update model with 1.0 safeguard

checkpoint-80000/trainer_state.json

@@ -18,9 +18,9 @@
       "eval_bleu": 38.1466,
       "eval_gen_len": 37.2387,
       "eval_loss": 0.1091395914554596,
-      "eval_runtime": 171.3952,
-      "eval_samples_per_second": 6.085,
-      "eval_steps_per_second": 0.193,
+      "eval_runtime": 169.6196,
+      "eval_samples_per_second": 6.149,
+      "eval_steps_per_second": 0.195,
       "step": 4000
     },
     {
@@ -34,9 +34,9 @@
       "eval_bleu": 40.8678,
       "eval_gen_len": 34.2282,
       "eval_loss": 0.1069067046046257,
-      "eval_runtime": 198.8833,
-      "eval_samples_per_second": 5.244,
-      "eval_steps_per_second": 0.166,
+      "eval_runtime": 187.3165,
+      "eval_samples_per_second": 5.568,
+      "eval_steps_per_second": 0.176,
       "step": 8000
     },
     {
@@ -50,9 +50,9 @@
       "eval_bleu": 41.2373,
       "eval_gen_len": 34.722,
       "eval_loss": 0.10556207597255707,
-      "eval_runtime": 189.0135,
-      "eval_samples_per_second": 5.518,
-      "eval_steps_per_second": 0.175,
+      "eval_runtime": 176.6792,
+      "eval_samples_per_second": 5.903,
+      "eval_steps_per_second": 0.187,
       "step": 12000
     },
     {
@@ -66,9 +66,9 @@
       "eval_bleu": 41.2236,
       "eval_gen_len": 35.4794,
       "eval_loss": 0.10456866770982742,
-      "eval_runtime": 136.4617,
-      "eval_samples_per_second": 7.643,
-      "eval_steps_per_second": 0.242,
+      "eval_runtime": 126.4686,
+      "eval_samples_per_second": 8.247,
+      "eval_steps_per_second": 0.261,
       "step": 16000
     },
     {
@@ -82,9 +82,9 @@
       "eval_bleu": 41.6373,
       "eval_gen_len": 35.6874,
       "eval_loss": 0.10391870141029358,
-      "eval_runtime": 153.8741,
-      "eval_samples_per_second": 6.778,
-      "eval_steps_per_second": 0.214,
+      "eval_runtime": 137.3482,
+      "eval_samples_per_second": 7.594,
+      "eval_steps_per_second": 0.24,
       "step": 20000
     },
     {
@@ -98,9 +98,9 @@
       "eval_bleu": 41.2704,
       "eval_gen_len": 35.7728,
       "eval_loss": 0.10389306396245956,
-      "eval_runtime": 144.2866,
-      "eval_samples_per_second": 7.229,
-      "eval_steps_per_second": 0.229,
+      "eval_runtime": 129.1131,
+      "eval_samples_per_second": 8.078,
+      "eval_steps_per_second": 0.256,
       "step": 24000
     },
     {
@@ -114,9 +114,9 @@
       "eval_bleu": 41.6866,
       "eval_gen_len": 35.348,
       "eval_loss": 0.10308653861284256,
-      "eval_runtime": 156.8578,
-      "eval_samples_per_second": 6.649,
-      "eval_steps_per_second": 0.21,
+      "eval_runtime": 149.5149,
+      "eval_samples_per_second": 6.976,
+      "eval_steps_per_second": 0.221,
       "step": 28000
     },
     {
@@ -130,9 +130,9 @@
       "eval_bleu": 41.3994,
       "eval_gen_len": 35.7287,
       "eval_loss": 0.10239578038454056,
-      "eval_runtime": 126.4909,
-      "eval_samples_per_second": 8.246,
-      "eval_steps_per_second": 0.261,
+      "eval_runtime": 125.8581,
+      "eval_samples_per_second": 8.287,
+      "eval_steps_per_second": 0.262,
       "step": 32000
     },
     {
@@ -146,9 +146,9 @@
       "eval_bleu": 39.5845,
       "eval_gen_len": 39.464,
       "eval_loss": 0.102129265666008,
-      "eval_runtime": 129.762,
-      "eval_samples_per_second": 8.038,
-      "eval_steps_per_second": 0.254,
+      "eval_runtime": 123.946,
+      "eval_samples_per_second": 8.415,
+      "eval_steps_per_second": 0.266,
       "step": 36000
     },
     {
@@ -162,9 +162,9 @@
       "eval_bleu": 42.2251,
       "eval_gen_len": 34.3845,
       "eval_loss": 0.10222680866718292,
-      "eval_runtime": 106.5884,
-      "eval_samples_per_second": 9.785,
-      "eval_steps_per_second": 0.31,
+      "eval_runtime": 101.5149,
+      "eval_samples_per_second": 10.274,
+      "eval_steps_per_second": 0.325,
       "step": 40000
     },
     {
@@ -178,9 +178,9 @@
       "eval_bleu": 41.9432,
       "eval_gen_len": 35.0316,
       "eval_loss": 0.10267173498868942,
-      "eval_runtime": 129.8219,
-      "eval_samples_per_second": 8.034,
