End of training

all_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 5.0,
-    "total_flos": 1.9922182705152e+18,
-    "train_loss": 0.11038042506926758,
-    "train_runtime": 7101.5946,
-    "train_samples_per_second": 56.325,
-    "train_steps_per_second": 0.111
+    "total_flos": 3.78521471397888e+18,
+    "train_loss": 0.055960290434986654,
+    "train_runtime": 18290.5686,
+    "train_samples_per_second": 41.551,
+    "train_steps_per_second": 0.054
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 5.0,
-    "total_flos": 1.9922182705152e+18,
-    "train_loss": 0.11038042506926758,
-    "train_runtime": 7101.5946,
-    "train_samples_per_second": 56.325,
-    "train_steps_per_second": 0.111
+    "total_flos": 3.78521471397888e+18,
+    "train_loss": 0.055960290434986654,
+    "train_runtime": 18290.5686,
+    "train_samples_per_second": 41.551,
+    "train_steps_per_second": 0.054
 }

trainer_state.json

@@ -1,443 +1,313 @@
 {
-  "best_metric": 0.05071697756648064,
-  "best_model_checkpoint": "./dinov2_Liveness_detection_v2.2.1/checkpoint-512",
+  "best_metric": 0.030054476112127304,
+  "best_model_checkpoint": "./dinov2_Liveness_detection_v2.2.1/checkpoint-960",
   "epoch": 5.0,
-  "eval_steps": 128,
-  "global_step": 785,
+  "eval_steps": 64,
+  "global_step": 990,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.10191082802547771,
-      "grad_norm": 3.0143625736236572,
-      "learning_rate": 4.923566878980892e-05,
-      "loss": 1.8573,
-      "step": 16
-    },
-    {
-      "epoch": 0.20382165605095542,
-      "grad_norm": 13.854424476623535,
-      "learning_rate": 4.821656050955414e-05,
-      "loss": 0.4389,
-      "step": 32
-    },
-    {
-      "epoch": 0.3057324840764331,
-      "grad_norm": 57.04545593261719,
-      "learning_rate": 4.7197452229299366e-05,
-      "loss": 0.3386,
-      "step": 48
-    },
-    {
-      "epoch": 0.40764331210191085,
-      "grad_norm": 26.68746566772461,
-      "learning_rate": 4.617834394904459e-05,
-      "loss": 0.3205,
+      "epoch": 0.32323232323232326,
+      "grad_norm": 5.1993489265441895,
+      "learning_rate": 4.696969696969697e-05,
+      "loss": 0.4052,
       "step": 64
     },
     {
-      "epoch": 0.5095541401273885,
-      "grad_norm": 28.5941219329834,
-      "learning_rate": 4.5159235668789814e-05,
-      "loss": 0.2216,
-      "step": 80
-    },
-    {
-      "epoch": 0.6114649681528662,
-      "grad_norm": 7.417737007141113,
-      "learning_rate": 4.414012738853504e-05,
-      "loss": 0.1841,
-      "step": 96
-    },
-    {
-      "epoch": 0.7133757961783439,
-      "grad_norm": 15.305898666381836,
-      "learning_rate": 4.312101910828026e-05,
-      "loss": 0.2161,
-      "step": 112
+      "epoch": 0.32323232323232326,
+      "eval_accuracy": 0.9712097686781231,
+      "eval_f1": 0.9713372966580719,
+      "eval_loss": 0.07746175676584244,
+      "eval_precision": 0.9712492471516934,
+      "eval_recall": 0.9712097686781231,
+      "eval_runtime": 513.7785,
+      "eval_samples_per_second": 73.962,
+      "eval_steps_per_second": 9.245,
+      "step": 64
     },
     {
-      "epoch": 0.8152866242038217,
-      "grad_norm": 65.58370208740234,
-      "learning_rate": 4.210191082802548e-05,
-      "loss": 0.183,
+      "epoch": 0.6464646464646465,
+      "grad_norm": 4.368253231048584,
+      "learning_rate": 4.3737373737373736e-05,
+      "loss": 0.0784,
       "step": 128
     },
     {
-      "epoch": 0.8152866242038217,
-      "eval_accuracy": 0.9016,
