🍻 cheers

all_results.json

@@ -1,12 +1,12 @@
 {
     "epoch": 7.0,
-    "eval_accuracy": 0.6974621395477492,
-    "eval_loss": 1.0670689344406128,
-    "eval_runtime": 150.7024,
-    "eval_samples_per_second": 191.915,
-    "eval_steps_per_second": 23.994,
-    "train_loss": 1.126858214650741,
-    "train_runtime": 24456.4619,
-    "train_samples_per_second": 66.225,
-    "train_steps_per_second": 2.07
+    "eval_accuracy": 0.7116727750501348,
+    "eval_loss": 0.9854417443275452,
+    "eval_runtime": 153.6785,
+    "eval_samples_per_second": 188.198,
+    "eval_steps_per_second": 23.53,
+    "train_loss": 0.6310260885072734,
+    "train_runtime": 24479.9902,
+    "train_samples_per_second": 66.162,
+    "train_steps_per_second": 2.068
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 7.0,
-    "eval_accuracy": 0.6974621395477492,
-    "eval_loss": 1.0670689344406128,
-    "eval_runtime": 150.7024,
-    "eval_samples_per_second": 191.915,
-    "eval_steps_per_second": 23.994
+    "eval_accuracy": 0.7116727750501348,
+    "eval_loss": 0.9854417443275452,
+    "eval_runtime": 153.6785,
+    "eval_samples_per_second": 188.198,
+    "eval_steps_per_second": 23.53
 }

runs/Jun01_15-30-11_adrian-development/events.out.tfevents.1685650870.adrian-development.134629.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d020070c2dd4a79194aea8b0abdc3fca9b8999fb7746988fb45660d5cb9754bf
+size 369

train_results.json

@@ -1,7 +1,7 @@
 {
     "epoch": 7.0,
-    "train_loss": 1.126858214650741,
-    "train_runtime": 24456.4619,
-    "train_samples_per_second": 66.225,
-    "train_steps_per_second": 2.07
+    "train_loss": 0.6310260885072734,
+    "train_runtime": 24479.9902,
+    "train_samples_per_second": 66.162,
+    "train_steps_per_second": 2.068
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 1.0670689344406128,
-  "best_model_checkpoint": "Transformers/vit-base-clothing-leafs-example-full-simple/checkpoint-47000",
+  "best_metric": 0.9854417443275452,
+  "best_model_checkpoint": "Transformers/vit-base-clothing-leafs-example-full-simple/checkpoint-14000",
   "epoch": 7.0,
   "global_step": 50617,
   "is_hyper_param_search": false,
@@ -9,762 +9,762 @@
   "log_history": [
     {
       "epoch": 0.14,
-      "learning_rate": 4.9012189580575696e-06,
-      "loss": 2.8236,
+      "learning_rate": 4.9012189580575696e-05,
+      "loss": 1.7292,
       "step": 1000
     },
     {
       "epoch": 0.14,
-      "eval_accuracy": 0.4710600926630247,
-      "eval_loss": 2.3487069606781006,
-      "eval_runtime": 155.6192,
-      "eval_samples_per_second": 185.851,
-      "eval_steps_per_second": 23.236,
+      "eval_accuracy": 0.6539312633981053,
+      "eval_loss": 1.3212652206420898,
+      "eval_runtime": 154.4552,
+      "eval_samples_per_second": 187.252,
+      "eval_steps_per_second": 23.411,
       "step": 1000
     },
     {
       "epoch": 0.28,
-      "learning_rate": 4.80243791611514e-06,
-      "loss": 2.1379,
+      "learning_rate": 4.8025366971570815e-05,
+      "loss": 1.2308,
       "step": 2000
     },
     {
       "epoch": 0.28,
-      "eval_accuracy": 0.5444644215476108,
-      "eval_loss": 1.9659085273742676,
-      "eval_runtime": 153.7302,
-      "eval_samples_per_second": 188.135,
-      "eval_steps_per_second": 23.522,
+      "eval_accuracy": 0.6633012931332549,
+      "eval_loss": 1.198843002319336,
+      "eval_runtime": 151.3412,
+      "eval_samples_per_second": 191.105,
+      "eval_steps_per_second": 23.893,
       "step": 2000
     },
     {
       "epoch": 0.41,
-      "learning_rate": 4.703755655214652e-06,
-      "loss": 1.8288,
+      "learning_rate": 4.7037556552146516e-05,
+      "loss": 1.1582,
       "step": 3000
     },
     {
       "epoch": 0.41,
-      "eval_accuracy": 0.6094322660950142,
-      "eval_loss": 1.736711025238037,
-      "eval_runtime": 154.9586,
-      "eval_samples_per_second": 186.643,
-      "eval_steps_per_second": 23.335,
+      "eval_accuracy": 0.6818684738261531,
+      "eval_loss": 1.1182668209075928,
+      "eval_runtime": 152.2609,
+      "eval_samples_per_second": 189.95,
+      "eval_steps_per_second": 23.749,
       "step": 3000
     },
     {
       "epoch": 0.55,
-      "learning_rate": 4.605073394314164e-06,
-      "loss": 1.6449,
+      "learning_rate": 4.604974613272221e-05,
+      "loss": 1.109,
       "step": 4000
     },
     {
       "epoch": 0.55,
-      "eval_accuracy": 0.6326325980222668,
