update model with 1.0 safeguard

checkpoint-80000/trainer_state.json

@@ -18,9 +18,9 @@
       "eval_bleu": 35.5597,
       "eval_gen_len": 39.8495,
       "eval_loss": 0.10906314104795456,
-      "eval_runtime": 172.1137,
-      "eval_samples_per_second": 6.06,
-      "eval_steps_per_second": 0.192,
+      "eval_runtime": 140.7392,
+      "eval_samples_per_second": 7.411,
+      "eval_steps_per_second": 0.234,
       "step": 4000
     },
     {
@@ -34,9 +34,9 @@
       "eval_bleu": 40.9319,
       "eval_gen_len": 34.6337,
       "eval_loss": 0.10657753050327301,
-      "eval_runtime": 155.9008,
-      "eval_samples_per_second": 6.69,
-      "eval_steps_per_second": 0.212,
+      "eval_runtime": 125.8016,
+      "eval_samples_per_second": 8.291,
+      "eval_steps_per_second": 0.262,
       "step": 8000
     },
     {
@@ -50,9 +50,9 @@
       "eval_bleu": 41.3844,
       "eval_gen_len": 34.6453,
       "eval_loss": 0.10533162206411362,
-      "eval_runtime": 169.7915,
-      "eval_samples_per_second": 6.143,
-      "eval_steps_per_second": 0.194,
+      "eval_runtime": 111.0041,
+      "eval_samples_per_second": 9.396,
+      "eval_steps_per_second": 0.297,
       "step": 12000
     },
     {
@@ -66,9 +66,9 @@
       "eval_bleu": 41.3616,
       "eval_gen_len": 35.1601,
       "eval_loss": 0.10429085791110992,
-      "eval_runtime": 173.7331,
-      "eval_samples_per_second": 6.003,
-      "eval_steps_per_second": 0.19,
+      "eval_runtime": 141.5674,
+      "eval_samples_per_second": 7.368,
+      "eval_steps_per_second": 0.233,
       "step": 16000
     },
     {
@@ -82,9 +82,9 @@
       "eval_bleu": 41.6835,
       "eval_gen_len": 34.9942,
       "eval_loss": 0.10373106598854065,
-      "eval_runtime": 138.2922,
-      "eval_samples_per_second": 7.542,
-      "eval_steps_per_second": 0.239,
+      "eval_runtime": 109.7842,
+      "eval_samples_per_second": 9.5,
+      "eval_steps_per_second": 0.301,
       "step": 20000
     },
     {
@@ -98,9 +98,9 @@
       "eval_bleu": 41.1613,
       "eval_gen_len": 36.2349,
       "eval_loss": 0.10356967151165009,
-      "eval_runtime": 161.9993,
-      "eval_samples_per_second": 6.438,
-      "eval_steps_per_second": 0.204,
+      "eval_runtime": 116.302,
+      "eval_samples_per_second": 8.968,
+      "eval_steps_per_second": 0.284,
       "step": 24000
     },
     {
@@ -114,9 +114,9 @@
       "eval_bleu": 41.8822,
       "eval_gen_len": 34.4698,
       "eval_loss": 0.10275906324386597,
-      "eval_runtime": 113.2908,
-      "eval_samples_per_second": 9.206,
-      "eval_steps_per_second": 0.291,
+      "eval_runtime": 88.2743,
+      "eval_samples_per_second": 11.815,
+      "eval_steps_per_second": 0.374,
       "step": 28000
     },
     {
@@ -130,9 +130,9 @@
       "eval_bleu": 41.7092,
       "eval_gen_len": 35.2943,
       "eval_loss": 0.10221899300813675,
-      "eval_runtime": 118.7545,
-      "eval_samples_per_second": 8.783,
-      "eval_steps_per_second": 0.278,
+      "eval_runtime": 93.7664,
+      "eval_samples_per_second": 11.123,
+      "eval_steps_per_second": 0.352,
       "step": 32000
     },
     {
@@ -146,9 +146,9 @@
       "eval_bleu": 40.1745,
       "eval_gen_len": 38.5484,
       "eval_loss": 0.10177244991064072,
-      "eval_runtime": 118.1882,
-      "eval_samples_per_second": 8.825,
-      "eval_steps_per_second": 0.279,
+      "eval_runtime": 92.869,
+      "eval_samples_per_second": 11.231,
+      "eval_steps_per_second": 0.355,
       "step": 36000
     },
     {
@@ -162,9 +162,9 @@
       "eval_bleu": 42.5078,
       "eval_gen_len": 34.3826,
       "eval_loss": 0.10191857814788818,
-      "eval_runtime": 109.0756,
-      "eval_samples_per_second": 9.562,
-      "eval_steps_per_second": 0.303,
+      "eval_runtime": 85.4817,
+      "eval_samples_per_second": 12.201,
+      "eval_steps_per_second": 0.386,
       "step": 40000
     },
     {
@@ -178,9 +178,9 @@
       "eval_bleu": 42.0937,
       "eval_gen_len": 35.7114,
       "eval_loss": 0.10230503231287003,
-      "eval_runtime": 120.9618,
-      "eval_samples_per_second": 8.623,
