Model save

trainer_state.json

@@ -10,175 +10,175 @@
   "log_history": [
     {
       "epoch": 0.04,
-      "grad_norm": 3.7410537349353388,
+      "grad_norm": 2.97260032828452,
       "learning_rate": 2e-05,
       "loss": 3.2188,
       "step": 1
     },
     {
       "epoch": 0.08,
-      "grad_norm": 3.724477625112808,
+      "grad_norm": 3.594459617966852,
       "learning_rate": 2e-05,
       "loss": 3.2031,
       "step": 2
     },
     {
       "epoch": 0.12,
-      "grad_norm": 4.847707450779626,
+      "grad_norm": 4.533624625545066,
       "learning_rate": 2e-05,
       "loss": 3.4062,
       "step": 3
     },
     {
       "epoch": 0.16,
-      "grad_norm": 5.609117941274957,
+      "grad_norm": 5.240356703299346,
       "learning_rate": 2e-05,
       "loss": 3.3906,
       "step": 4
     },
     {
       "epoch": 0.2,
-      "grad_norm": 5.086957233893569,
+      "grad_norm": 5.560506359364048,
       "learning_rate": 2e-05,
       "loss": 3.0156,
       "step": 5
     },
     {
       "epoch": 0.24,
-      "grad_norm": 5.227277254140017,
+      "grad_norm": 4.893164995962358,
       "learning_rate": 2e-05,
       "loss": 3.4844,
       "step": 6
     },
     {
       "epoch": 0.28,
-      "grad_norm": 4.351893594691178,
+      "grad_norm": 3.9500718110988577,
       "learning_rate": 2e-05,
       "loss": 3.3438,
       "step": 7
     },
     {
       "epoch": 0.32,
-      "grad_norm": 4.143312345927856,
+      "grad_norm": 3.769524663807585,
       "learning_rate": 2e-05,
       "loss": 3.2656,
       "step": 8
     },
     {
       "epoch": 0.36,
-      "grad_norm": 5.095207786632339,
+      "grad_norm": 4.641711022263081,
       "learning_rate": 2e-05,
       "loss": 3.5625,
       "step": 9
     },
     {
       "epoch": 0.4,
-      "grad_norm": 5.292856607844618,
+      "grad_norm": 5.045996826645826,
       "learning_rate": 2e-05,
       "loss": 3.3906,
       "step": 10
     },
     {
       "epoch": 0.44,
-      "grad_norm": 3.9969754196981433,
+      "grad_norm": 3.9379891091293446,
       "learning_rate": 2e-05,
       "loss": 3.5,
       "step": 11
     },
     {
       "epoch": 0.48,
-      "grad_norm": 6.349497527343764,
+      "grad_norm": 5.935094879674494,
       "learning_rate": 2e-05,
-      "loss": 3.0469,
+      "loss": 3.0312,
       "step": 12
     },
     {
       "epoch": 0.52,
-      "grad_norm": 5.2510974905297,
+      "grad_norm": 4.966274615752918,
       "learning_rate": 2e-05,
       "loss": 3.2031,
       "step": 13
     },
     {
       "epoch": 0.56,
-      "grad_norm": 4.472101838754883,
+      "grad_norm": 4.115242373102183,
       "learning_rate": 2e-05,
       "loss": 3.1875,
       "step": 14
     },
     {
       "epoch": 0.6,
-      "grad_norm": 4.03317126613206,
+      "grad_norm": 3.8744700920947444,
       "learning_rate": 2e-05,
       "loss": 3.3906,
       "step": 15
     },
     {
       "epoch": 0.64,
-      "grad_norm": 3.1679359694764315,
+      "grad_norm": 3.1817540324181763,
       "learning_rate": 2e-05,
       "loss": 3.2656,
       "step": 16
     },
     {
       "epoch": 0.68,
-      "grad_norm": 3.2017940861615557,
+      "grad_norm": 3.2254117611390924,
       "learning_rate": 2e-05,
       "loss": 3.0781,
       "step": 17
     },
     {
       "epoch": 0.72,
-      "grad_norm": 4.001252869474549,
+      "grad_norm": 4.082116872038495,
       "learning_rate": 2e-05,
       "loss": 3.3906,
       "step": 18
     },
     {
       "epoch": 0.76,
-      "grad_norm": 5.765014522502107,
+      "grad_norm": 5.513160685623603,
       "learning_rate": 2e-05,
       "loss": 3.0469,
       "step": 19
     },
     {
       "epoch": 0.8,
-      "grad_norm": 5.4796650093828685,
+      "grad_norm": 5.2179385595071155,
       "learning_rate": 2e-05,
       "loss": 3.2031,
       "step": 20
     },
     {
       "epoch": 0.84,
-      "grad_norm": 5.915029810230642,
+      "grad_norm": 5.4603287380267,
       "learning_rate": 2e-05,
       "loss": 3.5312,
       "step": 21
     },
     {
       "epoch": 0.88,
-      "grad_norm": 4.775032664581898,
+      "grad_norm": 4.596875242645009,
       "learning_rate": 2e-05,
       "loss": 3.0469,
       "step": 22
     },
     {
       "epoch": 0.92,
-      "grad_norm": 4.653673104315309,
+      "grad_norm": 4.546647529822756,
       "learning_rate": 2e-05,
       "loss": 3.0469,
       "step": 23
     },
     {
       "epoch": 0.96,
-      "grad_norm": 5.417622869264991,
+      "grad_norm": 5.146569623435821,
       "learning_rate": 2e-05,
       "loss": 3.2344,
       "step": 24
     },
     {
       "epoch": 1.0,
-      "grad_norm": 7.2444031678746645,
+      "grad_norm": 6.727524092488977,
       "learning_rate": 2e-05,
       "loss": 3.1562,
       "step": 25
@@ -187,10 +187,10 @@
       "epoch": 1.0,
       "step": 25,
       "total_flos": 478414897152.0,
-      "train_loss": 3.264375,
-      "train_runtime": 88.1114,
-      "train_samples_per_second": 2.236,
-      "train_steps_per_second": 0.284
+      "train_loss": 3.26375,
+      "train_runtime": 93.9312,
+      "train_samples_per_second": 2.097,
+      "train_steps_per_second": 0.266
     }
   ],
   "logging_steps": 1.0,