-      "eval_loss": 1.5849583148956299,
-      "eval_runtime": 154.5958,
-      "eval_samples_per_second": 187.081,
-      "eval_steps_per_second": 23.39,
+      "eval_accuracy": 0.6853952008851394,
+      "eval_loss": 1.0872125625610352,
+      "eval_runtime": 151.3068,
+      "eval_samples_per_second": 191.148,
+      "eval_steps_per_second": 23.898,
       "step": 4000
     },
     {
       "epoch": 0.69,
-      "learning_rate": 4.506292352371734e-06,
-      "loss": 1.5127,
+      "learning_rate": 4.5062923523717335e-05,
+      "loss": 1.0772,
       "step": 5000
     },
     {
       "epoch": 0.69,
-      "eval_accuracy": 0.6461517184150474,
-      "eval_loss": 1.4778083562850952,
-      "eval_runtime": 156.4372,
-      "eval_samples_per_second": 184.879,
-      "eval_steps_per_second": 23.115,
+      "eval_accuracy": 0.6994329576101238,
+      "eval_loss": 1.0495002269744873,
+      "eval_runtime": 150.7718,
+      "eval_samples_per_second": 191.826,
+      "eval_steps_per_second": 23.983,
       "step": 5000
     },
     {
       "epoch": 0.83,
-      "learning_rate": 4.407511310429303e-06,
-      "loss": 1.4122,
+      "learning_rate": 4.407511310429303e-05,
+      "loss": 1.0371,
       "step": 6000
     },
     {
       "epoch": 0.83,
-      "eval_accuracy": 0.6564898693036443,
-      "eval_loss": 1.399396300315857,
-      "eval_runtime": 154.7049,
-      "eval_samples_per_second": 186.949,
-      "eval_steps_per_second": 23.374,
+      "eval_accuracy": 0.6932093216236774,
+      "eval_loss": 1.05511474609375,
+      "eval_runtime": 151.6025,
+      "eval_samples_per_second": 190.775,
+      "eval_steps_per_second": 23.852,
       "step": 6000
     },
     {
       "epoch": 0.97,
-      "learning_rate": 4.3088290495288145e-06,
-      "loss": 1.3623,
+      "learning_rate": 4.308829049528815e-05,
+      "loss": 1.0287,
       "step": 7000
     },
     {
       "epoch": 0.97,
-      "eval_accuracy": 0.6619874144250052,
-      "eval_loss": 1.3486990928649902,
-      "eval_runtime": 153.9773,
-      "eval_samples_per_second": 187.833,
-      "eval_steps_per_second": 23.484,
+      "eval_accuracy": 0.7008159878293341,
+      "eval_loss": 1.0264155864715576,
+      "eval_runtime": 152.8647,
+      "eval_samples_per_second": 189.2,
+      "eval_steps_per_second": 23.655,
       "step": 7000
     },
     {
       "epoch": 1.11,
-      "learning_rate": 4.2100480075863845e-06,
-      "loss": 1.293,
+      "learning_rate": 4.210048007586384e-05,
+      "loss": 0.9387,
       "step": 8000
     },
     {
       "epoch": 1.11,
-      "eval_accuracy": 0.6671392019915635,
-      "eval_loss": 1.299405813217163,
-      "eval_runtime": 154.5924,
-      "eval_samples_per_second": 187.086,
-      "eval_steps_per_second": 23.391,
+      "eval_accuracy": 0.7021644422930641,
+      "eval_loss": 1.025620937347412,
+      "eval_runtime": 152.5576,
+      "eval_samples_per_second": 189.581,
+      "eval_steps_per_second": 23.703,
       "step": 8000
     },
     {
       "epoch": 1.24,
-      "learning_rate": 4.1112669656439545e-06,
-      "loss": 1.2382,
+      "learning_rate": 4.111464527727839e-05,
+      "loss": 0.9005,
       "step": 9000
     },
     {
       "epoch": 1.24,
-      "eval_accuracy": 0.6701818684738261,
-      "eval_loss": 1.27021062374115,
-      "eval_runtime": 153.5258,
-      "eval_samples_per_second": 188.385,
-      "eval_steps_per_second": 23.553,
+      "eval_accuracy": 0.7013346241615379,
+      "eval_loss": 1.0224932432174683,
+      "eval_runtime": 151.9509,
+      "eval_samples_per_second": 190.338,
+      "eval_steps_per_second": 23.797,
       "step": 9000
     },
     {
       "epoch": 1.38,
-      "learning_rate": 4.012683485785409e-06,
-      "loss": 1.2186,
+      "learning_rate": 4.012683485785408e-05,
+      "loss": 0.8957,
       "step": 10000
     },
     {
       "epoch": 1.38,
-      "eval_accuracy": 0.6728787774012862,
-      "eval_loss": 1.2421326637268066,
-      "eval_runtime": 155.3885,
-      "eval_samples_per_second": 186.127,
-      "eval_steps_per_second": 23.271,
+      "eval_accuracy": 0.6993638060991633,
+      "eval_loss": 1.0279977321624756,
+      "eval_runtime": 150.8357,
+      "eval_samples_per_second": 191.745,
+      "eval_steps_per_second": 23.973,
       "step": 10000
     },
     {
       "epoch": 1.52,
-      "learning_rate": 3.913902443842978e-06,
-      "loss": 1.1912,