-      "eval_f1": 0.9039152443229508,
-      "eval_loss": 0.24732650816440582,
-      "eval_precision": 0.9122675566006839,
-      "eval_recall": 0.9016,
-      "eval_runtime": 285.4094,
-      "eval_samples_per_second": 70.075,
-      "eval_steps_per_second": 8.759,
+      "epoch": 0.6464646464646465,
+      "eval_accuracy": 0.9802889549724992,
+      "eval_f1": 0.980265299265069,
+      "eval_loss": 0.054525312036275864,
+      "eval_precision": 0.9803559157542755,
+      "eval_recall": 0.9802889549724992,
+      "eval_runtime": 493.8537,
+      "eval_samples_per_second": 76.946,
+      "eval_steps_per_second": 9.618,
       "step": 128
     },
     {
-      "epoch": 0.9171974522292994,
-      "grad_norm": 13.938675880432129,
-      "learning_rate": 4.10828025477707e-05,
-      "loss": 0.1348,
-      "step": 144
-    },
-    {
-      "epoch": 1.019108280254777,
-      "grad_norm": 7.33504056930542,
-      "learning_rate": 4.0063694267515926e-05,
-      "loss": 0.1039,
-      "step": 160
-    },
-    {
-      "epoch": 1.1210191082802548,
-      "grad_norm": 29.384906768798828,
-      "learning_rate": 3.904458598726115e-05,
-      "loss": 0.1084,
-      "step": 176
-    },
-    {
-      "epoch": 1.2229299363057324,
-      "grad_norm": 14.105592727661133,
-      "learning_rate": 3.8025477707006374e-05,
-      "loss": 0.0855,
+      "epoch": 0.9696969696969697,
+      "grad_norm": 6.4717583656311035,
+      "learning_rate": 4.0505050505050506e-05,
+      "loss": 0.0639,
       "step": 192
     },
     {
-      "epoch": 1.3248407643312101,
-      "grad_norm": 4.494396209716797,
-      "learning_rate": 3.700636942675159e-05,
-      "loss": 0.1292,
-      "step": 208
-    },
-    {
-      "epoch": 1.426751592356688,
-      "grad_norm": 15.541342735290527,
-      "learning_rate": 3.5987261146496815e-05,
-      "loss": 0.1187,
-      "step": 224
-    },
-    {
-      "epoch": 1.5286624203821657,
-      "grad_norm": 37.96527862548828,
-      "learning_rate": 3.496815286624204e-05,
-      "loss": 0.0875,
-      "step": 240
+      "epoch": 0.9696969696969697,
+      "eval_accuracy": 0.9771046606489644,
+      "eval_f1": 0.9772131122819654,
+      "eval_loss": 0.0614825114607811,
+      "eval_precision": 0.9772293201056353,
+      "eval_recall": 0.9771046606489644,
+      "eval_runtime": 494.1919,
+      "eval_samples_per_second": 76.893,
+      "eval_steps_per_second": 9.612,
+      "step": 192
     },
     {
-      "epoch": 1.6305732484076434,
-      "grad_norm": 8.300914764404297,
-      "learning_rate": 3.394904458598726e-05,
-      "loss": 0.1022,
+      "epoch": 1.2929292929292928,
+      "grad_norm": 5.717361927032471,
+      "learning_rate": 3.7272727272727276e-05,
+      "loss": 0.0479,
       "step": 256
     },
     {
-      "epoch": 1.6305732484076434,
-      "eval_accuracy": 0.9729,
-      "eval_f1": 0.9727317036154979,
-      "eval_loss": 0.07502331584692001,
-      "eval_precision": 0.9737473833915566,
-      "eval_recall": 0.9729,
-      "eval_runtime": 271.7966,
-      "eval_samples_per_second": 73.584,
-      "eval_steps_per_second": 9.198,
+      "epoch": 1.2929292929292928,
+      "eval_accuracy": 0.9794994605121188,
+      "eval_f1": 0.9793994612943837,
+      "eval_loss": 0.057158615440130234,
+      "eval_precision": 0.9800037878399945,
+      "eval_recall": 0.9794994605121188,
+      "eval_runtime": 451.5225,
+      "eval_samples_per_second": 84.16,
+      "eval_steps_per_second": 10.52,
       "step": 256
     },
     {
-      "epoch": 1.732484076433121,
