Training in progress, step 10000

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2ef929787429c88c576570a6fdf928468aedd3cf5c685150c4d00c23ed2574b5
+oid sha256:c205aa1e4f5bca6eeb78422e0866131d9cc58c5a15fab7af01cc2505a790df94
 size 100170757

last-checkpoint/pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1b649041f4e942a1c71886ca3f83f11fc846bb4752950f41b84cd207e3cc556c
+oid sha256:f48f41b7b489ac24545b446eb3e28cf98daf67da888804a7513071641e3cb0ab
 size 146774203

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9e521fc0b39ea1a8ce6e0a681bc8a834da87e83fd7cd3754aacc6c4b558e0937
-size 14439
+oid sha256:8b635e00d41e3e6e46449e8609b1f91fbb1205e9587df7d15d6fa3ee2cfe466d
+size 14503

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:677543783c6139b35f47c70ebab6ad516807591d9ad24cec274ace85b8305217
+oid sha256:97030c307e5479014866fb832070891e4b01f3e0d70ac54668f29d2602b645e8
 size 246897640

last-checkpoint/trainer_state.json

@@ -1,916 +1,316 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.1289213579716373,
-  "global_step": 30000,
+  "epoch": 0.042973785990545764,
+  "global_step": 10000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 0.00040004211081201384,
-      "loss": 8.3496,
+      "learning_rate": 4.0004211081201405e-05,
+      "loss": 3.9197,
       "step": 200
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.000400168442509171,
-      "loss": 8.2272,
+      "learning_rate": 4.0016844250917146e-05,
+      "loss": 3.9108,
       "step": 400
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.000400378992874836,
-      "loss": 7.6879,
+      "learning_rate": 4.003789928748371e-05,
+      "loss": 3.8655,
       "step": 600
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.0004006737582146567,
-      "loss": 7.4747,
+      "learning_rate": 4.006737582146571e-05,
+      "loss": 3.8602,
       "step": 800
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.0004010527333566261,
-      "loss": 7.2829,
+      "learning_rate": 4.010527333566261e-05,
+      "loss": 3.8478,
       "step": 1000
     },
     {
       "epoch": 0.01,
-      "learning_rate": 0.0004015159116511832,
-      "loss": 7.1171,
+      "learning_rate": 4.0151591165118474e-05,
+      "loss": 3.8608,
       "step": 1200
     },
     {
       "epoch": 0.01,
-      "learning_rate": 0.00040206328497132196,
-      "loss": 6.9445,
+      "learning_rate": 4.020632849713237e-05,
+      "loss": 3.8431,
       "step": 1400
     },
     {
       "epoch": 0.01,
-      "learning_rate": 0.0004026948437127389,
-      "loss": 6.8391,
+      "learning_rate": 4.0269484371273996e-05,
+      "loss": 3.8447,
       "step": 1600
     },
     {
       "epoch": 0.01,
-      "learning_rate": 0.0004034105767939909,
-      "loss": 6.7131,
+      "learning_rate": 4.034105767939918e-05,
+      "loss": 3.8422,
       "step": 1800
     },
     {
       "epoch": 0.01,
-      "learning_rate": 0.00040421047165670534,
-      "loss": 6.6113,
+      "learning_rate": 4.0421047165670686e-05,
+      "loss": 3.8019,
       "step": 2000
     },
     {
       "epoch": 0.01,
-      "learning_rate": 0.0004050945142657896,
-      "loss": 6.4966,
+      "learning_rate": 4.050945142657905e-05,
+      "loss": 3.8109,
       "step": 2200
     },
     {
       "epoch": 0.01,
-      "learning_rate": 0.0004060626891096795,
-      "loss": 6.3979,
+      "learning_rate": 4.0606268910968035e-05,
+      "loss": 3.8192,
       "step": 2400
     },
     {
       "epoch": 0.01,
-      "learning_rate": 0.0004071149792006148,
-      "loss": 6.3116,
+      "learning_rate": 4.071149792006154e-05,
+      "loss": 3.8037,
       "step": 2600
     },
     {
       "epoch": 0.01,
-      "learning_rate": 0.00040825136607492915,
-      "loss": 6.2301,
+      "learning_rate": 4.082513660749298e-05,
+      "loss": 3.8074,
       "step": 2800
     },
     {
       "epoch": 0.01,
-      "learning_rate": 0.0004094718297933883,
-      "loss": 6.123,
+      "learning_rate": 4.0947182979338874e-05,
+      "loss": 3.801,
       "step": 3000
     },
     {
       "epoch": 0.01,
-      "learning_rate": 0.0004107763489415231,
-      "loss": 6.0802,
+      "learning_rate": 4.1077634894152375e-05,
+      "loss": 3.786,
       "step": 3200
     },
     {
       "epoch": 0.01,
-      "learning_rate": 0.00041216490063001633,
-      "loss": 6.0029,
+      "learning_rate": 4.121649006300163e-05,
+      "loss": 3.7799,
       "step": 3400
     },
     {
       "epoch": 0.02,
-      "learning_rate": 0.00041363746049510354,
-      "loss": 5.9471,
+      "learning_rate": 4.136374604951046e-05,
+      "loss": 3.7721,
       "step": 3600
     },
     {
       "epoch": 0.02,
-      "learning_rate": 0.0004151940026989945,
-      "loss": 5.9132,