+      "learning_rate": 3.913902443842978e-05,
+      "loss": 0.8924,
       "step": 11000
     },
     {
       "epoch": 1.52,
-      "eval_accuracy": 0.6746767166862596,
-      "eval_loss": 1.2220091819763184,
-      "eval_runtime": 155.3267,
-      "eval_samples_per_second": 186.201,
-      "eval_steps_per_second": 23.28,
+      "eval_accuracy": 0.7037203512896757,
+      "eval_loss": 1.016230821609497,
+      "eval_runtime": 151.3023,
+      "eval_samples_per_second": 191.154,
+      "eval_steps_per_second": 23.899,
       "step": 11000
     },
     {
       "epoch": 1.66,
-      "learning_rate": 3.815121401900548e-06,
-      "loss": 1.1798,
+      "learning_rate": 3.8151214019005474e-05,
+      "loss": 0.8934,
       "step": 12000
     },
     {
       "epoch": 1.66,
-      "eval_accuracy": 0.6796556254754167,
-      "eval_loss": 1.1974430084228516,
-      "eval_runtime": 153.7759,
-      "eval_samples_per_second": 188.079,
-      "eval_steps_per_second": 23.515,
+      "eval_accuracy": 0.7084918055459511,
+      "eval_loss": 1.0030243396759033,
+      "eval_runtime": 151.4775,
+      "eval_samples_per_second": 190.933,
+      "eval_steps_per_second": 23.872,
       "step": 12000
     },
     {
       "epoch": 1.8,
-      "learning_rate": 3.7164391410000594e-06,
-      "loss": 1.1605,
+      "learning_rate": 3.7164391410000594e-05,
+      "loss": 0.8959,
       "step": 13000
     },
     {
       "epoch": 1.8,
-      "eval_accuracy": 0.6826982919576793,
-      "eval_loss": 1.1833155155181885,
-      "eval_runtime": 153.6706,
-      "eval_samples_per_second": 188.208,
-      "eval_steps_per_second": 23.531,
+      "eval_accuracy": 0.7021298665375838,
+      "eval_loss": 1.017107367515564,
+      "eval_runtime": 151.0427,
+      "eval_samples_per_second": 191.482,
+      "eval_steps_per_second": 23.94,
       "step": 13000
     },
     {
       "epoch": 1.94,
-      "learning_rate": 3.6176580990576294e-06,
-      "loss": 1.1454,
+      "learning_rate": 3.6176580990576294e-05,
+      "loss": 0.8927,
       "step": 14000
     },
     {
       "epoch": 1.94,
-      "eval_accuracy": 0.6838047161330475,
-      "eval_loss": 1.1689262390136719,
-      "eval_runtime": 155.3074,
-      "eval_samples_per_second": 186.224,
-      "eval_steps_per_second": 23.283,
+      "eval_accuracy": 0.7116727750501348,
+      "eval_loss": 0.9854417443275452,
+      "eval_runtime": 151.5533,
+      "eval_samples_per_second": 190.837,
+      "eval_steps_per_second": 23.86,
       "step": 14000
     },
     {
       "epoch": 2.07,
-      "learning_rate": 3.518975838157141e-06,
-      "loss": 1.1076,
+      "learning_rate": 3.518877057115199e-05,
+      "loss": 0.8107,
       "step": 15000
     },
     {
       "epoch": 2.07,
-      "eval_accuracy": 0.6820759283590346,
-      "eval_loss": 1.1666451692581177,
-      "eval_runtime": 156.0771,
-      "eval_samples_per_second": 185.306,
-      "eval_steps_per_second": 23.168,
+      "eval_accuracy": 0.701887836249222,
+      "eval_loss": 1.0297770500183105,
+      "eval_runtime": 151.6336,
+      "eval_samples_per_second": 190.736,
+      "eval_steps_per_second": 23.847,
       "step": 15000
     },
     {
       "epoch": 2.21,
-      "learning_rate": 3.4201947962147106e-06,
-      "loss": 1.0882,
+      "learning_rate": 3.420194796214711e-05,
+      "loss": 0.7442,
       "step": 16000
     },
     {
       "epoch": 2.21,
-      "eval_accuracy": 0.6835626858446857,
-      "eval_loss": 1.1561516523361206,
-      "eval_runtime": 154.5444,
-      "eval_samples_per_second": 187.144,
-      "eval_steps_per_second": 23.398,
+      "eval_accuracy": 0.707523684392504,
+      "eval_loss": 1.0102756023406982,
+      "eval_runtime": 154.4177,
+      "eval_samples_per_second": 187.297,
+      "eval_steps_per_second": 23.417,
       "step": 16000
     },
     {
       "epoch": 2.35,
-      "learning_rate": 3.3214137542722802e-06,
-      "loss": 1.0832,
+      "learning_rate": 3.32141375427228e-05,
+      "loss": 0.7421,
       "step": 17000
     },
     {
       "epoch": 2.35,
-      "eval_accuracy": 0.6874005947029943,
-      "eval_loss": 1.1425527334213257,
-      "eval_runtime": 155.3385,
-      "eval_samples_per_second": 186.187,
-      "eval_steps_per_second": 23.278,
+      "eval_accuracy": 0.708595532812392,
+      "eval_loss": 1.0141746997833252,
+      "eval_runtime": 154.77,
+      "eval_samples_per_second": 186.871,