-      "grad_norm": 5.855906009674072,
-      "learning_rate": 3.2929936305732486e-05,
-      "loss": 0.0579,
-      "step": 272
-    },
-    {
-      "epoch": 1.8343949044585988,
-      "grad_norm": 6.417220115661621,
-      "learning_rate": 3.191082802547771e-05,
-      "loss": 0.0656,
-      "step": 288
-    },
-    {
-      "epoch": 1.9363057324840764,
-      "grad_norm": 5.988188743591309,
-      "learning_rate": 3.089171974522293e-05,
-      "loss": 0.0536,
-      "step": 304
-    },
-    {
-      "epoch": 2.038216560509554,
-      "grad_norm": 11.449820518493652,
-      "learning_rate": 2.9872611464968155e-05,
-      "loss": 0.05,
+      "epoch": 1.6161616161616161,
+      "grad_norm": 7.240455627441406,
+      "learning_rate": 3.4040404040404045e-05,
+      "loss": 0.0439,
       "step": 320
     },
     {
-      "epoch": 2.140127388535032,
-      "grad_norm": 8.527064323425293,
-      "learning_rate": 2.8853503184713375e-05,
-      "loss": 0.0399,
-      "step": 336
-    },
-    {
-      "epoch": 2.2420382165605095,
-      "grad_norm": 5.836983680725098,
-      "learning_rate": 2.78343949044586e-05,
-      "loss": 0.0518,
-      "step": 352
-    },
-    {
-      "epoch": 2.343949044585987,
-      "grad_norm": 26.618896484375,
-      "learning_rate": 2.6815286624203823e-05,
-      "loss": 0.0372,
-      "step": 368
+      "epoch": 1.6161616161616161,
+      "eval_accuracy": 0.984394326166478,
+      "eval_f1": 0.984415214874111,
+      "eval_loss": 0.042214084416627884,
+      "eval_precision": 0.9843782986693216,
+      "eval_recall": 0.984394326166478,
+      "eval_runtime": 528.4934,
+      "eval_samples_per_second": 71.903,
+      "eval_steps_per_second": 8.988,
+      "step": 320
     },
     {
-      "epoch": 2.445859872611465,
-      "grad_norm": 8.455528259277344,
-      "learning_rate": 2.5796178343949047e-05,
-      "loss": 0.0432,
+      "epoch": 1.9393939393939394,
+      "grad_norm": 3.2924530506134033,
+      "learning_rate": 3.080808080808081e-05,
+      "loss": 0.0392,
       "step": 384
     },
     {
-      "epoch": 2.445859872611465,
-      "eval_accuracy": 0.98205,
-      "eval_f1": 0.9819928803546507,
-      "eval_loss": 0.05750665441155434,
-      "eval_precision": 0.9823073850700162,
-      "eval_recall": 0.98205,
-      "eval_runtime": 276.5019,
-      "eval_samples_per_second": 72.332,
-      "eval_steps_per_second": 9.042,
+      "epoch": 1.9393939393939394,
+      "eval_accuracy": 0.9802626384904866,
+      "eval_f1": 0.9801401584145893,
+      "eval_loss": 0.05642416328191757,
+      "eval_precision": 0.980989991407714,
+      "eval_recall": 0.9802626384904866,
+      "eval_runtime": 541.8272,
+      "eval_samples_per_second": 70.133,
+      "eval_steps_per_second": 8.767,
       "step": 384
     },
     {
-      "epoch": 2.5477707006369426,
-      "grad_norm": 15.16251277923584,
-      "learning_rate": 2.477707006369427e-05,
-      "loss": 0.0592,
-      "step": 400
-    },
-    {
-      "epoch": 2.6496815286624202,
-      "grad_norm": 5.268926620483398,
-      "learning_rate": 2.375796178343949e-05,
-      "loss": 0.0323,
-      "step": 416
-    },
-    {
-      "epoch": 2.7515923566878984,
-      "grad_norm": 5.213953018188477,
-      "learning_rate": 2.2738853503184715e-05,
-      "loss": 0.0247,
-      "step": 432
-    },
-    {
-      "epoch": 2.853503184713376,
-      "grad_norm": 4.638315677642822,
-      "learning_rate": 2.171974522292994e-05,
-      "loss": 0.0306,
+      "epoch": 2.2626262626262625,
