      "loss": 0.3258,
       "step": 2500
     },
     {
-      "epoch": 0.29,
-      "eval_average_metrics": 88.34428223844282,
-      "eval_em": 88.02919708029196,
-      "eval_f1": 88.65936739659367,
-      "eval_loss": 0.23554831743240356,
-      "eval_runtime": 15.4728,
-      "eval_samples_per_second": 88.542,
-      "step": 2500
+      "epoch": 0.3,
+      "eval_average_metrics": 90.58880778588806,
+      "eval_em": 90.2919708029197,
+      "eval_f1": 90.88564476885644,
+      "eval_loss": 0.21875151991844177,
+      "eval_runtime": 36.4713,
+      "eval_samples_per_second": 37.564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.33,
+      "eval_average_metrics": 89.81630170316302,
+      "eval_em": 89.48905109489051,
+      "eval_f1": 90.14355231143553,
+      "eval_loss": 0.22119711339473724,
+      "eval_runtime": 36.7112,
+      "eval_samples_per_second": 37.318,
+      "step": 2800
     },
     {
       "epoch": 0.35,
       "learning_rate": 0.0002650878622134295,
-      "loss": 0.3514,
+      "loss": 0.3308,
       "step": 3000
     },
     {
       "epoch": 0.35,
-      "eval_average_metrics": 88.32290580465764,
-      "eval_em": 87.88321167883211,
-      "eval_f1": 88.76259993048315,
-      "eval_loss": 0.23650217056274414,
-      "eval_runtime": 15.254,
-      "eval_samples_per_second": 89.812,
+      "eval_average_metrics": 89.86253041362531,
+      "eval_em": 89.41605839416059,
+      "eval_f1": 90.30900243309003,
+      "eval_loss": 0.21812815964221954,
+      "eval_runtime": 35.9089,
+      "eval_samples_per_second": 38.152,
       "step": 3000
     },
+    {
+      "epoch": 0.37,
+      "eval_average_metrics": 90.68004866180048,
+      "eval_em": 90.2919708029197,
+      "eval_f1": 91.06812652068126,
+      "eval_loss": 0.21945150196552277,
+      "eval_runtime": 36.7647,
+      "eval_samples_per_second": 37.264,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4,
+      "eval_average_metrics": 88.52676399026764,
+      "eval_em": 88.24817518248175,
+      "eval_f1": 88.80535279805353,
+      "eval_loss": 0.22326913475990295,
+      "eval_runtime": 38.4494,
+      "eval_samples_per_second": 35.631,
+      "step": 3400
+    },
     {
       "epoch": 0.41,
       "learning_rate": 0.00025926917258233445,
-      "loss": 0.3468,
+      "loss": 0.3268,
       "step": 3500
     },
     {
-      "epoch": 0.41,
-      "eval_average_metrics": 89.25669099756689,
-      "eval_em": 88.97810218978101,
-      "eval_f1": 89.53527980535279,
-      "eval_loss": 0.22920837998390198,
-      "eval_runtime": 15.3356,
-      "eval_samples_per_second": 89.335,
-      "step": 3500
+      "epoch": 0.42,
+      "eval_average_metrics": 89.9014598540146,
+      "eval_em": 89.56204379562044,
+      "eval_f1": 90.24087591240875,
+      "eval_loss": 0.2204274833202362,
+      "eval_runtime": 39.4301,
+      "eval_samples_per_second": 34.745,
+      "step": 3600
+    },
+    {
+      "epoch": 0.44,
+      "eval_average_metrics": 88.8491484184915,
+      "eval_em": 88.54014598540147,
+      "eval_f1": 89.15815085158151,
+      "eval_loss": 0.22101299464702606,
+      "eval_runtime": 36.7124,
+      "eval_samples_per_second": 37.317,
+      "step": 3800
     },
     {
       "epoch": 0.47,
       "learning_rate": 0.0002534504829512394,
-      "loss": 0.3428,
+      "loss": 0.3266,
       "step": 4000
     },
     {
       "epoch": 0.47,
-      "eval_average_metrics": 88.33211678832117,
-      "eval_em": 88.02919708029196,
-      "eval_f1": 88.63503649635038,
-      "eval_loss": 0.2317986935377121,
-      "eval_runtime": 15.4226,
-      "eval_samples_per_second": 88.831,
+      "eval_average_metrics": 90.43673965936739,
+      "eval_em": 90.07299270072993,
+      "eval_f1": 90.80048661800487,
+      "eval_loss": 0.2189851850271225,
+      "eval_runtime": 38.8738,
+      "eval_samples_per_second": 35.242,
       "step": 4000
     },
+    {
+      "epoch": 0.49,
+      "eval_average_metrics": 88.7676399026764,
+      "eval_em": 88.39416058394161,
+      "eval_f1": 89.14111922141119,
+      "eval_loss": 0.22029492259025574,
+      "eval_runtime": 37.8194,
+      "eval_samples_per_second": 36.225,
+      "step": 4200
+    },
+    {
+      "epoch": 0.51,
+      "eval_average_metrics": 89.73722627737226,
+      "eval_em": 89.41605839416059,
+      "eval_f1": 90.05839416058394,
+      "eval_loss": 0.21841323375701904,
+      "eval_runtime": 39.6246,
+      "eval_samples_per_second": 34.575,
+      "step": 4400
+    },
     {
       "epoch": 0.52,
       "learning_rate": 0.0002476317933201443,
-      "loss": 0.3485,
+      "loss": 0.3309,
       "step": 4500
     },
     {
-      "epoch": 0.52,
-      "eval_average_metrics": 87.90024330900243,
-      "eval_em": 87.66423357664233,
-      "eval_f1": 88.13625304136254,
-      "eval_loss": 0.23495055735111237,
-      "eval_runtime": 15.4567,
-      "eval_samples_per_second": 88.635,
-      "step": 4500
+      "epoch": 0.54,
+      "eval_average_metrics": 89.49391727493918,
+      "eval_em": 89.27007299270073,
+      "eval_f1": 89.71776155717761,
+      "eval_loss": 0.2204442322254181,
+      "eval_runtime": 39.9817,
+      "eval_samples_per_second": 34.266,
+      "step": 4600
+    },
+    {
+      "epoch": 0.56,
+      "eval_average_metrics": 89.69464720194647,
+      "eval_em": 89.34306569343066,
+      "eval_f1": 90.04622871046227,
+      "eval_loss": 0.2244672179222107,
+      "eval_runtime": 38.4253,
+      "eval_samples_per_second": 35.654,
+      "step": 4800
     },
     {
       "epoch": 0.58,
       "learning_rate": 0.0002418131036890492,
-      "loss": 0.3509,
+      "loss": 0.3285,
       "step": 5000
     },
     {
       "epoch": 0.58,
-      "eval_average_metrics": 88.830900243309,
-      "eval_em": 88.61313868613139,
-      "eval_f1": 89.04866180048661,
-      "eval_loss": 0.2319142073392868,
-      "eval_runtime": 15.5322,
-      "eval_samples_per_second": 88.204,
+      "eval_average_metrics": 89.48783454987836,
+      "eval_em": 89.27007299270073,
+      "eval_f1": 89.70559610705597,
+      "eval_loss": 0.21830883622169495,
+      "eval_runtime": 38.7825,
+      "eval_samples_per_second": 35.325,
       "step": 5000
     },
+    {
+      "epoch": 0.61,
+      "eval_average_metrics": 88.56326034063261,
+      "eval_em": 88.32116788321169,
+      "eval_f1": 88.80535279805352,
+      "eval_loss": 0.22253794968128204,
+      "eval_runtime": 38.0561,
+      "eval_samples_per_second": 36.0,
+      "step": 5200
+    },
+    {
+      "epoch": 0.63,
+      "eval_average_metrics": 89.82846715328466,
+      "eval_em": 89.56204379562044,
+      "eval_f1": 90.09489051094889,
+      "eval_loss": 0.22128665447235107,
+      "eval_runtime": 39.7387,
+      "eval_samples_per_second": 34.475,
+      "step": 5400
+    },
     {
       "epoch": 0.64,
       "learning_rate": 0.0002359944140579541,
-      "loss": 0.3427,
+      "loss": 0.3296,
       "step": 5500
     },
     {
-      "epoch": 0.64,
-      "eval_average_metrics": 86.61678832116789,
-      "eval_em": 86.27737226277372,