+      "eval_steps_per_second": 23.364,
       "step": 17000
     },
     {
       "epoch": 2.49,
-      "learning_rate": 3.2227314933717923e-06,
-      "loss": 1.0698,
+      "learning_rate": 3.222731493371792e-05,
+      "loss": 0.7311,
       "step": 18000
     },
     {
       "epoch": 2.49,
-      "eval_accuracy": 0.6872622916810732,
-      "eval_loss": 1.1318212747573853,
-      "eval_runtime": 155.6543,
-      "eval_samples_per_second": 185.809,
-      "eval_steps_per_second": 23.231,
+      "eval_accuracy": 0.7110504114514902,
+      "eval_loss": 1.01434326171875,
+      "eval_runtime": 153.512,
+      "eval_samples_per_second": 188.402,
+      "eval_steps_per_second": 23.555,
       "step": 18000
     },
     {
       "epoch": 2.63,
-      "learning_rate": 3.123950451429362e-06,
-      "loss": 1.0752,
+      "learning_rate": 3.1239504514293614e-05,
+      "loss": 0.7362,
       "step": 19000
     },
     {
       "epoch": 2.63,
-      "eval_accuracy": 0.6842542009542909,
-      "eval_loss": 1.1395872831344604,
-      "eval_runtime": 155.6622,
-      "eval_samples_per_second": 185.8,
-      "eval_steps_per_second": 23.23,
+      "eval_accuracy": 0.709943987276122,
+      "eval_loss": 1.0113046169281006,
+      "eval_runtime": 153.941,
+      "eval_samples_per_second": 187.877,
+      "eval_steps_per_second": 23.49,
       "step": 19000
     },
     {
       "epoch": 2.77,
-      "learning_rate": 3.0251694094869315e-06,
-      "loss": 1.0659,
+      "learning_rate": 3.0252681905288737e-05,
+      "loss": 0.7392,
       "step": 20000
     },
     {
       "epoch": 2.77,
-      "eval_accuracy": 0.6903049581633358,
-      "eval_loss": 1.1166965961456299,
-      "eval_runtime": 154.6931,
-      "eval_samples_per_second": 186.964,
-      "eval_steps_per_second": 23.375,
+      "eval_accuracy": 0.7046193209321624,
+      "eval_loss": 1.03049898147583,
+      "eval_runtime": 154.7774,
+      "eval_samples_per_second": 186.862,
+      "eval_steps_per_second": 23.363,
       "step": 20000
     },
     {
       "epoch": 2.9,
-      "learning_rate": 2.926487148586444e-06,
-      "loss": 1.0561,
+      "learning_rate": 2.9264871485864437e-05,
+      "loss": 0.7515,
       "step": 21000
     },
     {
       "epoch": 2.9,
-      "eval_accuracy": 0.6879538067906784,
-      "eval_loss": 1.1178348064422607,
-      "eval_runtime": 152.6144,
-      "eval_samples_per_second": 189.51,
-      "eval_steps_per_second": 23.694,
+      "eval_accuracy": 0.7069013207938594,
+      "eval_loss": 1.0068926811218262,
+      "eval_runtime": 153.9311,
+      "eval_samples_per_second": 187.889,
+      "eval_steps_per_second": 23.491,
       "step": 21000
     },
     {
       "epoch": 3.04,
-      "learning_rate": 2.827706106644013e-06,
-      "loss": 1.0328,
+      "learning_rate": 2.827706106644013e-05,
+      "loss": 0.6945,
       "step": 22000
     },
     {
       "epoch": 3.04,
-      "eval_accuracy": 0.690581564207178,
-      "eval_loss": 1.1114201545715332,
-      "eval_runtime": 153.9011,
-      "eval_samples_per_second": 187.926,
-      "eval_steps_per_second": 23.496,
+      "eval_accuracy": 0.7095636539658392,
+      "eval_loss": 1.0314772129058838,
+      "eval_runtime": 151.7505,
+      "eval_samples_per_second": 190.589,
+      "eval_steps_per_second": 23.829,
       "step": 22000
     },
     {
       "epoch": 3.18,
-      "learning_rate": 2.729023845743525e-06,
-      "loss": 1.0299,
+      "learning_rate": 2.7289250647015823e-05,
+      "loss": 0.5912,
       "step": 23000
     },
     {
       "epoch": 3.18,
-      "eval_accuracy": 0.6917225641380265,
-      "eval_loss": 1.1057274341583252,
-      "eval_runtime": 155.3852,
-      "eval_samples_per_second": 186.131,
-      "eval_steps_per_second": 23.271,
+      "eval_accuracy": 0.7004702302745315,
+      "eval_loss": 1.0634562969207764,
+      "eval_runtime": 152.7645,
+      "eval_samples_per_second": 189.324,
+      "eval_steps_per_second": 23.67,
       "step": 23000
     },
     {
       "epoch": 3.32,
-      "learning_rate": 2.6302428038010947e-06,
-      "loss": 0.9961,
+      "learning_rate": 2.6302428038010946e-05,
+      "loss": 0.58,
       "step": 24000
     },
     {
       "epoch": 3.32,
-      "eval_accuracy": 0.6912730793167831,
-      "eval_loss": 1.1056290864944458,
-      "eval_runtime": 154.2366,
-      "eval_samples_per_second": 187.517,
-      "eval_steps_per_second": 23.445,