-      "eval_steps_per_second": 0.273,
+      "eval_runtime": 96.9439,
+      "eval_samples_per_second": 10.759,
+      "eval_steps_per_second": 0.34,
       "step": 44000
     },
     {
@@ -194,9 +194,9 @@
       "eval_bleu": 42.3842,
       "eval_gen_len": 35.3586,
       "eval_loss": 0.10215254127979279,
-      "eval_runtime": 145.2894,
-      "eval_samples_per_second": 7.179,
-      "eval_steps_per_second": 0.227,
+      "eval_runtime": 94.9854,
+      "eval_samples_per_second": 10.981,
+      "eval_steps_per_second": 0.347,
       "step": 48000
     },
     {
@@ -210,9 +210,9 @@
       "eval_bleu": 42.1993,
       "eval_gen_len": 34.0729,
       "eval_loss": 0.10184619575738907,
-      "eval_runtime": 129.3624,
-      "eval_samples_per_second": 8.063,
-      "eval_steps_per_second": 0.255,
+      "eval_runtime": 104.2027,
+      "eval_samples_per_second": 10.009,
+      "eval_steps_per_second": 0.317,
       "step": 52000
     },
     {
@@ -226,9 +226,9 @@
       "eval_bleu": 42.3348,
       "eval_gen_len": 35.5158,
       "eval_loss": 0.10167574882507324,
-      "eval_runtime": 118.7033,
-      "eval_samples_per_second": 8.787,
-      "eval_steps_per_second": 0.278,
+      "eval_runtime": 95.3873,
+      "eval_samples_per_second": 10.934,
+      "eval_steps_per_second": 0.346,
       "step": 56000
     },
     {
@@ -242,9 +242,9 @@
       "eval_bleu": 42.6183,
       "eval_gen_len": 34.5225,
       "eval_loss": 0.10100951045751572,
-      "eval_runtime": 129.9979,
-      "eval_samples_per_second": 8.023,
-      "eval_steps_per_second": 0.254,
+      "eval_runtime": 105.9409,
+      "eval_samples_per_second": 9.845,
+      "eval_steps_per_second": 0.311,
       "step": 60000
     },
     {
@@ -258,9 +258,9 @@
       "eval_bleu": 42.354,
       "eval_gen_len": 35.1323,
       "eval_loss": 0.10181604325771332,
-      "eval_runtime": 112.8931,
-      "eval_samples_per_second": 9.239,
-      "eval_steps_per_second": 0.292,
+      "eval_runtime": 90.7174,
+      "eval_samples_per_second": 11.497,
+      "eval_steps_per_second": 0.364,
       "step": 64000
     },
     {
@@ -274,9 +274,9 @@
       "eval_bleu": 41.9485,
       "eval_gen_len": 35.4727,
       "eval_loss": 0.10201819986104965,
-      "eval_runtime": 152.1334,
-      "eval_samples_per_second": 6.856,
-      "eval_steps_per_second": 0.217,
+      "eval_runtime": 99.9879,
+      "eval_samples_per_second": 10.431,
+      "eval_steps_per_second": 0.33,
       "step": 68000
     },
     {
@@ -290,9 +290,9 @@
       "eval_bleu": 42.3042,
       "eval_gen_len": 35.2704,
       "eval_loss": 0.10195478051900864,
-      "eval_runtime": 87.1208,
-      "eval_samples_per_second": 11.972,
-      "eval_steps_per_second": 0.379,
+      "eval_runtime": 70.0936,
+      "eval_samples_per_second": 14.88,
+      "eval_steps_per_second": 0.471,
       "step": 72000
     },
     {
@@ -306,9 +306,9 @@
       "eval_bleu": 39.7039,
       "eval_gen_len": 37.767,
       "eval_loss": 0.10168451070785522,
-      "eval_runtime": 120.3909,
-      "eval_samples_per_second": 8.663,
-      "eval_steps_per_second": 0.274,
+      "eval_runtime": 112.9245,
+      "eval_samples_per_second": 9.236,
+      "eval_steps_per_second": 0.292,
       "step": 76000
     },
     {
@@ -322,9 +322,9 @@
       "eval_bleu": 42.9732,
       "eval_gen_len": 34.5542,
       "eval_loss": 0.10154784470796585,
-      "eval_runtime": 146.5985,
-      "eval_samples_per_second": 7.115,
-      "eval_steps_per_second": 0.225,
+      "eval_runtime": 147.6247,
+      "eval_samples_per_second": 7.065,
+      "eval_steps_per_second": 0.224,
       "step": 80000
     }
   ],

checkpoint-80000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0e05e0105e3d4b9c07cc62cac7d00f8128a8ae004ebec6214a4683123704e19a
+oid sha256:00dba6608fa8aeace5d4d7a700208219e0f9dc795287f71cf75349dfe01fd209
 size 3771

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0e05e0105e3d4b9c07cc62cac7d00f8128a8ae004ebec6214a4683123704e19a
+oid sha256:00dba6608fa8aeace5d4d7a700208219e0f9dc795287f71cf75349dfe01fd209
 size 3771