-      "eval_f1": 86.95620437956205,
-      "eval_loss": 0.2410917580127716,
-      "eval_runtime": 15.3636,
-      "eval_samples_per_second": 89.172,
-      "step": 5500
+      "epoch": 0.65,
+      "eval_average_metrics": 89.02919708029196,
+      "eval_em": 88.6861313868613,
+      "eval_f1": 89.37226277372264,
+      "eval_loss": 0.2234881967306137,
+      "eval_runtime": 38.4221,
+      "eval_samples_per_second": 35.657,
+      "step": 5600
+    },
+    {
+      "epoch": 0.67,
+      "eval_average_metrics": 89.64355231143553,
+      "eval_em": 89.34306569343066,
+      "eval_f1": 89.9440389294404,
+      "eval_loss": 0.2219015508890152,
+      "eval_runtime": 39.7412,
+      "eval_samples_per_second": 34.473,
+      "step": 5800
     },
     {
       "epoch": 0.7,
       "learning_rate": 0.00023017572442685905,
-      "loss": 0.349,
+      "loss": 0.3377,
       "step": 6000
     },
     {
       "epoch": 0.7,
-      "eval_average_metrics": 88.17570385818561,
-      "eval_em": 87.8102189781022,
-      "eval_f1": 88.54118873826903,
-      "eval_loss": 0.23763984441757202,
-      "eval_runtime": 15.6242,
-      "eval_samples_per_second": 87.684,
+      "eval_average_metrics": 89.13503649635037,
+      "eval_em": 88.9051094890511,
+      "eval_f1": 89.36496350364963,
+      "eval_loss": 0.22565637528896332,
+      "eval_runtime": 38.9351,
+      "eval_samples_per_second": 35.187,
       "step": 6000
     },
+    {
+      "epoch": 0.72,
+      "eval_average_metrics": 89.66788321167883,
+      "eval_em": 89.34306569343066,
+      "eval_f1": 89.99270072992701,
+      "eval_loss": 0.2238757461309433,
+      "eval_runtime": 36.8334,
+      "eval_samples_per_second": 37.195,
+      "step": 6200
+    },
+    {
+      "epoch": 0.74,
+      "eval_average_metrics": 90.63138686131386,
+      "eval_em": 90.2919708029197,
+      "eval_f1": 90.970802919708,
+      "eval_loss": 0.2218635529279709,
+      "eval_runtime": 39.5931,
+      "eval_samples_per_second": 34.602,
+      "step": 6400
+    },
     {
       "epoch": 0.76,
       "learning_rate": 0.00022435703479576396,
-      "loss": 0.3404,
+      "loss": 0.3271,
       "step": 6500
     },
     {
-      "epoch": 0.76,
-      "eval_average_metrics": 88.27128953771289,
-      "eval_em": 88.02919708029196,
-      "eval_f1": 88.51338199513381,
-      "eval_loss": 0.23219799995422363,
-      "eval_runtime": 15.7753,
-      "eval_samples_per_second": 86.845,
-      "step": 6500
+      "epoch": 0.77,
+      "eval_average_metrics": 90.15450121654501,
+      "eval_em": 89.85401459854015,
+      "eval_f1": 90.45498783454988,
+      "eval_loss": 0.2218872755765915,
+      "eval_runtime": 36.9426,
+      "eval_samples_per_second": 37.085,
+      "step": 6600
+    },
+    {
+      "epoch": 0.79,
+      "eval_average_metrics": 90.49513381995133,
+      "eval_em": 90.14598540145985,
+      "eval_f1": 90.84428223844282,
+      "eval_loss": 0.2184651792049408,
+      "eval_runtime": 38.3744,
+      "eval_samples_per_second": 35.701,
+      "step": 6800
     },
     {
       "epoch": 0.81,
       "learning_rate": 0.0002185383451646689,
-      "loss": 0.3369,
+      "loss": 0.3221,
       "step": 7000
     },
     {
       "epoch": 0.81,
-      "eval_average_metrics": 89.59054570733403,
-      "eval_em": 89.12408759124088,
-      "eval_f1": 90.05700382342718,
-      "eval_loss": 0.2249041497707367,
-      "eval_runtime": 15.3646,
-      "eval_samples_per_second": 89.166,
+      "eval_average_metrics": 91.12773722627736,
+      "eval_em": 90.8029197080292,
+      "eval_f1": 91.45255474452554,
+      "eval_loss": 0.2153433859348297,
+      "eval_runtime": 38.0801,
+      "eval_samples_per_second": 35.977,
       "step": 7000
     },
+    {
+      "epoch": 0.84,
+      "eval_average_metrics": 90.32481751824818,
+      "eval_em": 90.0,
+      "eval_f1": 90.64963503649636,
+      "eval_loss": 0.21808864176273346,
+      "eval_runtime": 39.0051,
+      "eval_samples_per_second": 35.124,
+      "step": 7200
+    },
+    {
+      "epoch": 0.86,
+      "eval_average_metrics": 89.08394160583941,
+      "eval_em": 88.75912408759125,
+      "eval_f1": 89.4087591240876,
+      "eval_loss": 0.22276869416236877,
+      "eval_runtime": 36.749,
+      "eval_samples_per_second": 37.28,
+      "step": 7400
+    },
     {
       "epoch": 0.87,
       "learning_rate": 0.00021271965553357382,
-      "loss": 0.3348,
+      "loss": 0.3226,
       "step": 7500
     },
     {
-      "epoch": 0.87,
-      "eval_average_metrics": 89.87104622871047,
-      "eval_em": 89.56204379562044,
-      "eval_f1": 90.1800486618005,
-      "eval_loss": 0.22438949346542358,
-      "eval_runtime": 15.6922,
-      "eval_samples_per_second": 87.305,
-      "step": 7500
+      "epoch": 0.88,
+      "eval_average_metrics": 90.485401459854,
+      "eval_em": 90.21897810218978,
+      "eval_f1": 90.75182481751823,
+      "eval_loss": 0.21675343811511993,
+      "eval_runtime": 37.5554,
+      "eval_samples_per_second": 36.479,
+      "step": 7600
+    },
+    {
+      "epoch": 0.91,
+      "eval_average_metrics": 90.02311435523114,
+      "eval_em": 89.78102189781022,
+      "eval_f1": 90.26520681265205,
+      "eval_loss": 0.21614064276218414,
+      "eval_runtime": 33.4296,
+      "eval_samples_per_second": 40.982,
+      "step": 7800
     },
     {
       "epoch": 0.93,
       "learning_rate": 0.00020690096590247876,
-      "loss": 0.34,
+      "loss": 0.3262,
       "step": 8000
     },
     {
       "epoch": 0.93,
-      "eval_average_metrics": 87.99148418491484,
-      "eval_em": 87.66423357664233,
-      "eval_f1": 88.31873479318735,
-      "eval_loss": 0.23190169036388397,
-      "eval_runtime": 15.0876,
-      "eval_samples_per_second": 90.803,
+      "eval_average_metrics": 90.10583941605839,
+      "eval_em": 89.78102189781022,
+      "eval_f1": 90.43065693430657,
+      "eval_loss": 0.2208666056394577,
+      "eval_runtime": 35.7275,
+      "eval_samples_per_second": 38.346,
       "step": 8000
     },
+    {
+      "epoch": 0.95,
+      "eval_average_metrics": 88.86253041362531,
+      "eval_em": 88.46715328467153,
+      "eval_f1": 89.25790754257908,
+      "eval_loss": 0.22296273708343506,
+      "eval_runtime": 37.4484,
+      "eval_samples_per_second": 36.584,
+      "step": 8200
+    },
+    {
+      "epoch": 0.98,
+      "eval_average_metrics": 88.69708029197079,
+      "eval_em": 88.39416058394161,
+      "eval_f1": 88.99999999999999,
+      "eval_loss": 0.2256641685962677,
+      "eval_runtime": 36.1104,
+      "eval_samples_per_second": 37.939,
+      "step": 8400
+    },
     {
       "epoch": 0.99,
       "learning_rate": 0.00020108227627138367,
-      "loss": 0.3413,
+      "loss": 0.3246,
       "step": 8500
     },
     {
-      "epoch": 0.99,
-      "eval_average_metrics": 89.0498783454988,
-      "eval_em": 88.75912408759125,
-      "eval_f1": 89.34063260340633,
-      "eval_loss": 0.2290135771036148,
-      "eval_runtime": 15.6237,
-      "eval_samples_per_second": 87.688,
-      "step": 8500