+      "eval_accuracy": 0.7049996542424452,
+      "eval_loss": 1.0592378377914429,
+      "eval_runtime": 153.0117,
+      "eval_samples_per_second": 189.018,
+      "eval_steps_per_second": 23.632,
       "step": 24000
     },
     {
       "epoch": 3.46,
-      "learning_rate": 2.5315605429006068e-06,
-      "loss": 1.0128,
+      "learning_rate": 2.5314617618586643e-05,
+      "loss": 0.576,
       "step": 25000
     },
     {
       "epoch": 3.46,
-      "eval_accuracy": 0.6937971094668418,
-      "eval_loss": 1.0973228216171265,
-      "eval_runtime": 156.2066,
-      "eval_samples_per_second": 185.152,
-      "eval_steps_per_second": 23.149,
+      "eval_accuracy": 0.7018532604937417,
+      "eval_loss": 1.0623606443405151,
+      "eval_runtime": 154.2965,
+      "eval_samples_per_second": 187.444,
+      "eval_steps_per_second": 23.435,
       "step": 25000
     },
     {
       "epoch": 3.6,
-      "learning_rate": 2.4327795009581764e-06,
-      "loss": 1.0118,
+      "learning_rate": 2.4326807199162337e-05,
+      "loss": 0.5882,
       "step": 26000
     },
     {
       "epoch": 3.6,
-      "eval_accuracy": 0.6942465942880852,
-      "eval_loss": 1.0930811166763306,
-      "eval_runtime": 155.695,
-      "eval_samples_per_second": 185.761,
-      "eval_steps_per_second": 23.225,
+      "eval_accuracy": 0.7058986238849319,
+      "eval_loss": 1.0582630634307861,
+      "eval_runtime": 154.8374,
+      "eval_samples_per_second": 186.79,
+      "eval_steps_per_second": 23.354,
       "step": 26000
     },
     {
       "epoch": 3.73,
-      "learning_rate": 2.333998459015746e-06,
-      "loss": 1.0045,
+      "learning_rate": 2.3338996779738033e-05,
+      "loss": 0.5867,
       "step": 27000
     },
     {
       "epoch": 3.73,
-      "eval_accuracy": 0.6936933822004011,
-      "eval_loss": 1.089782953262329,
-      "eval_runtime": 155.3577,
-      "eval_samples_per_second": 186.164,
-      "eval_steps_per_second": 23.275,
+      "eval_accuracy": 0.7076965631699053,
+      "eval_loss": 1.0526437759399414,
+      "eval_runtime": 153.5073,
+      "eval_samples_per_second": 188.408,
+      "eval_steps_per_second": 23.556,
       "step": 27000
     },
     {
       "epoch": 3.87,
-      "learning_rate": 2.2354149791572e-06,
-      "loss": 0.9923,
+      "learning_rate": 2.2352174170733153e-05,
+      "loss": 0.593,
       "step": 28000
     },
     {
       "epoch": 3.87,
-      "eval_accuracy": 0.6958716547956573,
-      "eval_loss": 1.0858705043792725,
-      "eval_runtime": 155.5851,
-      "eval_samples_per_second": 185.892,
-      "eval_steps_per_second": 23.241,
+      "eval_accuracy": 0.7047921997095636,
+      "eval_loss": 1.0647141933441162,
+      "eval_runtime": 153.5176,
+      "eval_samples_per_second": 188.395,
+      "eval_steps_per_second": 23.554,
       "step": 28000
     },
     {
       "epoch": 4.01,
-      "learning_rate": 2.13663393721477e-06,
-      "loss": 0.9988,
+      "learning_rate": 2.136436375130885e-05,
+      "loss": 0.5745,
       "step": 29000
     },
     {
       "epoch": 4.01,
-      "eval_accuracy": 0.6943848973100062,
-      "eval_loss": 1.0852184295654297,
-      "eval_runtime": 154.9704,
-      "eval_samples_per_second": 186.629,
-      "eval_steps_per_second": 23.333,
+      "eval_accuracy": 0.7044810179102413,
+      "eval_loss": 1.0814189910888672,
+      "eval_runtime": 147.6133,
+      "eval_samples_per_second": 195.931,
+      "eval_steps_per_second": 24.496,
       "step": 29000
     },
     {
       "epoch": 4.15,
-      "learning_rate": 2.0378528952723396e-06,
-      "loss": 0.9773,
+      "learning_rate": 2.0377541142303973e-05,
+      "loss": 0.4332,
       "step": 30000
     },
     {
       "epoch": 4.15,
-      "eval_accuracy": 0.6930018670907959,
-      "eval_loss": 1.089293122291565,
-      "eval_runtime": 156.3077,
-      "eval_samples_per_second": 185.033,
-      "eval_steps_per_second": 23.134,
+      "eval_accuracy": 0.7036166240232349,
+      "eval_loss": 1.1231409311294556,
+      "eval_runtime": 147.2084,
+      "eval_samples_per_second": 196.47,
+      "eval_steps_per_second": 24.564,
       "step": 30000
     },
     {
       "epoch": 4.29,
-      "learning_rate": 1.9391706343718513e-06,
-      "loss": 0.9577,
+      "learning_rate": 1.9389730722879666e-05,
+      "loss": 0.4427,
       "step": 31000
     },
     {
       "epoch": 4.29,
-      "eval_accuracy": 0.6968052001936242,
