update model with 1.0 safeguard

checkpoint-80000/trainer_state.json

@@ -18,9 +18,9 @@
       "eval_bleu": 38.0976,
       "eval_gen_len": 37.2953,
       "eval_loss": 0.10893864184617996,
-      "eval_runtime": 163.5371,
-      "eval_samples_per_second": 6.378,
-      "eval_steps_per_second": 0.202,
+      "eval_runtime": 170.9124,
+      "eval_samples_per_second": 6.103,
+      "eval_steps_per_second": 0.193,
       "step": 4000
     },
     {
@@ -34,9 +34,9 @@
       "eval_bleu": 36.8776,
       "eval_gen_len": 39.7383,
       "eval_loss": 0.10664400458335876,
-      "eval_runtime": 149.6664,
-      "eval_samples_per_second": 6.969,
-      "eval_steps_per_second": 0.22,
+      "eval_runtime": 160.2997,
+      "eval_samples_per_second": 6.507,
+      "eval_steps_per_second": 0.206,
       "step": 8000
     },
     {
@@ -50,9 +50,9 @@
       "eval_bleu": 38.837,
       "eval_gen_len": 37.2205,
       "eval_loss": 0.1052979975938797,
-      "eval_runtime": 159.5413,
-      "eval_samples_per_second": 6.537,
-      "eval_steps_per_second": 0.207,
+      "eval_runtime": 164.3263,
+      "eval_samples_per_second": 6.347,
+      "eval_steps_per_second": 0.201,
       "step": 12000
     },
     {
@@ -66,9 +66,9 @@
       "eval_bleu": 41.3418,
       "eval_gen_len": 34.976,
       "eval_loss": 0.10423395782709122,
-      "eval_runtime": 118.5509,
-      "eval_samples_per_second": 8.798,
-      "eval_steps_per_second": 0.278,
+      "eval_runtime": 130.7322,
+      "eval_samples_per_second": 7.978,
+      "eval_steps_per_second": 0.252,
       "step": 16000
     },
     {
@@ -82,9 +82,9 @@
       "eval_bleu": 41.4557,
       "eval_gen_len": 36.4477,
       "eval_loss": 0.10369615256786346,
-      "eval_runtime": 132.3921,
-      "eval_samples_per_second": 7.878,
-      "eval_steps_per_second": 0.249,
+      "eval_runtime": 144.8968,
+      "eval_samples_per_second": 7.198,
+      "eval_steps_per_second": 0.228,
       "step": 20000
     },
     {
@@ -98,9 +98,9 @@
       "eval_bleu": 42.0105,
       "eval_gen_len": 35.0681,
       "eval_loss": 0.10366526991128922,
-      "eval_runtime": 129.295,
-      "eval_samples_per_second": 8.067,
-      "eval_steps_per_second": 0.255,
+      "eval_runtime": 140.889,
+      "eval_samples_per_second": 7.403,
+      "eval_steps_per_second": 0.234,
       "step": 24000
     },
     {
@@ -114,9 +114,9 @@
       "eval_bleu": 41.9312,
       "eval_gen_len": 35.4621,
       "eval_loss": 0.1028067097067833,
-      "eval_runtime": 119.6638,
-      "eval_samples_per_second": 8.716,
-      "eval_steps_per_second": 0.276,
+      "eval_runtime": 131.7196,
+      "eval_samples_per_second": 7.918,
+      "eval_steps_per_second": 0.251,
       "step": 28000
     },
     {
@@ -130,9 +130,9 @@
       "eval_bleu": 41.2791,
       "eval_gen_len": 36.2924,
       "eval_loss": 0.10221964865922928,
-      "eval_runtime": 141.8273,
-      "eval_samples_per_second": 7.354,
-      "eval_steps_per_second": 0.233,
+      "eval_runtime": 151.9451,
+      "eval_samples_per_second": 6.864,
+      "eval_steps_per_second": 0.217,
       "step": 32000
     },
     {
@@ -146,9 +146,9 @@
       "eval_bleu": 40.1286,
       "eval_gen_len": 38.2963,
       "eval_loss": 0.10198042541742325,
-      "eval_runtime": 134.6203,
-      "eval_samples_per_second": 7.748,
-      "eval_steps_per_second": 0.245,
+      "eval_runtime": 144.9715,
+      "eval_samples_per_second": 7.195,
+      "eval_steps_per_second": 0.228,
       "step": 36000
     },
     {
@@ -162,9 +162,9 @@
       "eval_bleu": 42.345,
       "eval_gen_len": 34.3241,
       "eval_loss": 0.10196036845445633,
-      "eval_runtime": 107.632,
-      "eval_samples_per_second": 9.69,
-      "eval_steps_per_second": 0.307,
+      "eval_runtime": 118.6741,
+      "eval_samples_per_second": 8.789,
+      "eval_steps_per_second": 0.278,
       "step": 40000
     },
     {
@@ -178,9 +178,9 @@
       "eval_bleu": 42.1017,
       "eval_gen_len": 35.5551,
       "eval_loss": 0.10244914144277573,
-      "eval_runtime": 122.4649,
-      "eval_samples_per_second": 8.517,
