update model with 1.0 safeguard

checkpoint-80000/trainer_state.json

@@ -18,9 +18,9 @@
       "eval_bleu": 38.0264,
       "eval_gen_len": 37.3058,
       "eval_loss": 0.10926879942417145,
-      "eval_runtime": 144.5824,
-      "eval_samples_per_second": 7.214,
-      "eval_steps_per_second": 0.228,
+      "eval_runtime": 185.5179,
+      "eval_samples_per_second": 5.622,
+      "eval_steps_per_second": 0.178,
       "step": 4000
     },
     {
@@ -34,9 +34,9 @@
       "eval_bleu": 37.1545,
       "eval_gen_len": 38.1074,
       "eval_loss": 0.10711430013179779,
-      "eval_runtime": 126.6602,
-      "eval_samples_per_second": 8.235,
-      "eval_steps_per_second": 0.261,
+      "eval_runtime": 160.2258,
+      "eval_samples_per_second": 6.51,
+      "eval_steps_per_second": 0.206,
       "step": 8000
     },
     {
@@ -50,9 +50,9 @@
       "eval_bleu": 40.7576,
       "eval_gen_len": 35.7632,
       "eval_loss": 0.10568097978830338,
-      "eval_runtime": 140.0097,
-      "eval_samples_per_second": 7.449,
-      "eval_steps_per_second": 0.236,
+      "eval_runtime": 172.6094,
+      "eval_samples_per_second": 6.043,
+      "eval_steps_per_second": 0.191,
       "step": 12000
     },
     {
@@ -66,9 +66,9 @@
       "eval_bleu": 41.1354,
       "eval_gen_len": 36.836,
       "eval_loss": 0.10463867336511612,
-      "eval_runtime": 115.2796,
-      "eval_samples_per_second": 9.048,
-      "eval_steps_per_second": 0.286,
+      "eval_runtime": 149.0366,
+      "eval_samples_per_second": 6.998,
+      "eval_steps_per_second": 0.221,
       "step": 16000
     },
     {
@@ -82,9 +82,9 @@
       "eval_bleu": 41.1989,
       "eval_gen_len": 36.8955,
       "eval_loss": 0.10405203700065613,
-      "eval_runtime": 112.8843,
-      "eval_samples_per_second": 9.24,
-      "eval_steps_per_second": 0.292,
+      "eval_runtime": 142.4153,
+      "eval_samples_per_second": 7.324,
+      "eval_steps_per_second": 0.232,
       "step": 20000
     },
     {
@@ -98,9 +98,9 @@
       "eval_bleu": 41.8132,
       "eval_gen_len": 34.7565,
       "eval_loss": 0.10406998544931412,
-      "eval_runtime": 80.9256,
-      "eval_samples_per_second": 12.888,
-      "eval_steps_per_second": 0.408,
+      "eval_runtime": 107.8168,
+      "eval_samples_per_second": 9.674,
+      "eval_steps_per_second": 0.306,
       "step": 24000
     },
     {
@@ -114,9 +114,9 @@
       "eval_bleu": 41.8593,
       "eval_gen_len": 35.1361,
       "eval_loss": 0.10320573300123215,
-      "eval_runtime": 123.7907,
-      "eval_samples_per_second": 8.426,
-      "eval_steps_per_second": 0.267,
+      "eval_runtime": 152.0565,
+      "eval_samples_per_second": 6.859,
+      "eval_steps_per_second": 0.217,
       "step": 28000
     },
     {
@@ -130,9 +130,9 @@
       "eval_bleu": 41.806,
       "eval_gen_len": 35.8591,
       "eval_loss": 0.10261942446231842,
-      "eval_runtime": 119.4076,
-      "eval_samples_per_second": 8.735,
-      "eval_steps_per_second": 0.276,
+      "eval_runtime": 154.646,
+      "eval_samples_per_second": 6.744,
+      "eval_steps_per_second": 0.213,
       "step": 32000
     },
     {
@@ -146,9 +146,9 @@
       "eval_bleu": 41.7632,
       "eval_gen_len": 36.094,
       "eval_loss": 0.10236303508281708,
-      "eval_runtime": 114.5766,
-      "eval_samples_per_second": 9.103,
-      "eval_steps_per_second": 0.288,
+      "eval_runtime": 142.1476,
+      "eval_samples_per_second": 7.337,
+      "eval_steps_per_second": 0.232,
       "step": 36000
     },
     {
@@ -162,9 +162,9 @@
       "eval_bleu": 41.9832,
       "eval_gen_len": 35.2454,
       "eval_loss": 0.10227732360363007,
-      "eval_runtime": 85.7342,
-      "eval_samples_per_second": 12.166,
-      "eval_steps_per_second": 0.385,
+      "eval_runtime": 113.1327,
+      "eval_samples_per_second": 9.219,
+      "eval_steps_per_second": 0.292,
       "step": 40000
     },
     {
@@ -178,9 +178,9 @@
       "eval_bleu": 41.4408,
       "eval_gen_len": 36.2848,
       "eval_loss": 0.10282173752784729,
-      "eval_runtime": 110.3637,
-      "eval_samples_per_second": 9.451,