-      "eval_loss": 1.0807169675827026,
-      "eval_runtime": 154.1329,
-      "eval_samples_per_second": 187.643,
-      "eval_steps_per_second": 23.46,
+      "eval_accuracy": 0.6981536546573542,
+      "eval_loss": 1.1329883337020874,
+      "eval_runtime": 156.5248,
+      "eval_samples_per_second": 184.776,
+      "eval_steps_per_second": 23.102,
       "step": 31000
     },
     {
       "epoch": 4.43,
-      "learning_rate": 1.840389592429421e-06,
-      "loss": 0.9748,
+      "learning_rate": 1.8402908113874786e-05,
+      "loss": 0.4475,
       "step": 32000
     },
     {
       "epoch": 4.43,
-      "eval_accuracy": 0.6957333517737363,
-      "eval_loss": 1.0788837671279907,
-      "eval_runtime": 154.5313,
-      "eval_samples_per_second": 187.16,
-      "eval_steps_per_second": 23.4,
+      "eval_accuracy": 0.7000207454532882,
+      "eval_loss": 1.126924991607666,
+      "eval_runtime": 155.4925,
+      "eval_samples_per_second": 186.003,
+      "eval_steps_per_second": 23.255,
       "step": 32000
     },
     {
       "epoch": 4.56,
-      "learning_rate": 1.7416085504869907e-06,
-      "loss": 0.9777,
+      "learning_rate": 1.7415097694450482e-05,
+      "loss": 0.4488,
       "step": 33000
     },
     {
       "epoch": 4.56,
-      "eval_accuracy": 0.6924486550031118,
-      "eval_loss": 1.0864237546920776,
-      "eval_runtime": 155.2284,
-      "eval_samples_per_second": 186.319,
-      "eval_steps_per_second": 23.295,
+      "eval_accuracy": 0.7025101998478667,
+      "eval_loss": 1.1300145387649536,
+      "eval_runtime": 153.2715,
+      "eval_samples_per_second": 188.698,
+      "eval_steps_per_second": 23.592,
       "step": 33000
     },
     {
       "epoch": 4.7,
-      "learning_rate": 1.6428275085445602e-06,
-      "loss": 0.9536,
+      "learning_rate": 1.6428275085445602e-05,
+      "loss": 0.4431,
       "step": 34000
     },
     {
       "epoch": 4.7,
-      "eval_accuracy": 0.6949035336422101,
-      "eval_loss": 1.0813385248184204,
-      "eval_runtime": 153.9831,
-      "eval_samples_per_second": 187.826,
-      "eval_steps_per_second": 23.483,
+      "eval_accuracy": 0.701887836249222,
+      "eval_loss": 1.1304574012756348,
+      "eval_runtime": 153.1234,
+      "eval_samples_per_second": 188.88,
+      "eval_steps_per_second": 23.615,
       "step": 34000
     },
     {
       "epoch": 4.84,
-      "learning_rate": 1.544145247644072e-06,
-      "loss": 0.9507,
+      "learning_rate": 1.54404646660213e-05,
+      "loss": 0.4514,
       "step": 35000
     },
     {
       "epoch": 4.84,
-      "eval_accuracy": 0.6950418366641311,
-      "eval_loss": 1.0795472860336304,
-      "eval_runtime": 156.0888,
-      "eval_samples_per_second": 185.292,
-      "eval_steps_per_second": 23.166,
+      "eval_accuracy": 0.6990872000553212,
+      "eval_loss": 1.1445332765579224,
+      "eval_runtime": 155.0001,
+      "eval_samples_per_second": 186.593,
+      "eval_steps_per_second": 23.329,
       "step": 35000
     },
     {
       "epoch": 4.98,
-      "learning_rate": 1.4453642057016417e-06,
-      "loss": 0.9627,
+      "learning_rate": 1.4452654246596994e-05,
+      "loss": 0.4427,
       "step": 36000
     },
     {
       "epoch": 4.98,
-      "eval_accuracy": 0.6955258972408547,
-      "eval_loss": 1.0754951238632202,
-      "eval_runtime": 156.2467,
-      "eval_samples_per_second": 185.105,
-      "eval_steps_per_second": 23.143,
+      "eval_accuracy": 0.702821381647189,
+      "eval_loss": 1.1224807500839233,
+      "eval_runtime": 153.3684,
+      "eval_samples_per_second": 188.579,
+      "eval_steps_per_second": 23.577,
       "step": 36000
     },
     {
       "epoch": 5.12,
-      "learning_rate": 1.3465831637592115e-06,
-      "loss": 0.9399,
+      "learning_rate": 1.346681944801154e-05,
+      "loss": 0.3391,
       "step": 37000
     },
     {
       "epoch": 5.12,
-      "eval_accuracy": 0.6960791093285388,
-      "eval_loss": 1.0770469903945923,
-      "eval_runtime": 153.8614,
-      "eval_samples_per_second": 187.974,
-      "eval_steps_per_second": 23.502,
+      "eval_accuracy": 0.6964248668833414,
+      "eval_loss": 1.1918517351150513,
+      "eval_runtime": 152.8672,
+      "eval_samples_per_second": 189.197,
+      "eval_steps_per_second": 23.655,
       "step": 37000
     },
     {
       "epoch": 5.26,
-      "learning_rate": 1.2479009028587235e-06,
-      "loss": 0.9357,