+      "grad_norm": 5.49556827545166,
+      "learning_rate": 2.7575757575757578e-05,
+      "loss": 0.0374,
       "step": 448
     },
     {
-      "epoch": 2.9554140127388537,
-      "grad_norm": 6.414985179901123,
-      "learning_rate": 2.0700636942675162e-05,
-      "loss": 0.0302,
-      "step": 464
-    },
-    {
-      "epoch": 3.0573248407643314,
-      "grad_norm": 4.792835235595703,
-      "learning_rate": 1.9681528662420383e-05,
-      "loss": 0.0239,
-      "step": 480
-    },
-    {
-      "epoch": 3.159235668789809,
-      "grad_norm": 3.82145357131958,
-      "learning_rate": 1.8662420382165603e-05,
-      "loss": 0.016,
-      "step": 496
+      "epoch": 2.2626262626262625,
+      "eval_accuracy": 0.9837100976341483,
+      "eval_f1": 0.9837088988305677,
+      "eval_loss": 0.04635660722851753,
+      "eval_precision": 0.983712570988981,
+      "eval_recall": 0.9837100976341483,
+      "eval_runtime": 523.4635,
+      "eval_samples_per_second": 72.593,
+      "eval_steps_per_second": 9.074,
+      "step": 448
     },
     {
-      "epoch": 3.261146496815287,
-      "grad_norm": 6.392252445220947,
-      "learning_rate": 1.7643312101910827e-05,
-      "loss": 0.0247,
+      "epoch": 2.5858585858585856,
+      "grad_norm": 7.576784133911133,
+      "learning_rate": 2.4343434343434344e-05,
+      "loss": 0.0273,
       "step": 512
     },
     {
-      "epoch": 3.261146496815287,
-      "eval_accuracy": 0.9832,
-      "eval_f1": 0.9831797846161753,
-      "eval_loss": 0.05071697756648064,
-      "eval_precision": 0.983261693079801,
-      "eval_recall": 0.9832,
-      "eval_runtime": 271.679,
-      "eval_samples_per_second": 73.616,
-      "eval_steps_per_second": 9.202,
+      "epoch": 2.5858585858585856,
+      "eval_accuracy": 0.9861048974973026,
+      "eval_f1": 0.9861281800402698,
+      "eval_loss": 0.03779837116599083,
+      "eval_precision": 0.9860990328572948,
+      "eval_recall": 0.9861048974973026,
+      "eval_runtime": 526.8186,
+      "eval_samples_per_second": 72.131,
+      "eval_steps_per_second": 9.016,
       "step": 512
     },
     {
-      "epoch": 3.3630573248407645,
-      "grad_norm": 7.162614345550537,
-      "learning_rate": 1.662420382165605e-05,
-      "loss": 0.0135,
-      "step": 528
-    },
-    {
-      "epoch": 3.464968152866242,
-      "grad_norm": 3.0641708374023438,
-      "learning_rate": 1.5605095541401275e-05,
-      "loss": 0.0172,
-      "step": 544
-    },
-    {
-      "epoch": 3.56687898089172,
-      "grad_norm": 11.316327095031738,
-      "learning_rate": 1.4585987261146497e-05,
-      "loss": 0.0129,
-      "step": 560
-    },
-    {
-      "epoch": 3.6687898089171975,
-      "grad_norm": 9.987749099731445,
-      "learning_rate": 1.356687898089172e-05,
-      "loss": 0.0087,
+      "epoch": 2.909090909090909,
+      "grad_norm": 8.414101600646973,
+      "learning_rate": 2.111111111111111e-05,
+      "loss": 0.0271,
       "step": 576
     },
     {
-      "epoch": 3.770700636942675,
-      "grad_norm": 1.408227562904358,
-      "learning_rate": 1.2547770700636943e-05,
-      "loss": 0.0145,
-      "step": 592
-    },
-    {
-      "epoch": 3.872611464968153,
-      "grad_norm": 4.08975076675415,
-      "learning_rate": 1.1528662420382167e-05,
-      "loss": 0.0169,
-      "step": 608
-    },
-    {
-      "epoch": 3.9745222929936306,
-      "grad_norm": 11.745902061462402,
-      "learning_rate": 1.0509554140127389e-05,
-      "loss": 0.0122,
-      "step": 624
+      "epoch": 2.909090909090909,