+      "learning_rate": 4.1519400269899535e-05,
+      "loss": 3.7829,
       "step": 3800
     },
     {
       "epoch": 0.02,
-      "learning_rate": 0.0004168344999303346,
-      "loss": 5.8561,
+      "learning_rate": 4.1683449993033545e-05,
+      "loss": 3.7922,
       "step": 4000
     },
     {
       "epoch": 0.02,
-      "learning_rate": 0.00041855892340467854,
-      "loss": 5.8044,
+      "learning_rate": 4.185589234046783e-05,
+      "loss": 3.7686,
       "step": 4200
     },
     {
       "epoch": 0.02,
-      "learning_rate": 0.0004203672428649916,
-      "loss": 5.734,
+      "learning_rate": 4.203672428649923e-05,
+      "loss": 3.7742,
       "step": 4400
     },
     {
       "epoch": 0.02,
-      "learning_rate": 0.0004222594265821944,
-      "loss": 5.7245,
+      "learning_rate": 4.2225942658219505e-05,
+      "loss": 3.753,
       "step": 4600
     },
     {
       "epoch": 0.02,
-      "learning_rate": 0.0004242354413557057,
-      "loss": 5.6867,
+      "learning_rate": 4.242354413557057e-05,
+      "loss": 3.7663,
       "step": 4800
     },
     {
       "epoch": 0.02,
-      "learning_rate": 0.00042629525251402893,
-      "loss": 5.6387,
+      "learning_rate": 4.262952525140289e-05,
+      "loss": 3.7589,
       "step": 5000
     },
     {
       "epoch": 0.02,
-      "learning_rate": 0.0004284388239153662,
-      "loss": 5.6119,
+      "learning_rate": 4.2843882391536666e-05,
+      "loss": 3.7652,
       "step": 5200
     },
     {
       "epoch": 0.02,
-      "learning_rate": 0.0004306661179482429,
-      "loss": 5.5533,
+      "learning_rate": 4.306661179482435e-05,
+      "loss": 3.7618,
       "step": 5400
     },
     {
       "epoch": 0.02,
-      "learning_rate": 0.0004329770955321787,
-      "loss": 5.517,
+      "learning_rate": 4.329770955321802e-05,
+      "loss": 3.7441,
       "step": 5600
     },
     {
       "epoch": 0.02,
-      "learning_rate": 0.0004353717161183629,
-      "loss": 5.4864,
+      "learning_rate": 4.3537171611836355e-05,
+      "loss": 3.7364,
       "step": 5800
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.0004378499376903721,
-      "loss": 5.4671,
+      "learning_rate": 4.3784993769037384e-05,
+      "loss": 3.7501,
       "step": 6000
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.00044041171676490604,
-      "loss": 5.4412,
+      "learning_rate": 4.404117167649071e-05,
+      "loss": 3.7363,
       "step": 6200
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.0004430570083925455,
-      "loss": 5.4108,
+      "learning_rate": 4.43057008392547e-05,
+      "loss": 3.7279,
       "step": 6400
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.0004457857661585539,
-      "loss": 5.3807,
+      "learning_rate": 4.457857661585541e-05,
+      "loss": 3.7158,
       "step": 6600
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.0004485979421836768,
-      "loss": 5.3353,
+      "learning_rate": 4.485979421836779e-05,
+      "loss": 3.74,
       "step": 6800
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.0004514934871249904,
-      "loss": 5.3277,
+      "learning_rate": 4.514934871249906e-05,
+      "loss": 3.7441,
       "step": 7000
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.00045447235017676696,
-      "loss": 5.2979,
+      "learning_rate": 4.544723501767687e-05,
+      "loss": 3.7141,
       "step": 7200
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.00045753447907136494,
-      "loss": 5.2791,
+      "learning_rate": 4.575344790713656e-05,
+      "loss": 3.7264,
       "step": 7400
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.000460679820080143,
-      "loss": 5.2494,
+      "learning_rate": 4.6067982008014406e-05,
+      "loss": 3.7221,
       "step": 7600
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.00046390831801440893,
-      "loss": 5.2175,
+      "learning_rate": 4.639083180144098e-05,
+      "loss": 3.7363,
       "step": 7800
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.0004672199162263843,
-      "loss": 5.2038,
+      "learning_rate": 4.672199162263839e-05,
+      "loss": 3.73,
       "step": 8000
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.0004706145566101966,
-      "loss": 5.1835,
+      "learning_rate": 4.706145566101981e-05,
+      "loss": 3.7205,
       "step": 8200
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.0004740921796029061,
-      "loss": 5.1691,
+      "learning_rate": 4.740921796029065e-05,
+      "loss": 3.7192,
       "step": 8400
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.0004776527241855382,
-      "loss": 5.1582,
+      "learning_rate": 4.776527241855393e-05,
+      "loss": 3.7095,
       "step": 8600