+      "learning_rate": 1.2479009028587233e-05,
+      "loss": 0.3222,
       "step": 38000
     },
     {
       "epoch": 5.26,
-      "eval_accuracy": 0.6960791093285388,
-      "eval_loss": 1.0759111642837524,
-      "eval_runtime": 153.1358,
-      "eval_samples_per_second": 188.865,
-      "eval_steps_per_second": 23.613,
+      "eval_accuracy": 0.6971163819929466,
+      "eval_loss": 1.2108075618743896,
+      "eval_runtime": 155.4241,
+      "eval_samples_per_second": 186.084,
+      "eval_steps_per_second": 23.265,
       "step": 38000
     },
     {
       "epoch": 5.39,
-      "learning_rate": 1.1491198609162931e-06,
-      "loss": 0.943,
+      "learning_rate": 1.149119860916293e-05,
+      "loss": 0.315,
       "step": 39000
     },
     {
       "epoch": 5.39,
-      "eval_accuracy": 0.6965977456607427,
-      "eval_loss": 1.0720691680908203,
-      "eval_runtime": 154.5633,
-      "eval_samples_per_second": 187.121,
-      "eval_steps_per_second": 23.395,
+      "eval_accuracy": 0.6985685637231174,
+      "eval_loss": 1.2174683809280396,
+      "eval_runtime": 154.5842,
+      "eval_samples_per_second": 187.095,
+      "eval_steps_per_second": 23.392,
       "step": 39000
     },
     {
       "epoch": 5.53,
-      "learning_rate": 1.0504376000158052e-06,
-      "loss": 0.9244,
+      "learning_rate": 1.0503388189738627e-05,
+      "loss": 0.3307,
       "step": 40000
     },
     {
       "epoch": 5.53,
-      "eval_accuracy": 0.696908927460065,
-      "eval_loss": 1.0704323053359985,
-      "eval_runtime": 154.315,
-      "eval_samples_per_second": 187.422,
-      "eval_steps_per_second": 23.433,
+      "eval_accuracy": 0.6943848973100062,
+      "eval_loss": 1.219739556312561,
+      "eval_runtime": 155.288,
+      "eval_samples_per_second": 186.247,
+      "eval_steps_per_second": 23.286,
       "step": 40000
     },
     {
       "epoch": 5.67,
-      "learning_rate": 9.516565580733748e-07,
-      "loss": 0.9231,
+      "learning_rate": 9.516565580733746e-06,
+      "loss": 0.3187,
       "step": 41000
     },
     {
       "epoch": 5.67,
-      "eval_accuracy": 0.6960445335730586,
-      "eval_loss": 1.0727081298828125,
-      "eval_runtime": 153.587,
-      "eval_samples_per_second": 188.31,
-      "eval_steps_per_second": 23.544,
+      "eval_accuracy": 0.6988105940114792,
+      "eval_loss": 1.228104591369629,
+      "eval_runtime": 154.3425,
+      "eval_samples_per_second": 187.388,
+      "eval_steps_per_second": 23.428,
       "step": 41000
     },
     {
       "epoch": 5.81,
-      "learning_rate": 8.528755161309442e-07,
-      "loss": 0.9294,
+      "learning_rate": 8.528755161309441e-06,
+      "loss": 0.3327,
       "step": 42000
     },
     {
       "epoch": 5.81,
-      "eval_accuracy": 0.6969780789710255,
-      "eval_loss": 1.0715699195861816,
-      "eval_runtime": 153.1528,
-      "eval_samples_per_second": 188.844,
-      "eval_steps_per_second": 23.61,
+      "eval_accuracy": 0.6966668971717032,
+      "eval_loss": 1.2378689050674438,
+      "eval_runtime": 147.7901,
+      "eval_samples_per_second": 195.696,
+      "eval_steps_per_second": 24.467,
       "step": 42000
     },
     {
       "epoch": 5.95,
-      "learning_rate": 7.540944741885138e-07,
-      "loss": 0.9416,
+      "learning_rate": 7.540944741885138e-06,
+      "loss": 0.3213,
       "step": 43000
     },
     {
       "epoch": 5.95,
-      "eval_accuracy": 0.6980845031463937,
-      "eval_loss": 1.0693832635879517,
-      "eval_runtime": 151.6066,
-      "eval_samples_per_second": 190.77,
-      "eval_steps_per_second": 23.851,
+      "eval_accuracy": 0.6972201092593873,
+      "eval_loss": 1.2366853952407837,
+      "eval_runtime": 155.5771,
+      "eval_samples_per_second": 185.901,
+      "eval_steps_per_second": 23.242,
       "step": 43000
     },
     {
       "epoch": 6.08,
-      "learning_rate": 6.554122132880259e-07,
-      "loss": 0.9248,
+      "learning_rate": 6.553134322460834e-06,
+      "loss": 0.2688,
       "step": 44000
     },
     {
       "epoch": 6.08,
-      "eval_accuracy": 0.6991217758108015,
-      "eval_loss": 1.0678476095199585,
-      "eval_runtime": 151.367,
-      "eval_samples_per_second": 191.072,
-      "eval_steps_per_second": 23.889,
+      "eval_accuracy": 0.6940737155106839,
+      "eval_loss": 1.2731064558029175,
+      "eval_runtime": 155.6374,
+      "eval_samples_per_second": 185.829,
+      "eval_steps_per_second": 23.233,