+      "eval_accuracy": 0.9883417984683808,
+      "eval_f1": 0.9883335392334167,
+      "eval_loss": 0.03357573598623276,
+      "eval_precision": 0.988365298180521,
+      "eval_recall": 0.9883417984683808,
+      "eval_runtime": 528.7169,
+      "eval_samples_per_second": 71.872,
+      "eval_steps_per_second": 8.984,
+      "step": 576
     },
     {
-      "epoch": 4.076433121019108,
-      "grad_norm": 1.0680924654006958,
-      "learning_rate": 9.490445859872613e-06,
-      "loss": 0.0115,
+      "epoch": 3.2323232323232323,
+      "grad_norm": 4.41958475112915,
+      "learning_rate": 1.787878787878788e-05,
+      "loss": 0.021,
       "step": 640
     },
     {
-      "epoch": 4.076433121019108,
-      "eval_accuracy": 0.98645,
-      "eval_f1": 0.9864145632656558,
-      "eval_loss": 0.053590141236782074,
-      "eval_precision": 0.9866191209109798,
-      "eval_recall": 0.98645,
-      "eval_runtime": 270.7038,
-      "eval_samples_per_second": 73.881,
-      "eval_steps_per_second": 9.235,
+      "epoch": 3.2323232323232323,
+      "eval_accuracy": 0.985894365641201,
+      "eval_f1": 0.9859268508760648,
+      "eval_loss": 0.04180537536740303,
+      "eval_precision": 0.9859071918052413,
+      "eval_recall": 0.985894365641201,
+      "eval_runtime": 523.9915,
+      "eval_samples_per_second": 72.52,
+      "eval_steps_per_second": 9.065,
       "step": 640
     },
     {
-      "epoch": 4.178343949044586,
-      "grad_norm": 7.269558429718018,
-      "learning_rate": 8.471337579617835e-06,
-      "loss": 0.0092,
-      "step": 656
+      "epoch": 3.5555555555555554,
+      "grad_norm": 10.126025199890137,
+      "learning_rate": 1.4646464646464647e-05,
+      "loss": 0.019,
+      "step": 704
     },
     {
-      "epoch": 4.280254777070064,
-      "grad_norm": 4.04400110244751,
-      "learning_rate": 7.452229299363057e-06,
-      "loss": 0.0065,
-      "step": 672
+      "epoch": 3.5555555555555554,
+      "eval_accuracy": 0.9848417063606937,
+      "eval_f1": 0.9848982600089184,
+      "eval_loss": 0.04540397599339485,
+      "eval_precision": 0.9849383014159727,
+      "eval_recall": 0.9848417063606937,
+      "eval_runtime": 525.5117,
+      "eval_samples_per_second": 72.31,
+      "eval_steps_per_second": 9.039,
+      "step": 704
     },
     {
-      "epoch": 4.382165605095541,
-      "grad_norm": 0.2636276185512543,
-      "learning_rate": 6.433121019108281e-06,
-      "loss": 0.0028,
-      "step": 688
+      "epoch": 3.878787878787879,
+      "grad_norm": 3.757537603378296,
+      "learning_rate": 1.1414141414141415e-05,
+      "loss": 0.0177,
+      "step": 768
     },
     {
-      "epoch": 4.484076433121019,
-      "grad_norm": 11.116532325744629,
-      "learning_rate": 5.414012738853504e-06,
-      "loss": 0.0033,
-      "step": 704
+      "epoch": 3.878787878787879,
+      "eval_accuracy": 0.9883154819863681,
+      "eval_f1": 0.9883223194841958,
+      "eval_loss": 0.03590450435876846,
+      "eval_precision": 0.9883066841617396,
+      "eval_recall": 0.9883154819863681,
+      "eval_runtime": 522.9339,
+      "eval_samples_per_second": 72.667,
+      "eval_steps_per_second": 9.083,
+      "step": 768
     },
     {
-      "epoch": 4.585987261146497,
-      "grad_norm": 0.07053454965353012,
-      "learning_rate": 4.394904458598726e-06,
-      "loss": 0.0023,
-      "step": 720
+      "epoch": 4.202020202020202,
+      "grad_norm": 4.959455966949463,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 0.0134,
+      "step": 832
     },