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.0004812961278841711,
-      "loss": 5.1504,
+      "learning_rate": 4.812961278841707e-05,
+      "loss": 3.7311,
       "step": 8800
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.0004850223267710129,
-      "loss": 5.1162,
+      "learning_rate": 4.85022326771014e-05,
+      "loss": 3.7031,
       "step": 9000
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.0004888312554655432,
-      "loss": 5.0957,
+      "learning_rate": 4.888312554655438e-05,
+      "loss": 3.6934,
       "step": 9200
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.0004927228471356421,
-      "loss": 5.079,
+      "learning_rate": 4.92722847135643e-05,
+      "loss": 3.7218,
       "step": 9400
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.0004966970334987757,
-      "loss": 5.0572,
+      "learning_rate": 4.9669703349877704e-05,
+      "loss": 3.711,
       "step": 9600
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.0005007537448231871,
-      "loss": 5.0342,
+      "learning_rate": 5.0075374482318863e-05,
+      "loss": 3.7002,
       "step": 9800
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.0005048929099291249,
-      "loss": 5.0106,
+      "learning_rate": 5.0489290992912625e-05,
+      "loss": 3.698,
       "step": 10000
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 0.0005091144561900837,
-      "loss": 5.0155,
-      "step": 10200
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 0.0005134183095340927,
-      "loss": 4.9817,
-      "step": 10400
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 0.0005178043944449977,
-      "loss": 4.9742,
-      "step": 10600
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 0.0005222726339638023,
-      "loss": 4.9299,
-      "step": 10800
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 0.0005268229496900086,
-      "loss": 4.9208,
-      "step": 11000
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 0.0005314552617829947,
-      "loss": 4.8617,
-      "step": 11200
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 0.0005361694889634196,
-      "loss": 4.7952,
-      "step": 11400
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 0.0005409655485146408,
-      "loss": 4.7641,
-      "step": 11600
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 0.0005458433562841782,
-      "loss": 4.7361,
-      "step": 11800
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 0.0005508028266851747,
-      "loss": 4.7023,
-      "step": 12000
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 0.000555843872697916,
-      "loss": 4.6561,
-      "step": 12200
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 0.0005609664058713396,
-      "loss": 4.63,
-      "step": 12400
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 0.0005661703363245996,
-      "loss": 4.6307,
-      "step": 12600
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 0.0005714555727486404,
-      "loss": 4.5881,
-      "step": 12800
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 0.0005768220224077955,
-      "loss": 4.5489,
-      "step": 13000
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 0.0005822695911414169,
-      "loss": 4.5521,
-      "step": 13200
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 0.0005877981833655298,
-      "loss": 4.5165,
-      "step": 13400
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 0.0005934077020745051,
-      "loss": 4.505,
-      "step": 13600
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 0.0005990980488427659,
-      "loss": 4.4863,
-      "step": 13800
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 0.000604869123826509,
-      "loss": 4.5071,
-      "step": 14000
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 0.0006107208257654633,
-      "loss": 4.4501,
-      "step": 14200
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 0.0006166530519846631,
-      "loss": 4.4623,
-      "step": 14400
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 0.0006226656983962468,
-      "loss": 4.4336,
-      "step": 14600
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 0.0006287586595012887,
-      "loss": 4.4335,
-      "step": 14800
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 0.000634931828391647,
-      "loss": 4.4142,
-      "step": 15000
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 0.0006411850967518416,
-      "loss": 4.4145,
-      "step": 15200
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 0.0006475183548609511,
-      "loss": 4.3842,
-      "step": 15400
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 0.0006539314915945428,
-      "loss": 4.3748,
-      "step": 15600
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 0.0006604243944266178,