+      "epoch": 1.0,
+      "eval_average_metrics": 90.43673965936739,
+      "eval_em": 90.14598540145985,
+      "eval_f1": 90.72749391727494,
+      "eval_loss": 0.2168925255537033,
+      "eval_runtime": 36.8057,
+      "eval_samples_per_second": 37.223,
+      "step": 8600
+    },
+    {
+      "epoch": 1.02,
+      "eval_average_metrics": 90.17883211678833,
+      "eval_em": 89.85401459854015,
+      "eval_f1": 90.5036496350365,
+      "eval_loss": 0.21890240907669067,
+      "eval_runtime": 36.2138,
+      "eval_samples_per_second": 37.831,
+      "step": 8800
     },
     {
       "epoch": 1.05,
       "learning_rate": 0.00019526358664028858,
-      "loss": 0.3207,
+      "loss": 0.3085,
       "step": 9000
     },
     {
       "epoch": 1.05,
-      "eval_average_metrics": 89.54257907542579,
-      "eval_em": 89.34306569343066,
-      "eval_f1": 89.74209245742092,
-      "eval_loss": 0.22638055682182312,
-      "eval_runtime": 15.2056,
-      "eval_samples_per_second": 90.098,
+      "eval_average_metrics": 90.25182481751824,
+      "eval_em": 90.0,
+      "eval_f1": 90.5036496350365,
+      "eval_loss": 0.21791431307792664,
+      "eval_runtime": 36.94,
+      "eval_samples_per_second": 37.087,
       "step": 9000
     },
+    {
+      "epoch": 1.07,
+      "eval_average_metrics": 90.31508515815085,
+      "eval_em": 90.07299270072993,
+      "eval_f1": 90.55717761557177,
+      "eval_loss": 0.2160734087228775,
+      "eval_runtime": 35.6639,
+      "eval_samples_per_second": 38.414,
+      "step": 9200
+    },
+    {
+      "epoch": 1.09,
+      "eval_average_metrics": 89.48783454987836,
+      "eval_em": 89.27007299270073,
+      "eval_f1": 89.70559610705597,
+      "eval_loss": 0.22037485241889954,
+      "eval_runtime": 37.8808,
+      "eval_samples_per_second": 36.166,
+      "step": 9400
+    },
     {
       "epoch": 1.11,
       "learning_rate": 0.00018944489700919352,
-      "loss": 0.333,
+      "loss": 0.3185,
       "step": 9500
     },
     {
-      "epoch": 1.11,
-      "eval_average_metrics": 89.5912408759124,
-      "eval_em": 89.34306569343066,
-      "eval_f1": 89.83941605839416,
-      "eval_loss": 0.22846655547618866,
-      "eval_runtime": 15.7388,
-      "eval_samples_per_second": 87.046,
-      "step": 9500
+      "epoch": 1.12,
+      "eval_average_metrics": 88.35401459854015,
+      "eval_em": 88.02919708029196,
+      "eval_f1": 88.67883211678833,
+      "eval_loss": 0.22213077545166016,
+      "eval_runtime": 36.2657,
+      "eval_samples_per_second": 37.777,
+      "step": 9600
+    },
+    {
+      "epoch": 1.14,
+      "eval_average_metrics": 87.77007299270073,
+      "eval_em": 87.44525547445255,
+      "eval_f1": 88.09489051094891,
+      "eval_loss": 0.22431711852550507,
+      "eval_runtime": 34.0786,
+      "eval_samples_per_second": 40.201,
+      "step": 9800
     },
     {
       "epoch": 1.16,
       "learning_rate": 0.00018362620737809844,
-      "loss": 0.3313,
+      "loss": 0.3149,
       "step": 10000
     },
     {
       "epoch": 1.16,
-      "eval_average_metrics": 88.52485227667711,
-      "eval_em": 88.17518248175182,
-      "eval_f1": 88.87452207160239,
-      "eval_loss": 0.23135380446910858,
-      "eval_runtime": 15.5843,
-      "eval_samples_per_second": 87.909,
+      "eval_average_metrics": 88.28953771289537,
+      "eval_em": 88.02919708029196,
+      "eval_f1": 88.54987834549878,
+      "eval_loss": 0.2239065319299698,
+      "eval_runtime": 37.3216,
+      "eval_samples_per_second": 36.708,
       "step": 10000
     },
+    {
+      "epoch": 1.19,
+      "eval_average_metrics": 90.12043795620437,
+      "eval_em": 89.85401459854015,
+      "eval_f1": 90.3868613138686,
+      "eval_loss": 0.21789653599262238,
+      "eval_runtime": 35.8306,
+      "eval_samples_per_second": 38.235,
+      "step": 10200
+    },
+    {
+      "epoch": 1.21,
+      "eval_average_metrics": 89.71654501216544,
+      "eval_em": 89.41605839416059,
+      "eval_f1": 90.0170316301703,
+      "eval_loss": 0.2222936451435089,
+      "eval_runtime": 35.4154,
+      "eval_samples_per_second": 38.684,
+      "step": 10400
+    },
     {
       "epoch": 1.22,
       "learning_rate": 0.00017780751774700338,
-      "loss": 0.3326,
+      "loss": 0.3167,
       "step": 10500
     },
     {
-      "epoch": 1.22,
-      "eval_average_metrics": 89.61730969760167,
-      "eval_em": 89.12408759124088,
-      "eval_f1": 90.11053180396246,
-      "eval_loss": 0.22540691494941711,
-      "eval_runtime": 15.4318,
-      "eval_samples_per_second": 88.777,
-      "step": 10500
+      "epoch": 1.23,
+      "eval_average_metrics": 89.19586374695864,
+      "eval_em": 88.83211678832117,
+      "eval_f1": 89.5596107055961,
+      "eval_loss": 0.22271069884300232,
+      "eval_runtime": 35.7044,
+      "eval_samples_per_second": 38.371,
+      "step": 10600
+    },
+    {
+      "epoch": 1.26,
+      "eval_average_metrics": 89.97445255474452,
+      "eval_em": 89.7080291970803,
+      "eval_f1": 90.24087591240875,
+      "eval_loss": 0.22066636383533478,
+      "eval_runtime": 34.2308,
+      "eval_samples_per_second": 40.022,
+      "step": 10800
     },
     {
       "epoch": 1.28,
       "learning_rate": 0.0001719888281159083,
-      "loss": 0.3349,
+      "loss": 0.3179,
       "step": 11000
     },
     {
       "epoch": 1.28,
-      "eval_average_metrics": 88.49808828640946,
-      "eval_em": 88.17518248175182,
-      "eval_f1": 88.82099409106708,
-      "eval_loss": 0.2274966984987259,
-      "eval_runtime": 15.7884,
-      "eval_samples_per_second": 86.773,
+      "eval_average_metrics": 90.05717761557177,
+      "eval_em": 89.78102189781022,
+      "eval_f1": 90.33333333333333,
+      "eval_loss": 0.2182897925376892,
+      "eval_runtime": 36.8191,
+      "eval_samples_per_second": 37.209,
       "step": 11000
     },
+    {
+      "epoch": 1.3,
+      "eval_average_metrics": 88.34428223844282,
+      "eval_em": 88.02919708029196,
+      "eval_f1": 88.65936739659367,
+      "eval_loss": 0.22293207049369812,
+      "eval_runtime": 38.8183,
+      "eval_samples_per_second": 35.293,
+      "step": 11200
+    },
+    {
+      "epoch": 1.33,
+      "eval_average_metrics": 90.60705596107056,
+      "eval_em": 90.36496350364963,
+      "eval_f1": 90.84914841849148,
+      "eval_loss": 0.2161208987236023,
+      "eval_runtime": 38.6684,
+      "eval_samples_per_second": 35.429,
+      "step": 11400
+    },
     {
       "epoch": 1.34,
       "learning_rate": 0.00016617013848481323,
-      "loss": 0.3268,
+      "loss": 0.3153,
       "step": 11500
     },
     {
-      "epoch": 1.34,
-      "eval_average_metrics": 90.12043795620437,
-      "eval_em": 89.7080291970803,
-      "eval_f1": 90.53284671532846,
-      "eval_loss": 0.22236795723438263,
-      "eval_runtime": 15.5257,
-      "eval_samples_per_second": 88.241,
-      "step": 11500
+      "epoch": 1.35,
+      "eval_average_metrics": 90.38807785888078,
+      "eval_em": 90.14598540145985,
+      "eval_f1": 90.6301703163017,