-      "eval_steps_per_second": 0.269,
+      "eval_runtime": 129.7053,
+      "eval_samples_per_second": 8.041,
+      "eval_steps_per_second": 0.254,
       "step": 44000
     },
     {
@@ -194,9 +194,9 @@
       "eval_bleu": 41.5599,
       "eval_gen_len": 36.302,
       "eval_loss": 0.10217992961406708,
-      "eval_runtime": 135.1924,
-      "eval_samples_per_second": 7.715,
-      "eval_steps_per_second": 0.244,
+      "eval_runtime": 139.5859,
+      "eval_samples_per_second": 7.472,
+      "eval_steps_per_second": 0.236,
       "step": 48000
     },
     {
@@ -210,9 +210,9 @@
       "eval_bleu": 42.1594,
       "eval_gen_len": 34.8734,
       "eval_loss": 0.10176843404769897,
-      "eval_runtime": 130.1772,
-      "eval_samples_per_second": 8.012,
-      "eval_steps_per_second": 0.254,
+      "eval_runtime": 143.2813,
+      "eval_samples_per_second": 7.279,
+      "eval_steps_per_second": 0.23,
       "step": 52000
     },
     {
@@ -226,9 +226,9 @@
       "eval_bleu": 42.3226,
       "eval_gen_len": 35.5129,
       "eval_loss": 0.10162311047315598,
-      "eval_runtime": 151.058,
-      "eval_samples_per_second": 6.905,
-      "eval_steps_per_second": 0.218,
+      "eval_runtime": 155.5764,
+      "eval_samples_per_second": 6.704,
+      "eval_steps_per_second": 0.212,
       "step": 56000
     },
     {
@@ -242,9 +242,9 @@
       "eval_bleu": 42.76,
       "eval_gen_len": 34.7421,
       "eval_loss": 0.10091908276081085,
-      "eval_runtime": 141.7745,
-      "eval_samples_per_second": 7.357,
-      "eval_steps_per_second": 0.233,
+      "eval_runtime": 150.4127,
+      "eval_samples_per_second": 6.934,
+      "eval_steps_per_second": 0.219,
       "step": 60000
     },
     {
@@ -258,9 +258,9 @@
       "eval_bleu": 42.0603,
       "eval_gen_len": 34.3337,
       "eval_loss": 0.10173938423395157,
-      "eval_runtime": 118.5645,
-      "eval_samples_per_second": 8.797,
-      "eval_steps_per_second": 0.278,
+      "eval_runtime": 116.9913,
+      "eval_samples_per_second": 8.915,
+      "eval_steps_per_second": 0.282,
       "step": 64000
     },
     {
@@ -274,9 +274,9 @@
       "eval_bleu": 42.6668,
       "eval_gen_len": 34.0575,
       "eval_loss": 0.10194776207208633,
-      "eval_runtime": 106.9067,
-      "eval_samples_per_second": 9.756,
-      "eval_steps_per_second": 0.309,
+      "eval_runtime": 115.372,
+      "eval_samples_per_second": 9.04,
+      "eval_steps_per_second": 0.286,
       "step": 68000
     },
     {
@@ -290,9 +290,9 @@
       "eval_bleu": 42.3805,
       "eval_gen_len": 34.7498,
       "eval_loss": 0.1018945649266243,
-      "eval_runtime": 86.6094,
-      "eval_samples_per_second": 12.043,
-      "eval_steps_per_second": 0.381,
+      "eval_runtime": 87.7456,
+      "eval_samples_per_second": 11.887,
+      "eval_steps_per_second": 0.376,
       "step": 72000
     },
     {
@@ -306,9 +306,9 @@
       "eval_bleu": 42.3891,
       "eval_gen_len": 35.1179,
       "eval_loss": 0.10174746811389923,
-      "eval_runtime": 108.0214,
-      "eval_samples_per_second": 9.655,
-      "eval_steps_per_second": 0.305,
+      "eval_runtime": 123.341,
+      "eval_samples_per_second": 8.456,
+      "eval_steps_per_second": 0.268,
       "step": 76000
     },
     {
@@ -322,9 +322,9 @@
       "eval_bleu": 42.8456,
       "eval_gen_len": 34.791,
       "eval_loss": 0.1016170084476471,
-      "eval_runtime": 104.9692,
-      "eval_samples_per_second": 9.936,
-      "eval_steps_per_second": 0.314,
+      "eval_runtime": 121.3922,
+      "eval_samples_per_second": 8.592,
+      "eval_steps_per_second": 0.272,
       "step": 80000
     }
   ],

checkpoint-80000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:75348bcfc056147819c47ff6d4e14e03f359f11d3c4f02f59a51bfa89c291c59
+oid sha256:34fec518c9aaf6e1dcbf7cbe9131e5915cb9c8225a5a430b145445a64c83b5ca
 size 3771

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:75348bcfc056147819c47ff6d4e14e03f359f11d3c4f02f59a51bfa89c291c59
+oid sha256:34fec518c9aaf6e1dcbf7cbe9131e5915cb9c8225a5a430b145445a64c83b5ca
 size 3771