-      "loss": 4.3815,
-      "step": 15800
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 0.0006669969494315867,
-      "loss": 4.352,
-      "step": 16000
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 0.0006736490412862749,
-      "loss": 4.3575,
-      "step": 16200
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 0.000680380553271933,
-      "loss": 4.3416,
-      "step": 16400
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 0.0006871913672762998,
-      "loss": 4.341,
-      "step": 16600
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 0.0006940813637956594,
-      "loss": 4.3183,
-      "step": 16800
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 0.0007010504219369541,
-      "loss": 4.3145,
-      "step": 17000
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 0.0007080984194198885,
-      "loss": 4.3065,
-      "step": 17200
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 0.0007152252325790948,
-      "loss": 4.2805,
-      "step": 17400
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 0.0007224307363662818,
-      "loss": 4.2804,
-      "step": 17600
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 0.0007297148043524434,
-      "loss": 4.2996,
-      "step": 17800
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 0.0007370773087300737,
-      "loss": 4.2743,
-      "step": 18000
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 0.0007445181203154048,
-      "loss": 4.2621,
-      "step": 18200
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 0.0007520371085506811,
-      "loss": 4.2548,
-      "step": 18400
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 0.0007596341415064441,
-      "loss": 4.2643,
-      "step": 18600
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 0.0007673090858838494,
-      "loss": 4.266,
-      "step": 18800
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 0.0007750618070170041,
-      "loss": 4.2503,
-      "step": 19000
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 0.0007828921688753324,
-      "loss": 4.2093,
-      "step": 19200
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 0.0007908000340659631,
-      "loss": 4.2449,
-      "step": 19400
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 0.0007987852638361333,
-      "loss": 4.2158,
-      "step": 19600
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 0.0008068477180756314,
-      "loss": 4.202,
-      "step": 19800
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 0.0008149872553192515,
-      "loss": 4.2065,
-      "step": 20000
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 0.0008232037327492777,
-      "loss": 4.1773,
-      "step": 20200
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 0.0008314970061979818,
-      "loss": 4.1904,
-      "step": 20400
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 0.0008398669301501703,
-      "loss": 4.1868,
-      "step": 20600
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 0.0008483133577457148,
-      "loss": 4.2006,
-      "step": 20800
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 0.0008568361407821495,
-      "loss": 4.1467,
-      "step": 21000
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 0.0008654351297172607,
-      "loss": 4.1585,
-      "step": 21200
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 0.0008741101736717116,
-      "loss": 4.1547,
-      "step": 21400
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 0.0008828611204316911,
-      "loss": 4.1557,
-      "step": 21600
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 0.0008916878164515838,
-      "loss": 4.1496,
-      "step": 21800
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 0.0009005901068566691,
-      "loss": 4.1434,
-      "step": 22000
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 0.0009095678354458306,
-      "loss": 4.1173,
-      "step": 22200
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 0.0009186208446943008,
-      "loss": 4.1364,
-      "step": 22400
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 0.0009277489757564244,
-      "loss": 4.1445,
-      "step": 22600
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 0.0009369520684684475,
-      "loss": 4.1156,
-      "step": 22800
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 0.0009462299613513248,
-      "loss": 4.1033,
-      "step": 23000
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 0.0009555824916135536,
-      "loss": 4.1187,
-      "step": 23200
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 0.0009650094951540386,
-      "loss": 4.0823,
-      "step": 23400
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 0.0009745108065649499,