+      "eval_loss": 0.2135632187128067,
+      "eval_runtime": 38.983,
+      "eval_samples_per_second": 35.144,
+      "step": 11600
+    },
+    {
+      "epoch": 1.37,
+      "eval_average_metrics": 89.14720194647202,
+      "eval_em": 88.9051094890511,
+      "eval_f1": 89.38929440389293,
+      "eval_loss": 0.22012485563755035,
+      "eval_runtime": 39.657,
+      "eval_samples_per_second": 34.546,
+      "step": 11800
     },
     {
       "epoch": 1.4,
       "learning_rate": 0.00016035144885371815,
-      "loss": 0.3315,
+      "loss": 0.3216,
       "step": 12000
     },
     {
       "epoch": 1.4,
-      "eval_average_metrics": 88.9403892944039,
-      "eval_em": 88.54014598540147,
-      "eval_f1": 89.34063260340633,
-      "eval_loss": 0.22285930812358856,
-      "eval_runtime": 15.103,
-      "eval_samples_per_second": 90.71,
+      "eval_average_metrics": 89.6581508515815,
+      "eval_em": 89.41605839416059,
+      "eval_f1": 89.90024330900242,
+      "eval_loss": 0.21762309968471527,
+      "eval_runtime": 40.4674,
+      "eval_samples_per_second": 33.854,
       "step": 12000
     },
+    {
+      "epoch": 1.42,
+      "eval_average_metrics": 89.30535279805352,
+      "eval_em": 89.05109489051095,
+      "eval_f1": 89.5596107055961,
+      "eval_loss": 0.21773971617221832,
+      "eval_runtime": 38.4174,
+      "eval_samples_per_second": 35.661,
+      "step": 12200
+    },
+    {
+      "epoch": 1.44,
+      "eval_average_metrics": 90.169099756691,
+      "eval_em": 89.92700729927007,
+      "eval_f1": 90.41119221411194,
+      "eval_loss": 0.21440057456493378,
+      "eval_runtime": 37.9225,
+      "eval_samples_per_second": 36.126,
+      "step": 12400
+    },
     {
       "epoch": 1.45,
       "learning_rate": 0.00015453275922262303,
-      "loss": 0.3262,
+      "loss": 0.3151,
       "step": 12500
     },
     {
-      "epoch": 1.45,
-      "eval_average_metrics": 86.66718804310045,
-      "eval_em": 86.27737226277372,
-      "eval_f1": 87.0570038234272,
-      "eval_loss": 0.23210552334785461,
-      "eval_runtime": 15.5946,
-      "eval_samples_per_second": 87.851,
-      "step": 12500
+      "epoch": 1.47,
+      "eval_average_metrics": 89.58515815085158,
+      "eval_em": 89.34306569343066,
+      "eval_f1": 89.82725060827251,
+      "eval_loss": 0.21768692135810852,
+      "eval_runtime": 36.5526,
+      "eval_samples_per_second": 37.48,
+      "step": 12600
+    },
+    {
+      "epoch": 1.49,
+      "eval_average_metrics": 89.99878345498783,
+      "eval_em": 89.78102189781022,
+      "eval_f1": 90.21654501216545,
+      "eval_loss": 0.2161240428686142,
+      "eval_runtime": 39.5978,
+      "eval_samples_per_second": 34.598,
+      "step": 12800
     },
     {
       "epoch": 1.51,
       "learning_rate": 0.00014871406959152797,
-      "loss": 0.3389,
+      "loss": 0.3255,
       "step": 13000
     },
     {
       "epoch": 1.51,
-      "eval_average_metrics": 89.97445255474452,
-      "eval_em": 89.7080291970803,
-      "eval_f1": 90.24087591240875,
-      "eval_loss": 0.22476908564567566,
-      "eval_runtime": 15.3804,
-      "eval_samples_per_second": 89.075,
+      "eval_average_metrics": 89.70681265206812,
+      "eval_em": 89.48905109489051,
+      "eval_f1": 89.92457420924575,
+      "eval_loss": 0.21710088849067688,
+      "eval_runtime": 38.5743,
+      "eval_samples_per_second": 35.516,
       "step": 13000
     },
+    {
+      "epoch": 1.54,
+      "eval_average_metrics": 89.3515815085158,
+      "eval_em": 89.05109489051095,
+      "eval_f1": 89.65206812652067,
+      "eval_loss": 0.21740640699863434,
+      "eval_runtime": 38.7743,
+      "eval_samples_per_second": 35.333,
+      "step": 13200
+    },
+    {
+      "epoch": 1.56,
+      "eval_average_metrics": 90.29075425790754,
+      "eval_em": 90.07299270072993,
+      "eval_f1": 90.50851581508516,
+      "eval_loss": 0.21489615738391876,
+      "eval_runtime": 41.332,
+      "eval_samples_per_second": 33.146,
+      "step": 13400
+    },
     {
       "epoch": 1.57,
       "learning_rate": 0.00014289537996043292,
-      "loss": 0.324,
+      "loss": 0.3129,
       "step": 13500
     },
     {
-      "epoch": 1.57,
-      "eval_average_metrics": 88.98470629127564,
-      "eval_em": 88.54014598540147,
-      "eval_f1": 89.42926659714982,
-      "eval_loss": 0.22203609347343445,
-      "eval_runtime": 15.4489,
-      "eval_samples_per_second": 88.679,
-      "step": 13500
+      "epoch": 1.58,
+      "eval_average_metrics": 89.81386861313868,
+      "eval_em": 89.48905109489051,
+      "eval_f1": 90.13868613138686,
+      "eval_loss": 0.21436673402786255,
+      "eval_runtime": 39.3785,
+      "eval_samples_per_second": 34.791,
+      "step": 13600
+    },
+    {
+      "epoch": 1.61,
+      "eval_average_metrics": 89.77980535279806,
+      "eval_em": 89.48905109489051,
+      "eval_f1": 90.0705596107056,
+      "eval_loss": 0.21632547676563263,
+      "eval_runtime": 36.2514,
+      "eval_samples_per_second": 37.792,
+      "step": 13800
     },
     {
       "epoch": 1.63,
       "learning_rate": 0.00013707669032933783,
-      "loss": 0.327,
+      "loss": 0.317,
       "step": 14000
     },
     {
       "epoch": 1.63,
-      "eval_average_metrics": 89.79370872436566,
-      "eval_em": 89.41605839416059,
-      "eval_f1": 90.17135905457073,
-      "eval_loss": 0.22265847027301788,
-      "eval_runtime": 15.2397,
-      "eval_samples_per_second": 89.897,
+      "eval_average_metrics": 90.14476885644768,
+      "eval_em": 89.85401459854015,
+      "eval_f1": 90.43552311435523,
+      "eval_loss": 0.21579378843307495,
+      "eval_runtime": 38.9315,
+      "eval_samples_per_second": 35.19,
       "step": 14000
     },
+    {
+      "epoch": 1.65,
+      "eval_average_metrics": 89.69221411192214,
+      "eval_em": 89.34306569343066,
+      "eval_f1": 90.04136253041362,
+      "eval_loss": 0.21815675497055054,
+      "eval_runtime": 38.141,
+      "eval_samples_per_second": 35.919,
+      "step": 14200
+    },
+    {
+      "epoch": 1.68,
+      "eval_average_metrics": 90.26642335766422,
+      "eval_em": 90.0,
+      "eval_f1": 90.53284671532845,
+      "eval_loss": 0.21684660017490387,
+      "eval_runtime": 36.4617,
+      "eval_samples_per_second": 37.574,
+      "step": 14400
+    },
     {
       "epoch": 1.69,
       "learning_rate": 0.00013125800069824274,
-      "loss": 0.3315,
+      "loss": 0.3204,
       "step": 14500
     },
     {
-      "epoch": 1.69,
-      "eval_average_metrics": 89.11678832116787,
-      "eval_em": 88.83211678832117,
-      "eval_f1": 89.40145985401458,
-      "eval_loss": 0.22402139008045197,
-      "eval_runtime": 15.9941,
-      "eval_samples_per_second": 85.657,
-      "step": 14500
+      "epoch": 1.7,
+      "eval_average_metrics": 90.54379562043796,
+      "eval_em": 90.21897810218978,
+      "eval_f1": 90.86861313868614,
+      "eval_loss": 0.21547624468803406,
+      "eval_runtime": 39.6007,
+      "eval_samples_per_second": 34.595,
+      "step": 14600
+    },
+    {