-      "eval_steps_per_second": 0.254,
+      "eval_runtime": 129.0759,
+      "eval_samples_per_second": 8.081,
+      "eval_steps_per_second": 0.256,
       "step": 44000
     },
     {
@@ -194,9 +194,9 @@
       "eval_bleu": 42.028,
       "eval_gen_len": 35.8677,
       "eval_loss": 0.102423757314682,
-      "eval_runtime": 130.0386,
-      "eval_samples_per_second": 8.021,
-      "eval_steps_per_second": 0.254,
+      "eval_runtime": 128.7526,
+      "eval_samples_per_second": 8.101,
+      "eval_steps_per_second": 0.256,
       "step": 48000
     },
     {
@@ -210,9 +210,9 @@
       "eval_bleu": 41.9107,
       "eval_gen_len": 35.6884,
       "eval_loss": 0.1020052582025528,
-      "eval_runtime": 105.0208,
-      "eval_samples_per_second": 9.931,
-      "eval_steps_per_second": 0.314,
+      "eval_runtime": 98.4633,
+      "eval_samples_per_second": 10.593,
+      "eval_steps_per_second": 0.335,
       "step": 52000
     },
     {
@@ -226,9 +226,9 @@
       "eval_bleu": 41.898,
       "eval_gen_len": 35.7948,
       "eval_loss": 0.10190638899803162,
-      "eval_runtime": 154.2922,
-      "eval_samples_per_second": 6.76,
-      "eval_steps_per_second": 0.214,
+      "eval_runtime": 155.172,
+      "eval_samples_per_second": 6.722,
+      "eval_steps_per_second": 0.213,
       "step": 56000
     },
     {
@@ -242,9 +242,9 @@
       "eval_bleu": 42.4405,
       "eval_gen_len": 34.8159,
       "eval_loss": 0.10122686624526978,
-      "eval_runtime": 118.3439,
-      "eval_samples_per_second": 8.813,
-      "eval_steps_per_second": 0.279,
+      "eval_runtime": 109.5432,
+      "eval_samples_per_second": 9.521,
+      "eval_steps_per_second": 0.301,
       "step": 60000
     },
     {
@@ -258,9 +258,9 @@
       "eval_bleu": 42.1656,
       "eval_gen_len": 35.1611,
       "eval_loss": 0.1020784005522728,
-      "eval_runtime": 134.6489,
-      "eval_samples_per_second": 7.746,
-      "eval_steps_per_second": 0.245,
+      "eval_runtime": 127.0613,
+      "eval_samples_per_second": 8.209,
+      "eval_steps_per_second": 0.26,
       "step": 64000
     },
     {
@@ -274,9 +274,9 @@
       "eval_bleu": 42.2236,
       "eval_gen_len": 34.6663,
       "eval_loss": 0.10225772112607956,
-      "eval_runtime": 127.5545,
-      "eval_samples_per_second": 8.177,
-      "eval_steps_per_second": 0.259,
+      "eval_runtime": 118.3874,
+      "eval_samples_per_second": 8.81,
+      "eval_steps_per_second": 0.279,
       "step": 68000
     },
     {
@@ -290,9 +290,9 @@
       "eval_bleu": 42.1593,
       "eval_gen_len": 34.5043,
       "eval_loss": 0.10208923369646072,
-      "eval_runtime": 88.1144,
-      "eval_samples_per_second": 11.837,
-      "eval_steps_per_second": 0.375,
+      "eval_runtime": 84.9237,
+      "eval_samples_per_second": 12.282,
+      "eval_steps_per_second": 0.389,
       "step": 72000
     },
     {
@@ -306,9 +306,9 @@
       "eval_bleu": 42.0667,
       "eval_gen_len": 35.1371,
       "eval_loss": 0.10195324569940567,
-      "eval_runtime": 120.6638,
-      "eval_samples_per_second": 8.644,
-      "eval_steps_per_second": 0.273,
+      "eval_runtime": 115.0559,
+      "eval_samples_per_second": 9.065,
+      "eval_steps_per_second": 0.287,
       "step": 76000
     },
     {
@@ -322,9 +322,9 @@
       "eval_bleu": 42.7449,
       "eval_gen_len": 35.2848,
       "eval_loss": 0.10178153961896896,
-      "eval_runtime": 116.145,
-      "eval_samples_per_second": 8.98,
-      "eval_steps_per_second": 0.284,
+      "eval_runtime": 108.8465,
+      "eval_samples_per_second": 9.582,
+      "eval_steps_per_second": 0.303,
       "step": 80000
     }
   ],

checkpoint-80000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:038425ce924f4f9135467ce2b852d43fabd43f14b098635a051c844c20458e63
+oid sha256:3ffc182f2e76a1a072a43bb175e1b6d8176501f76e795c82d01e5ab1d2f0a83c
 size 3771

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:038425ce924f4f9135467ce2b852d43fabd43f14b098635a051c844c20458e63
+oid sha256:3ffc182f2e76a1a072a43bb175e1b6d8176501f76e795c82d01e5ab1d2f0a83c
 size 3771