       "step": 44000
     },
     {
       "epoch": 6.22,
-      "learning_rate": 5.566311713455954e-07,
-      "loss": 0.9137,
+      "learning_rate": 5.56532390303653e-06,
+      "loss": 0.2395,
       "step": 45000
     },
     {
       "epoch": 6.22,
-      "eval_accuracy": 0.6976350183251504,
-      "eval_loss": 1.0700552463531494,
-      "eval_runtime": 152.3071,
-      "eval_samples_per_second": 189.893,
-      "eval_steps_per_second": 23.742,
+      "eval_accuracy": 0.696632321416223,
+      "eval_loss": 1.2903701066970825,
+      "eval_runtime": 153.3038,
+      "eval_samples_per_second": 188.658,
+      "eval_steps_per_second": 23.587,
       "step": 45000
     },
     {
       "epoch": 6.36,
-      "learning_rate": 4.5785012940316495e-07,
-      "loss": 0.91,
+      "learning_rate": 4.5785012940316495e-06,
+      "loss": 0.2407,
       "step": 46000
     },
     {
       "epoch": 6.36,
-      "eval_accuracy": 0.6971855335039071,
-      "eval_loss": 1.0688731670379639,
-      "eval_runtime": 151.0293,
-      "eval_samples_per_second": 191.499,
-      "eval_steps_per_second": 23.942,
+      "eval_accuracy": 0.6951455639305719,
+      "eval_loss": 1.2933955192565918,
+      "eval_runtime": 155.5305,
+      "eval_samples_per_second": 185.957,
+      "eval_steps_per_second": 23.249,
       "step": 46000
     },
     {
       "epoch": 6.5,
-      "learning_rate": 3.59167868502677e-07,
-      "loss": 0.9256,
+      "learning_rate": 3.5906908746073454e-06,
+      "loss": 0.2312,
       "step": 47000
     },
     {
       "epoch": 6.5,
-      "eval_accuracy": 0.6974621395477492,
-      "eval_loss": 1.0670689344406128,
-      "eval_runtime": 150.9586,
-      "eval_samples_per_second": 191.589,
-      "eval_steps_per_second": 23.954,
+      "eval_accuracy": 0.6955258972408547,
+      "eval_loss": 1.2975932359695435,
+      "eval_runtime": 155.813,
+      "eval_samples_per_second": 185.62,
+      "eval_steps_per_second": 23.207,
       "step": 47000
     },
     {
       "epoch": 6.64,
-      "learning_rate": 2.6048560760218905e-07,
-      "loss": 0.9085,
+      "learning_rate": 2.6028804551830413e-06,
+      "loss": 0.2369,
       "step": 48000
     },
     {
       "epoch": 6.64,
-      "eval_accuracy": 0.6985339879676371,
-      "eval_loss": 1.067813754081726,
-      "eval_runtime": 151.0599,
-      "eval_samples_per_second": 191.46,
-      "eval_steps_per_second": 23.938,
+      "eval_accuracy": 0.6934859276675195,
+      "eval_loss": 1.3018196821212769,
+      "eval_runtime": 150.7052,
+      "eval_samples_per_second": 191.911,
+      "eval_steps_per_second": 23.994,
       "step": 48000
     },
     {
       "epoch": 6.78,
-      "learning_rate": 1.617045656597586e-07,
-      "loss": 0.9169,
+      "learning_rate": 1.6170456565975861e-06,
+      "loss": 0.2317,
       "step": 49000
     },
     {
       "epoch": 6.78,
-      "eval_accuracy": 0.6984302607011963,
-      "eval_loss": 1.0689929723739624,
-      "eval_runtime": 151.6049,
-      "eval_samples_per_second": 190.772,
-      "eval_steps_per_second": 23.851,
+      "eval_accuracy": 0.6949381093976903,
+      "eval_loss": 1.3031286001205444,
+      "eval_runtime": 154.6116,
+      "eval_samples_per_second": 187.062,
+      "eval_steps_per_second": 23.388,
       "step": 49000
     },
     {
       "epoch": 6.91,
-      "learning_rate": 6.292352371732817e-08,
-      "loss": 0.9087,
+      "learning_rate": 6.292352371732817e-07,
+      "loss": 0.2395,
       "step": 50000
     },
     {
       "epoch": 6.91,
-      "eval_accuracy": 0.6972546850148675,
-      "eval_loss": 1.069164752960205,
-      "eval_runtime": 151.0946,
-      "eval_samples_per_second": 191.417,
-      "eval_steps_per_second": 23.932,
+      "eval_accuracy": 0.6957679275292166,
+      "eval_loss": 1.3016529083251953,
+      "eval_runtime": 153.6886,
+      "eval_samples_per_second": 188.186,
+      "eval_steps_per_second": 23.528,
       "step": 50000
     },
     {
       "epoch": 7.0,
       "step": 50617,
       "total_flos": 1.2555854407514107e+20,
-      "train_loss": 1.126858214650741,
-      "train_runtime": 24456.4619,
-      "train_samples_per_second": 66.225,
-      "train_steps_per_second": 2.07
+      "train_loss": 0.6310260885072734,
+      "train_runtime": 24479.9902,
+      "train_samples_per_second": 66.162,
+      "train_steps_per_second": 2.068
     }
   ],
   "max_steps": 50617,