+      "epoch": 1.72,
+      "eval_average_metrics": 90.76277372262774,
+      "eval_em": 90.43795620437956,
+      "eval_f1": 91.08759124087592,
+      "eval_loss": 0.2154303342103958,
+      "eval_runtime": 37.9795,
+      "eval_samples_per_second": 36.072,
+      "step": 14800
     },
     {
       "epoch": 1.75,
       "learning_rate": 0.00012543931106714766,
-      "loss": 0.3236,
+      "loss": 0.3153,
       "step": 15000
     },
     {
       "epoch": 1.75,
-      "eval_average_metrics": 89.00729927007299,
-      "eval_em": 88.6861313868613,
-      "eval_f1": 89.32846715328466,
-      "eval_loss": 0.22223517298698425,
-      "eval_runtime": 15.5255,
-      "eval_samples_per_second": 88.242,
+      "eval_average_metrics": 90.34914841849148,
+      "eval_em": 90.0,
+      "eval_f1": 90.69829683698296,
+      "eval_loss": 0.21602213382720947,
+      "eval_runtime": 40.2922,
+      "eval_samples_per_second": 34.002,
       "step": 15000
     },
+    {
+      "epoch": 1.77,
+      "eval_average_metrics": 89.31751824817519,
+      "eval_em": 89.05109489051095,
+      "eval_f1": 89.58394160583941,
+      "eval_loss": 0.2179516851902008,
+      "eval_runtime": 36.6124,
+      "eval_samples_per_second": 37.419,
+      "step": 15200
+    },
+    {
+      "epoch": 1.79,
+      "eval_average_metrics": 89.24452554744525,
+      "eval_em": 88.97810218978101,
+      "eval_f1": 89.51094890510949,
+      "eval_loss": 0.2189921736717224,
+      "eval_runtime": 37.6709,
+      "eval_samples_per_second": 36.368,
+      "step": 15400
+    },
     {
       "epoch": 1.8,
       "learning_rate": 0.00011962062143605258,
-      "loss": 0.3214,
+      "loss": 0.3104,
       "step": 15500
     },
     {
-      "epoch": 1.8,
-      "eval_average_metrics": 88.32603406326032,
-      "eval_em": 88.02919708029196,
-      "eval_f1": 88.6228710462287,
-      "eval_loss": 0.22402779757976532,
-      "eval_runtime": 15.9032,
-      "eval_samples_per_second": 86.146,
-      "step": 15500
+      "epoch": 1.82,
+      "eval_average_metrics": 89.02554744525548,
+      "eval_em": 88.75912408759125,
+      "eval_f1": 89.29197080291972,
+      "eval_loss": 0.21728622913360596,
+      "eval_runtime": 37.6931,
+      "eval_samples_per_second": 36.346,
+      "step": 15600
+    },
+    {
+      "epoch": 1.84,
+      "eval_average_metrics": 90.48540145985402,
+      "eval_em": 90.21897810218978,
+      "eval_f1": 90.75182481751824,
+      "eval_loss": 0.21391963958740234,
+      "eval_runtime": 37.4604,
+      "eval_samples_per_second": 36.572,
+      "step": 15800
     },
     {
       "epoch": 1.86,
       "learning_rate": 0.00011380193180495751,
-      "loss": 0.3254,
+      "loss": 0.3221,
       "step": 16000
     },
     {
       "epoch": 1.86,
-      "eval_average_metrics": 89.72506082725062,
-      "eval_em": 89.48905109489051,
-      "eval_f1": 89.96107055961072,
-      "eval_loss": 0.21747739613056183,
-      "eval_runtime": 15.403,
-      "eval_samples_per_second": 88.944,
+      "eval_average_metrics": 90.92335766423358,
+      "eval_em": 90.65693430656935,
+      "eval_f1": 91.18978102189779,
+      "eval_loss": 0.21324890851974487,
+      "eval_runtime": 36.0923,
+      "eval_samples_per_second": 37.958,
       "step": 16000
     },
+    {
+      "epoch": 1.89,
+      "eval_average_metrics": 90.36374695863748,
+      "eval_em": 90.07299270072993,
+      "eval_f1": 90.65450121654501,
+      "eval_loss": 0.21571427583694458,
+      "eval_runtime": 36.3418,
+      "eval_samples_per_second": 37.698,
+      "step": 16200
+    },
+    {
+      "epoch": 1.91,
+      "eval_average_metrics": 90.58272506082724,
+      "eval_em": 90.2919708029197,
+      "eval_f1": 90.8734793187348,
+      "eval_loss": 0.21374623477458954,
+      "eval_runtime": 35.332,
+      "eval_samples_per_second": 38.775,
+      "step": 16400
+    },
     {
       "epoch": 1.92,
       "learning_rate": 0.00010798324217386244,
-      "loss": 0.3216,
+      "loss": 0.3142,
       "step": 16500
     },
     {
-      "epoch": 1.92,
-      "eval_average_metrics": 90.24209245742091,
-      "eval_em": 90.0,
-      "eval_f1": 90.48418491484183,
-      "eval_loss": 0.21551567316055298,
-      "eval_runtime": 15.0664,
-      "eval_samples_per_second": 90.931,
-      "step": 16500
+      "epoch": 1.93,
+      "eval_average_metrics": 90.41240875912409,
+      "eval_em": 90.14598540145985,
+      "eval_f1": 90.67883211678833,
+      "eval_loss": 0.21370555460453033,
+      "eval_runtime": 37.2326,
+      "eval_samples_per_second": 36.796,
+      "step": 16600
+    },
+    {
+      "epoch": 1.96,
+      "eval_average_metrics": 89.75547445255475,
+      "eval_em": 89.48905109489051,
+      "eval_f1": 90.02189781021897,
+      "eval_loss": 0.21746164560317993,
+      "eval_runtime": 35.4995,
+      "eval_samples_per_second": 38.592,
+      "step": 16800
     },
     {
       "epoch": 1.98,
       "learning_rate": 0.00010216455254276737,
-      "loss": 0.327,
+      "loss": 0.3206,
       "step": 17000
     },
     {
       "epoch": 1.98,
-      "eval_average_metrics": 89.62773722627736,
-      "eval_em": 89.34306569343066,
-      "eval_f1": 89.91240875912406,
-      "eval_loss": 0.22303573787212372,
-      "eval_runtime": 15.6955,
-      "eval_samples_per_second": 87.286,
+      "eval_average_metrics": 89.82846715328466,
+      "eval_em": 89.56204379562044,
+      "eval_f1": 90.0948905109489,
+      "eval_loss": 0.2167847603559494,
+      "eval_runtime": 37.0794,
+      "eval_samples_per_second": 36.948,
       "step": 17000
     },
+    {
+      "epoch": 2.0,
+      "eval_average_metrics": 90.26642335766422,
+      "eval_em": 90.0,
+      "eval_f1": 90.53284671532846,
+      "eval_loss": 0.21559016406536102,
+      "eval_runtime": 37.9031,
+      "eval_samples_per_second": 36.145,
+      "step": 17200
+    },
+    {
+      "epoch": 2.02,
+      "eval_average_metrics": 89.97445255474452,
+      "eval_em": 89.7080291970803,
+      "eval_f1": 90.24087591240874,
+      "eval_loss": 0.21499326825141907,
+      "eval_runtime": 33.4521,
+      "eval_samples_per_second": 40.954,
+      "step": 17400
+    },
     {
       "epoch": 2.04,
       "learning_rate": 9.634586291167229e-05,
-      "loss": 0.3216,
+      "loss": 0.3109,
       "step": 17500
     },
     {
-      "epoch": 2.04,
-      "eval_average_metrics": 89.6581508515815,
-      "eval_em": 89.34306569343066,
-      "eval_f1": 89.97323600973235,
-      "eval_loss": 0.21931786835193634,
-      "eval_runtime": 16.0777,
-      "eval_samples_per_second": 85.211,
-      "step": 17500
+      "epoch": 2.05,
+      "eval_average_metrics": 90.14476885644768,
+      "eval_em": 89.85401459854015,
+      "eval_f1": 90.43552311435523,
+      "eval_loss": 0.21627335250377655,
+      "eval_runtime": 35.646,
+      "eval_samples_per_second": 38.434,
+      "step": 17600
+    },
+    {
+      "epoch": 2.07,
+      "eval_average_metrics": 90.48540145985402,
+      "eval_em": 90.14598540145985,
+      "eval_f1": 90.8248175182482,
+      "eval_loss": 0.21596278250217438,