     {
-      "epoch": 4.687898089171974,
-      "grad_norm": 2.811331033706665,
-      "learning_rate": 3.375796178343949e-06,
-      "loss": 0.0057,
-      "step": 736
+      "epoch": 4.202020202020202,
+      "eval_accuracy": 0.9874470380799495,
+      "eval_f1": 0.987400014046709,
+      "eval_loss": 0.04097573831677437,
+      "eval_precision": 0.9877259518894549,
+      "eval_recall": 0.9874470380799495,
+      "eval_runtime": 521.4997,
+      "eval_samples_per_second": 72.867,
+      "eval_steps_per_second": 9.108,
+      "step": 832
     },
     {
-      "epoch": 4.789808917197452,
-      "grad_norm": 0.1817609667778015,
-      "learning_rate": 2.3566878980891724e-06,
-      "loss": 0.0043,
-      "step": 752
+      "epoch": 4.525252525252525,
+      "grad_norm": 3.8119397163391113,
+      "learning_rate": 4.949494949494949e-06,
+      "loss": 0.0102,
+      "step": 896
     },
     {
-      "epoch": 4.89171974522293,
-      "grad_norm": 0.5368472933769226,
-      "learning_rate": 1.337579617834395e-06,
-      "loss": 0.002,
-      "step": 768
+      "epoch": 4.525252525252525,
+      "eval_accuracy": 0.9909997631516619,
+      "eval_f1": 0.9910005190250708,
+      "eval_loss": 0.031363558024168015,
+      "eval_precision": 0.9909983512838277,
+      "eval_recall": 0.9909997631516619,
+      "eval_runtime": 523.3329,
+      "eval_samples_per_second": 72.612,
+      "eval_steps_per_second": 9.076,
+      "step": 896
     },
     {
-      "epoch": 4.89171974522293,
-      "eval_accuracy": 0.98685,
-      "eval_f1": 0.9868193085416297,
-      "eval_loss": 0.06709764152765274,
-      "eval_precision": 0.9869900769633103,
-      "eval_recall": 0.98685,
-      "eval_runtime": 275.4886,
-      "eval_samples_per_second": 72.598,
-      "eval_steps_per_second": 9.075,
-      "step": 768
+      "epoch": 4.848484848484849,
+      "grad_norm": 3.6070823669433594,
+      "learning_rate": 1.7171717171717171e-06,
+      "loss": 0.0103,
+      "step": 960
     },
     {
-      "epoch": 4.993630573248407,
-      "grad_norm": 2.216209888458252,
-      "learning_rate": 3.1847133757961787e-07,
-      "loss": 0.0012,
-      "step": 784
+      "epoch": 4.848484848484849,
+      "eval_accuracy": 0.9909997631516619,
+      "eval_f1": 0.9909972985219008,
+      "eval_loss": 0.030054476112127304,
+      "eval_precision": 0.9910057473707247,
+      "eval_recall": 0.9909997631516619,
+      "eval_runtime": 534.5024,
+      "eval_samples_per_second": 71.094,
+      "eval_steps_per_second": 8.887,
+      "step": 960
     },
     {
       "epoch": 5.0,
-      "step": 785,
-      "total_flos": 1.9922182705152e+18,
-      "train_loss": 0.11038042506926758,
-      "train_runtime": 7101.5946,
-      "train_samples_per_second": 56.325,
-      "train_steps_per_second": 0.111
+      "step": 990,
+      "total_flos": 3.78521471397888e+18,
+      "train_loss": 0.055960290434986654,
+      "train_runtime": 18290.5686,
+      "train_samples_per_second": 41.551,
+      "train_steps_per_second": 0.054
     }
   ],
-  "logging_steps": 16,
-  "max_steps": 785,
+  "logging_steps": 64,
+  "max_steps": 990,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 5,
-  "save_steps": 128,
+  "save_steps": 64,
   "stateful_callbacks": {
     "TrainerControl": {
       "args": {
@@ -450,8 +320,8 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.9922182705152e+18,
-  "train_batch_size": 512,
+  "total_flos": 3.78521471397888e+18,
+  "train_batch_size": 768,
   "trial_name": null,
   "trial_params": null
 }