-      "eval_steps_per_second": 0.299,
+      "eval_runtime": 139.7824,
+      "eval_samples_per_second": 7.462,
+      "eval_steps_per_second": 0.236,
       "step": 44000
     },
     {
@@ -194,9 +194,9 @@
       "eval_bleu": 41.1816,
       "eval_gen_len": 36.6194,
       "eval_loss": 0.10262637585401535,
-      "eval_runtime": 120.1566,
-      "eval_samples_per_second": 8.68,
-      "eval_steps_per_second": 0.275,
+      "eval_runtime": 153.9201,
+      "eval_samples_per_second": 6.776,
+      "eval_steps_per_second": 0.214,
       "step": 48000
     },
     {
@@ -210,9 +210,9 @@
       "eval_bleu": 41.7264,
       "eval_gen_len": 35.7603,
       "eval_loss": 0.10216823220252991,
-      "eval_runtime": 89.1094,
-      "eval_samples_per_second": 11.705,
-      "eval_steps_per_second": 0.37,
+      "eval_runtime": 120.762,
+      "eval_samples_per_second": 8.637,
+      "eval_steps_per_second": 0.273,
       "step": 52000
     },
     {
@@ -226,9 +226,9 @@
       "eval_bleu": 40.7372,
       "eval_gen_len": 37.8715,
       "eval_loss": 0.10200918465852737,
-      "eval_runtime": 134.6413,
-      "eval_samples_per_second": 7.747,
-      "eval_steps_per_second": 0.245,
+      "eval_runtime": 167.8456,
+      "eval_samples_per_second": 6.214,
+      "eval_steps_per_second": 0.197,
       "step": 56000
     },
     {
@@ -242,9 +242,9 @@
       "eval_bleu": 42.6732,
       "eval_gen_len": 34.7517,
       "eval_loss": 0.10124800354242325,
-      "eval_runtime": 111.3971,
-      "eval_samples_per_second": 9.363,
-      "eval_steps_per_second": 0.296,
+      "eval_runtime": 145.0169,
+      "eval_samples_per_second": 7.192,
+      "eval_steps_per_second": 0.228,
       "step": 60000
     },
     {
@@ -258,9 +258,9 @@
       "eval_bleu": 42.345,
       "eval_gen_len": 34.5609,
       "eval_loss": 0.10216101258993149,
-      "eval_runtime": 107.5188,
-      "eval_samples_per_second": 9.701,
-      "eval_steps_per_second": 0.307,
+      "eval_runtime": 136.5147,
+      "eval_samples_per_second": 7.64,
+      "eval_steps_per_second": 0.242,
       "step": 64000
     },
     {
@@ -274,9 +274,9 @@
       "eval_bleu": 42.6795,
       "eval_gen_len": 34.442,
       "eval_loss": 0.10226498544216156,
-      "eval_runtime": 101.1175,
-      "eval_samples_per_second": 10.315,
-      "eval_steps_per_second": 0.326,
+      "eval_runtime": 125.0786,
+      "eval_samples_per_second": 8.339,
+      "eval_steps_per_second": 0.264,
       "step": 68000
     },
     {
@@ -290,9 +290,9 @@
       "eval_bleu": 42.1965,
       "eval_gen_len": 34.9348,
       "eval_loss": 0.10218390822410583,
-      "eval_runtime": 80.2313,
-      "eval_samples_per_second": 13.0,
-      "eval_steps_per_second": 0.411,
+      "eval_runtime": 107.2258,
+      "eval_samples_per_second": 9.727,
+      "eval_steps_per_second": 0.308,
       "step": 72000
     },
     {
@@ -306,9 +306,9 @@
       "eval_bleu": 40.6959,
       "eval_gen_len": 38.1342,
       "eval_loss": 0.10209660232067108,
-      "eval_runtime": 181.0599,
-      "eval_samples_per_second": 5.761,
-      "eval_steps_per_second": 0.182,
+      "eval_runtime": 211.7076,
+      "eval_samples_per_second": 4.927,
+      "eval_steps_per_second": 0.156,
       "step": 76000
     },
     {
@@ -322,9 +322,9 @@
       "eval_bleu": 42.729,
       "eval_gen_len": 35.0566,
       "eval_loss": 0.10193432867527008,
-      "eval_runtime": 83.5606,
-      "eval_samples_per_second": 12.482,
-      "eval_steps_per_second": 0.395,
+      "eval_runtime": 104.8859,
+      "eval_samples_per_second": 9.944,
+      "eval_steps_per_second": 0.315,
       "step": 80000
     }
   ],

checkpoint-80000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:360679c7fad415b3c48425a6babc6418951bc55ca3821631fd52a04f2c42e833
+oid sha256:19b52f3d420b071910bee5599c056de4ab2b8a451358cbb9a4e7fdac32ab97c2
 size 3771

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:360679c7fad415b3c48425a6babc6418951bc55ca3821631fd52a04f2c42e833
+oid sha256:19b52f3d420b071910bee5599c056de4ab2b8a451358cbb9a4e7fdac32ab97c2
 size 3771