+      "eval_runtime": 36.6834,
+      "eval_samples_per_second": 37.347,
+      "step": 17800
     },
     {
       "epoch": 2.09,
       "learning_rate": 9.052717328057722e-05,
-      "loss": 0.3166,
+      "loss": 0.3035,
       "step": 18000
     },
     {
       "epoch": 2.09,
-      "eval_average_metrics": 88.78397636426834,
-      "eval_em": 88.39416058394161,
-      "eval_f1": 89.17379214459508,
-      "eval_loss": 0.2221771478652954,
-      "eval_runtime": 15.5735,
-      "eval_samples_per_second": 87.97,
+      "eval_average_metrics": 89.97445255474453,
+      "eval_em": 89.7080291970803,
+      "eval_f1": 90.24087591240877,
+      "eval_loss": 0.21806372702121735,
+      "eval_runtime": 34.0399,
+      "eval_samples_per_second": 40.247,
       "step": 18000
     },
+    {
+      "epoch": 2.12,
+      "eval_average_metrics": 89.58515815085158,
+      "eval_em": 89.34306569343066,
+      "eval_f1": 89.82725060827251,
+      "eval_loss": 0.2175503671169281,
+      "eval_runtime": 35.3257,
+      "eval_samples_per_second": 38.782,
+      "step": 18200
+    },
+    {
+      "epoch": 2.14,
+      "eval_average_metrics": 90.29075425790754,
+      "eval_em": 90.0,
+      "eval_f1": 90.58150851581509,
+      "eval_loss": 0.21528899669647217,
+      "eval_runtime": 37.4995,
+      "eval_samples_per_second": 36.534,
+      "step": 18400
+    },
     {
       "epoch": 2.15,
       "learning_rate": 8.470848364948212e-05,
-      "loss": 0.321,
+      "loss": 0.3102,
       "step": 18500
     },
     {
-      "epoch": 2.15,
-      "eval_average_metrics": 90.1508515815085,
-      "eval_em": 89.92700729927007,
-      "eval_f1": 90.37469586374695,
-      "eval_loss": 0.2213568538427353,
-      "eval_runtime": 15.5146,
-      "eval_samples_per_second": 88.304,
-      "step": 18500
+      "epoch": 2.16,
+      "eval_average_metrics": 89.9501216545012,
+      "eval_em": 89.7080291970803,
+      "eval_f1": 90.19221411192213,
+      "eval_loss": 0.21668794751167297,
+      "eval_runtime": 36.8393,
+      "eval_samples_per_second": 37.189,
+      "step": 18600
+    },
+    {
+      "epoch": 2.19,
+      "eval_average_metrics": 90.04744525547446,
+      "eval_em": 89.78102189781022,
+      "eval_f1": 90.31386861313868,
+      "eval_loss": 0.2147464007139206,
+      "eval_runtime": 36.4341,
+      "eval_samples_per_second": 37.602,
+      "step": 18800
     },
     {
       "epoch": 2.21,
       "learning_rate": 7.888979401838705e-05,
-      "loss": 0.3099,
+      "loss": 0.2992,
       "step": 19000
     },
     {
       "epoch": 2.21,
-      "eval_average_metrics": 90.41240875912409,
-      "eval_em": 90.14598540145985,
-      "eval_f1": 90.67883211678833,
-      "eval_loss": 0.21670033037662506,
-      "eval_runtime": 15.2815,
-      "eval_samples_per_second": 89.651,
+      "eval_average_metrics": 89.56082725060827,
+      "eval_em": 89.34306569343066,
+      "eval_f1": 89.77858880778588,
+      "eval_loss": 0.21566100418567657,
+      "eval_runtime": 37.3837,
+      "eval_samples_per_second": 36.647,
       "step": 19000
     },
+    {
+      "epoch": 2.23,
+      "eval_average_metrics": 89.26885644768856,
+      "eval_em": 89.05109489051095,
+      "eval_f1": 89.48661800486619,
+      "eval_loss": 0.21698147058486938,
+      "eval_runtime": 34.5662,
+      "eval_samples_per_second": 39.634,
+      "step": 19200
+    },
+    {
+      "epoch": 2.26,
+      "eval_average_metrics": 89.9501216545012,
+      "eval_em": 89.7080291970803,
+      "eval_f1": 90.19221411192213,
+      "eval_loss": 0.21671971678733826,
+      "eval_runtime": 39.0865,
+      "eval_samples_per_second": 35.05,
+      "step": 19400
+    },
     {
       "epoch": 2.27,
       "learning_rate": 7.307110438729197e-05,
-      "loss": 0.3178,
+      "loss": 0.3104,
       "step": 19500
     },
     {
-      "epoch": 2.27,
-      "eval_average_metrics": 89.41484184914842,
-      "eval_em": 89.19708029197079,
-      "eval_f1": 89.63260340632604,
-      "eval_loss": 0.21756400167942047,
-      "eval_runtime": 15.5668,
-      "eval_samples_per_second": 88.008,
-      "step": 19500
+      "epoch": 2.28,
+      "eval_average_metrics": 88.92822384428223,
+      "eval_em": 88.6861313868613,
+      "eval_f1": 89.17031630170315,
+      "eval_loss": 0.21831820905208588,
+      "eval_runtime": 38.0065,
+      "eval_samples_per_second": 36.046,
+      "step": 19600
+    },
+    {
+      "epoch": 2.3,
+      "eval_average_metrics": 89.82846715328466,
+      "eval_em": 89.56204379562044,
+      "eval_f1": 90.09489051094889,
+      "eval_loss": 0.2175111472606659,
+      "eval_runtime": 38.7133,
+      "eval_samples_per_second": 35.388,
+      "step": 19800
     },
     {
       "epoch": 2.33,
       "learning_rate": 6.72524147561969e-05,
-      "loss": 0.3082,
+      "loss": 0.299,
       "step": 20000
     },
     {
       "epoch": 2.33,
-      "eval_average_metrics": 90.38807785888078,
-      "eval_em": 90.14598540145985,
-      "eval_f1": 90.6301703163017,
-      "eval_loss": 0.213778555393219,
-      "eval_runtime": 15.3113,
-      "eval_samples_per_second": 89.476,
+      "eval_average_metrics": 90.04744525547446,
+      "eval_em": 89.78102189781022,
+      "eval_f1": 90.31386861313868,
+      "eval_loss": 0.21775414049625397,
+      "eval_runtime": 37.1614,
+      "eval_samples_per_second": 36.866,
       "step": 20000
     },
+    {
+      "epoch": 2.35,
+      "eval_average_metrics": 90.33941605839416,
+      "eval_em": 90.07299270072993,
+      "eval_f1": 90.60583941605839,
+      "eval_loss": 0.21791471540927887,
+      "eval_runtime": 38.7185,
+      "eval_samples_per_second": 35.384,
+      "step": 20200
+    },
+    {
+      "epoch": 2.37,
+      "eval_average_metrics": 90.85036496350364,
+      "eval_em": 90.58394160583941,
+      "eval_f1": 91.11678832116786,
+      "eval_loss": 0.21669700741767883,
+      "eval_runtime": 36.9688,
+      "eval_samples_per_second": 37.058,
+      "step": 20400
+    },
     {
       "epoch": 2.39,
       "learning_rate": 6.143372512510183e-05,
-      "loss": 0.3185,
+      "loss": 0.3106,
       "step": 20500
     },
     {
-      "epoch": 2.39,
-      "eval_average_metrics": 89.44091067083768,
-      "eval_em": 89.05109489051095,
-      "eval_f1": 89.83072645116441,
-      "eval_loss": 0.21920597553253174,
-      "eval_runtime": 15.0292,
-      "eval_samples_per_second": 91.156,
-      "step": 20500
+      "epoch": 2.4,
+      "eval_average_metrics": 90.99635036496349,
+      "eval_em": 90.72992700729927,
+      "eval_f1": 91.26277372262773,
+      "eval_loss": 0.21628382802009583,
+      "eval_runtime": 36.1926,
+      "eval_samples_per_second": 37.853,
+      "step": 20600
+    },
+    {
+      "epoch": 2.42,
+      "eval_average_metrics": 90.485401459854,
+      "eval_em": 90.21897810218978,
+      "eval_f1": 90.75182481751823,
+      "eval_loss": 0.2179671972990036,
+      "eval_runtime": 35.7701,
+      "eval_samples_per_second": 38.3,
+      "step": 20800
     },
     {
       "epoch": 2.44,