-      "loss": 4.0624,
-      "step": 23600
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 0.0009840862591346498,
-      "loss": 4.0845,
-      "step": 23800
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 0.0009937356848506058,
-      "loss": 4.0483,
-      "step": 24000
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 0.001003458914402332,
-      "loss": 4.0512,
-      "step": 24200
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 0.0010132557771843787,
-      "loss": 4.0606,
-      "step": 24400
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 0.0010231261012993067,
-      "loss": 4.046,
-      "step": 24600
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 0.0010330697135607168,
-      "loss": 4.0315,
-      "step": 24800
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 0.00104308643949628,
-      "loss": 4.0179,
-      "step": 25000
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 0.001053176103350803,
-      "loss": 4.0351,
-      "step": 25200
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 0.0010633385280893123,
-      "loss": 4.02,
-      "step": 25400
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 0.0010735735354001595,
-      "loss": 4.0201,
-      "step": 25600
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 0.0010838809456981471,
-      "loss": 4.0044,
-      "step": 25800
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 0.001094260578127686,
-      "loss": 3.9914,
-      "step": 26000
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 0.0011047122505659646,
-      "loss": 3.9991,
-      "step": 26200
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 0.0011152357796261423,
-      "loss": 4.0109,
-      "step": 26400
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 0.0011258309806605731,
-      "loss": 4.0405,
-      "step": 26600
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 0.0011364976677640387,
-      "loss": 4.0349,
-      "step": 26800
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 0.0011472356537770186,
-      "loss": 4.0312,
-      "step": 27000
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 0.0011580447502889633,
-      "loss": 4.0185,
-      "step": 27200
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 0.0011689247676416152,
-      "loss": 4.011,
-      "step": 27400
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 0.0011798755149323176,
-      "loss": 3.9898,
-      "step": 27600
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 0.001190896800017379,
-      "loss": 3.981,
-      "step": 27800
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 0.0012019884295154416,
-      "loss": 3.949,
-      "step": 28000
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 0.0012131502088108658,
-      "loss": 3.9896,
-      "step": 28200
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 0.0012243819420571598,
-      "loss": 3.9951,
-      "step": 28400
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 0.0012356834321804039,
-      "loss": 4.0361,
-      "step": 28600
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 0.0012470544808827113,
-      "loss": 4.1212,
-      "step": 28800
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 0.001258494888645708,
-      "loss": 4.0721,
-      "step": 29000
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 0.0012700044547340368,
-      "loss": 4.0311,
-      "step": 29200
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 0.0012815829771988738,
-      "loss": 4.0114,
-      "step": 29400
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 0.001293230252881479,
-      "loss": 3.9868,
-      "step": 29600
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 0.0013049460774167514,
-      "loss": 3.9881,
-      "step": 29800
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 0.0013167302452368242,
-      "loss": 3.9705,
-      "step": 30000
     }
   ],
   "max_steps": 500000,
   "num_train_epochs": 3,
-  "total_flos": 4.781489946624e+16,
+  "total_flos": 1.593829982208e+16,
   "trial_name": null,
   "trial_params": null
 }

last-checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3b76655be47d6fa9534e270ebad89f26359197ddd96685592630dba92c71f881
+oid sha256:db1575e3c0c4f23842bbe1ba840257aad501aecba3d1800f59d7014cd73b0cee
 size 3375

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1b649041f4e942a1c71886ca3f83f11fc846bb4752950f41b84cd207e3cc556c
+oid sha256:f48f41b7b489ac24545b446eb3e28cf98daf67da888804a7513071641e3cb0ab
 size 146774203

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3b76655be47d6fa9534e270ebad89f26359197ddd96685592630dba92c71f881
+oid sha256:db1575e3c0c4f23842bbe1ba840257aad501aecba3d1800f59d7014cd73b0cee
 size 3375