       "learning_rate": 5.561503549400675e-05,
-      "loss": 0.3214,
+      "loss": 0.3131,
       "step": 21000
     },
     {
       "epoch": 2.44,
-      "eval_average_metrics": 89.58515815085158,
-      "eval_em": 89.27007299270073,
-      "eval_f1": 89.90024330900242,
-      "eval_loss": 0.21777711808681488,
-      "eval_runtime": 15.5974,
-      "eval_samples_per_second": 87.835,
+      "eval_average_metrics": 89.90145985401459,
+      "eval_em": 89.63503649635037,
+      "eval_f1": 90.16788321167881,
+      "eval_loss": 0.21815823018550873,
+      "eval_runtime": 38.6202,
+      "eval_samples_per_second": 35.474,
       "step": 21000
     },
+    {
+      "epoch": 2.47,
+      "eval_average_metrics": 89.81386861313868,
+      "eval_em": 89.48905109489051,
+      "eval_f1": 90.13868613138686,
+      "eval_loss": 0.21742023527622223,
+      "eval_runtime": 38.6698,
+      "eval_samples_per_second": 35.428,
+      "step": 21200
+    },
+    {
+      "epoch": 2.49,
+      "eval_average_metrics": 90.54379562043796,
+      "eval_em": 90.21897810218978,
+      "eval_f1": 90.86861313868614,
+      "eval_loss": 0.21656812727451324,
+      "eval_runtime": 37.7889,
+      "eval_samples_per_second": 36.254,
+      "step": 21400
+    },
     {
       "epoch": 2.5,
       "learning_rate": 4.9796345862911663e-05,
-      "loss": 0.3204,
+      "loss": 0.3109,
       "step": 21500
     },
     {
-      "epoch": 2.5,
-      "eval_average_metrics": 90.63138686131387,
-      "eval_em": 90.36496350364963,
-      "eval_f1": 90.89781021897811,
-      "eval_loss": 0.2141435146331787,
-      "eval_runtime": 15.7201,
-      "eval_samples_per_second": 87.149,
-      "step": 21500
+      "epoch": 2.51,
+      "eval_average_metrics": 90.47080291970804,
+      "eval_em": 90.14598540145985,
+      "eval_f1": 90.79562043795622,
+      "eval_loss": 0.21554183959960938,
+      "eval_runtime": 40.3847,
+      "eval_samples_per_second": 33.924,
+      "step": 21600
+    },
+    {
+      "epoch": 2.54,
+      "eval_average_metrics": 90.55839416058393,
+      "eval_em": 90.2919708029197,
+      "eval_f1": 90.82481751824815,
+      "eval_loss": 0.21551626920700073,
+      "eval_runtime": 40.9505,
+      "eval_samples_per_second": 33.455,
+      "step": 21800
     },
     {
       "epoch": 2.56,
       "learning_rate": 4.397765623181659e-05,
-      "loss": 0.3219,
+      "loss": 0.3139,
       "step": 22000
     },
     {
       "epoch": 2.56,
-      "eval_average_metrics": 89.92579075425792,
-      "eval_em": 89.56204379562044,
-      "eval_f1": 90.28953771289538,
-      "eval_loss": 0.21593248844146729,
-      "eval_runtime": 15.5402,
-      "eval_samples_per_second": 88.158,
+      "eval_average_metrics": 90.53406326034063,
+      "eval_em": 90.21897810218978,
+      "eval_f1": 90.84914841849148,
+      "eval_loss": 0.21652992069721222,
+      "eval_runtime": 39.2455,
+      "eval_samples_per_second": 34.908,
       "step": 22000
     },
+    {
+      "epoch": 2.58,
+      "eval_average_metrics": 90.38807785888076,
+      "eval_em": 90.07299270072993,
+      "eval_f1": 90.70316301703161,
+      "eval_loss": 0.2176700383424759,
+      "eval_runtime": 39.6231,
+      "eval_samples_per_second": 34.576,
+      "step": 22200
+    },
+    {
+      "epoch": 2.61,
+      "eval_average_metrics": 90.82603406326034,
+      "eval_em": 90.51094890510949,
+      "eval_f1": 91.14111922141119,
+      "eval_loss": 0.21595974266529083,
+      "eval_runtime": 38.5872,
+      "eval_samples_per_second": 35.504,
+      "step": 22400
+    },
     {
       "epoch": 2.62,
       "learning_rate": 3.815896660072152e-05,
-      "loss": 0.3082,
+      "loss": 0.3012,
       "step": 22500
     },
     {
-      "epoch": 2.62,
-      "eval_average_metrics": 90.19343065693431,
-      "eval_em": 89.92700729927007,
-      "eval_f1": 90.45985401459853,
-      "eval_loss": 0.2151171863079071,
-      "eval_runtime": 15.4013,
-      "eval_samples_per_second": 88.953,
-      "step": 22500
+      "epoch": 2.63,
+      "eval_average_metrics": 90.60705596107056,
+      "eval_em": 90.2919708029197,
+      "eval_f1": 90.9221411192214,
+      "eval_loss": 0.2160317748785019,
+      "eval_runtime": 38.122,
+      "eval_samples_per_second": 35.937,
+      "step": 22600
+    },
+    {
+      "epoch": 2.65,
+      "eval_average_metrics": 90.65571776155718,
+      "eval_em": 90.36496350364963,
+      "eval_f1": 90.94647201946472,
+      "eval_loss": 0.21644768118858337,
+      "eval_runtime": 34.8398,
+      "eval_samples_per_second": 39.323,
+      "step": 22800
     },
     {
       "epoch": 2.68,
       "learning_rate": 3.234027696962644e-05,
-      "loss": 0.3139,
+      "loss": 0.3084,
       "step": 23000
     },
     {
       "epoch": 2.68,
-      "eval_average_metrics": 90.58272506082724,
-      "eval_em": 90.2919708029197,
-      "eval_f1": 90.8734793187348,
-      "eval_loss": 0.21431832015514374,
-      "eval_runtime": 15.4496,
-      "eval_samples_per_second": 88.675,
+      "eval_average_metrics": 90.169099756691,
+      "eval_em": 89.85401459854015,
+      "eval_f1": 90.48418491484183,
+      "eval_loss": 0.21626520156860352,
+      "eval_runtime": 34.9713,
+      "eval_samples_per_second": 39.175,
       "step": 23000
     },
+    {
+      "epoch": 2.7,
+      "eval_average_metrics": 89.58515815085158,
+      "eval_em": 89.27007299270073,
+      "eval_f1": 89.90024330900242,
+      "eval_loss": 0.2173507809638977,
+      "eval_runtime": 34.5287,
+      "eval_samples_per_second": 39.677,
+      "step": 23200
+    },
+    {
+      "epoch": 2.72,
+      "eval_average_metrics": 89.58515815085158,
+      "eval_em": 89.27007299270073,
+      "eval_f1": 89.90024330900242,
+      "eval_loss": 0.21798554062843323,
+      "eval_runtime": 32.6335,
+      "eval_samples_per_second": 41.981,
+      "step": 23400
+    },
     {
       "epoch": 2.73,
       "learning_rate": 2.6521587338531362e-05,
-      "loss": 0.3111,
+      "loss": 0.3034,
       "step": 23500
     },
     {
-      "epoch": 2.73,
-      "eval_average_metrics": 89.7433090024331,
-      "eval_em": 89.41605839416059,
-      "eval_f1": 90.0705596107056,
-      "eval_loss": 0.21385075151920319,
-      "eval_runtime": 15.7553,
-      "eval_samples_per_second": 86.955,
-      "step": 23500
+      "epoch": 2.75,
+      "eval_average_metrics": 89.6581508515815,
+      "eval_em": 89.34306569343066,
+      "eval_f1": 89.97323600973235,
+      "eval_loss": 0.2169240415096283,
+      "eval_runtime": 31.2981,
+      "eval_samples_per_second": 43.773,
+      "step": 23600
+    },
+    {
+      "epoch": 2.77,
+      "eval_average_metrics": 90.02311435523114,
+      "eval_em": 89.7080291970803,
+      "eval_f1": 90.338199513382,
+      "eval_loss": 0.2169690728187561,
+      "eval_runtime": 31.3076,
+      "eval_samples_per_second": 43.759,
+      "step": 23800
     },
     {
       "epoch": 2.79,
       "learning_rate": 2.0702897707436283e-05,
-      "loss": 0.3168,
+      "loss": 0.3087,
       "step": 24000
     },
     {
       "epoch": 2.79,
-      "eval_average_metrics": 89.81630170316302,
-      "eval_em": 89.56204379562044,
-      "eval_f1": 90.0705596107056,
-      "eval_loss": 0.2146376520395279,
-      "eval_runtime": 15.4091,
-      "eval_samples_per_second": 88.908,
+      "eval_average_metrics": 90.53406326034063,
+      "eval_em": 90.21897810218978,
+      "eval_f1": 90.84914841849148,
+      "eval_loss": 0.2168567031621933,
+      "eval_runtime": 32.1016,
+      "eval_samples_per_second": 42.677,
       "step": 24000
     },
+    {
+      "epoch": 2.82,
+      "eval_average_metrics": 90.169099756691,
+      "eval_em": 89.85401459854015,
+      "eval_f1": 90.48418491484183,
+      "eval_loss": 0.21707168221473694,
+      "eval_runtime": 34.5668,
+      "eval_samples_per_second": 39.633,
+      "step": 24200
+    },
+    {
+      "epoch": 2.84,
+      "eval_average_metrics": 90.169099756691,
+      "eval_em": 89.85401459854015,
+      "eval_f1": 90.48418491484183,
+      "eval_loss": 0.2168397605419159,
+      "eval_runtime": 35.1948,
+      "eval_samples_per_second": 38.926,
+      "step": 24400
+    },
     {
       "epoch": 2.85,
       "learning_rate": 1.4884208076341207e-05,
-      "loss": 0.3115,
+      "loss": 0.3038,
       "step": 24500
     },
     {
-      "epoch": 2.85,
-      "eval_average_metrics": 89.26885644768856,
-      "eval_em": 88.97810218978101,
-      "eval_f1": 89.5596107055961,
-      "eval_loss": 0.21739177405834198,
-      "eval_runtime": 15.7449,
-      "eval_samples_per_second": 87.012,
-      "step": 24500
+      "epoch": 2.86,
+      "eval_average_metrics": 89.14720194647202,
+      "eval_em": 88.83211678832117,
+      "eval_f1": 89.46228710462286,
+      "eval_loss": 0.21729852259159088,
+      "eval_runtime": 35.4024,
+      "eval_samples_per_second": 38.698,
+      "step": 24600
+    },
+    {
+      "epoch": 2.89,
+      "eval_average_metrics": 89.99878345498783,
+      "eval_em": 89.7080291970803,
+      "eval_f1": 90.28953771289538,
+      "eval_loss": 0.21667222678661346,
+      "eval_runtime": 34.8507,
+      "eval_samples_per_second": 39.311,
+      "step": 24800
     },
     {
       "epoch": 2.91,
       "learning_rate": 9.065518445246129e-06,
-      "loss": 0.3144,
+      "loss": 0.3083,
       "step": 25000
     },
     {
       "epoch": 2.91,
-      "eval_average_metrics": 90.08394160583941,
-      "eval_em": 89.78102189781022,
-      "eval_f1": 90.38686131386861,
-      "eval_loss": 0.21506215631961823,
-      "eval_runtime": 15.5375,
-      "eval_samples_per_second": 88.174,
+      "eval_average_metrics": 90.21776155717762,
+      "eval_em": 89.92700729927007,
+      "eval_f1": 90.50851581508516,
+      "eval_loss": 0.216518834233284,
+      "eval_runtime": 34.7968,
+      "eval_samples_per_second": 39.371,
       "step": 25000
     },
+    {
+      "epoch": 2.93,
+      "eval_average_metrics": 90.07177615571777,
+      "eval_em": 89.78102189781022,
+      "eval_f1": 90.36253041362531,
+      "eval_loss": 0.21596592664718628,
+      "eval_runtime": 34.8982,
+      "eval_samples_per_second": 39.257,
+      "step": 25200
+    },
+    {
+      "epoch": 2.96,
+      "eval_average_metrics": 90.07177615571777,
+      "eval_em": 89.78102189781022,
+      "eval_f1": 90.36253041362531,
+      "eval_loss": 0.21597927808761597,
+      "eval_runtime": 35.1856,
+      "eval_samples_per_second": 38.936,
+      "step": 25400
+    },
     {
       "epoch": 2.97,
       "learning_rate": 3.2468288141510528e-06,
-      "loss": 0.3119,
+      "loss": 0.3048,
       "step": 25500
     },
     {
-      "epoch": 2.97,
-      "eval_average_metrics": 90.10827250608273,
-      "eval_em": 89.85401459854015,
+      "epoch": 2.98,
+      "eval_average_metrics": 90.07177615571777,
+      "eval_em": 89.78102189781022,
       "eval_f1": 90.36253041362531,
-      "eval_loss": 0.21424712240695953,
-      "eval_runtime": 15.4741,
-      "eval_samples_per_second": 88.535,
-      "step": 25500
+      "eval_loss": 0.21600975096225739,
+      "eval_runtime": 33.008,
+      "eval_samples_per_second": 41.505,
+      "step": 25600
     },
     {
       "epoch": 3.0,
       "step": 25779,
-      "total_flos": 2.1011901579256954e+17,
-      "train_loss": 0.3297047741756515,
-      "train_runtime": 5659.5381,
-      "train_samples_per_second": 72.877,
-      "train_steps_per_second": 4.555
+      "total_flos": 2.1177253912543027e+17,
+      "train_loss": 0.3175360437943145,
+      "train_runtime": 17672.1841,
+      "train_samples_per_second": 23.339,
+      "train_steps_per_second": 1.459
     }
   ],
   "max_steps": 25779,
   "num_train_epochs": 3,
-  "total_flos": 2.1011901579256954e+17,
+  "total_flos": 2.1177253912543027e+17,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a9fc0b2c802946c2d87bb004854bb82ca322e9997328e616e2973d3f37a9dacc
+oid sha256:6b54ef46ceeb8bfabb09b43e3ec6d062336e479231fa78b4d934c2be81827529
 size 3183

training_config.json

@@ -1 +1 @@
-{"bottleneck_dim": 24, "dataset_config_name": ["en"], "delta_type": "adapter", "do_eval": true, "do_test": true, "do_train": true, "eval_dataset_config_name": ["en"], "eval_dataset_name": "superglue-record", "eval_steps": 200, "evaluation_strategy": "steps", "greater_is_better": true, "learning_rate": 0.0003, "load_best_model_at_end": true, "max_source_length": 512, "metric_for_best_model": "average_metrics", "model_name_or_path": "../../../../plm_cache/t5-base", "num_train_epochs": 3, "output_dir": "outputs/bitfit/t5-base/superglue-record", "overwrite_output_dir": true, "per_device_eval_batch_size": 16, "per_device_train_batch_size": 16, "predict_with_generate": true, "push_to_hub": true, "save_steps": 200, "save_strategy": "steps", "save_total_limit": 1, "seed": 42, "split_validation_test": true, "task_name": "superglue-record", "test_dataset_config_name": ["en"], "test_dataset_name": "superglue-record", "tokenizer_name": "../../../../plm_cache/t5-base", "unfrozen_modules": ["deltas", "layer_norm", "final_layer_norm"], "warmup_steps": 0}
+{"dataset_config_name": ["en"], "delta_type": "lora", "do_eval": true, "do_test": true, "do_train": true, "eval_dataset_config_name": ["en"], "eval_dataset_name": "superglue-record", "eval_steps": 200, "evaluation_strategy": "steps", "greater_is_better": true, "learning_rate": 0.0003, "load_best_model_at_end": true, "lora_r": 8, "max_source_length": 512, "metric_for_best_model": "average_metrics", "model_name_or_path": "../../../../plm_cache/t5-base", "num_train_epochs": 3, "output_dir": "outputs/bitfit/t5-base/superglue-record", "overwrite_output_dir": true, "per_device_eval_batch_size": 16, "per_device_train_batch_size": 16, "predict_with_generate": true, "push_to_hub": true, "save_steps": 200, "save_strategy": "steps", "save_total_limit": 1, "seed": 42, "split_validation_test": true, "task_name": "superglue-record", "test_dataset_config_name": ["en"], "test_dataset_name": "superglue-record", "tokenizer_name": "../../../../plm_cache/t5-base", "unfrozen_modules": ["deltas", "layer_norm", "final_layer_norm"], "warmup_steps": 0}