End of training

all_results.json

@@ -1,19 +1,19 @@
 {
     "epoch": 3.0,
     "eval_bleu": 0.2510920700175718,
-    "eval_loss": 2.5200681686401367,
-    "eval_rouge1": 0.45339177611687703,
-    "eval_rouge2": 0.21369428759421394,
-    "eval_rougeL": 0.39838760416988706,
-    "eval_runtime": 29.7946,
+    "eval_loss": 2.5200693607330322,
+    "eval_rouge1": 0.45339120041307046,
+    "eval_rouge2": 0.21369562389429725,
+    "eval_rougeL": 0.3983722007696566,
+    "eval_runtime": 28.0888,
     "eval_samples": 884,
-    "eval_samples_per_second": 29.67,
-    "eval_steps_per_second": 3.726,
-    "perplexity": 12.42944393298918,
+    "eval_samples_per_second": 31.472,
+    "eval_steps_per_second": 3.952,
+    "perplexity": 12.42945875004982,
     "total_flos": 4151799742464000.0,
-    "train_loss": 0.0,
-    "train_runtime": 0.0936,
+    "train_loss": 3.14277273785115,
+    "train_runtime": 586.188,
     "train_samples": 3531,
-    "train_samples_per_second": 113198.697,
-    "train_steps_per_second": 14169.874
+    "train_samples_per_second": 18.071,
+    "train_steps_per_second": 2.262
 }

egy_training_log.txt

@@ -844,3 +844,5 @@ INFO:root:Epoch 2.0: Train Loss = 4.0723, Eval Loss = 2.958087921142578
 INFO:absl:Using default tokenizer.
 INFO:root:Epoch 3.0: Train Loss = 2.8055, Eval Loss = 2.606330394744873
 INFO:absl:Using default tokenizer.
+INFO:__main__:*** Evaluate ***
+INFO:absl:Using default tokenizer.

eval_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 3.0,
     "eval_bleu": 0.2510920700175718,
-    "eval_loss": 2.5200681686401367,
-    "eval_rouge1": 0.45339177611687703,
-    "eval_rouge2": 0.21369428759421394,
-    "eval_rougeL": 0.39838760416988706,
-    "eval_runtime": 29.7946,
+    "eval_loss": 2.5200693607330322,
+    "eval_rouge1": 0.45339120041307046,
+    "eval_rouge2": 0.21369562389429725,
+    "eval_rougeL": 0.3983722007696566,
+    "eval_runtime": 28.0888,
     "eval_samples": 884,
-    "eval_samples_per_second": 29.67,
-    "eval_steps_per_second": 3.726,
-    "perplexity": 12.42944393298918
+    "eval_samples_per_second": 31.472,
+    "eval_steps_per_second": 3.952,
+    "perplexity": 12.42945875004982
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 3.0,
     "total_flos": 4151799742464000.0,
-    "train_loss": 0.0,
-    "train_runtime": 0.0936,
+    "train_loss": 3.14277273785115,
+    "train_runtime": 586.188,
     "train_samples": 3531,
-    "train_samples_per_second": 113198.697,
-    "train_steps_per_second": 14169.874
+    "train_samples_per_second": 18.071,
+    "train_steps_per_second": 2.262
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 2.5200681686401367,
+  "best_metric": 2.5200693607330322,
   "best_model_checkpoint": "/home/iais_marenpielka/Bouthaina/res_nw_egy/checkpoint-1326",
   "epoch": 3.0,
   "eval_steps": 500,
@@ -10,7 +10,7 @@
   "log_history": [
     {
       "epoch": 1.0,
-      "grad_norm": 1.6992121934890747,
+      "grad_norm": 1.6992123126983643,
       "learning_rate": 4.4200000000000004e-05,
       "loss": 4.0723,
       "step": 442
@@ -22,33 +22,33 @@
       "eval_rouge1": 0.3912678429308313,
       "eval_rouge2": 0.16481568948149655,
       "eval_rougeL": 0.32008248461173605,
-      "eval_runtime": 28.6854,
-      "eval_samples_per_second": 30.817,
-      "eval_steps_per_second": 3.87,
+      "eval_runtime": 27.8657,
+      "eval_samples_per_second": 31.724,
+      "eval_steps_per_second": 3.983,
       "step": 442
     },
     {
       "epoch": 2.0,
-      "grad_norm": 1.676450490951538,
+      "grad_norm": 1.676450252532959,
       "learning_rate": 2.6755447941888623e-05,
       "loss": 2.8055,
       "step": 884
     },
     {
       "epoch": 2.0,
-      "eval_bleu": 0.24154541958578588,
-      "eval_loss": 2.6063287258148193,
-      "eval_rouge1": 0.43470063350740673,
+      "eval_bleu": 0.24154624358915514,
+      "eval_loss": 2.606330394744873,
+      "eval_rouge1": 0.43470628711649895,
       "eval_rouge2": 0.19710102219495218,
-      "eval_rougeL": 0.3776728970262111,
-      "eval_runtime": 28.998,
-      "eval_samples_per_second": 30.485,
-      "eval_steps_per_second": 3.828,
+      "eval_rougeL": 0.3776869994743499,
+      "eval_runtime": 28.1779,
+      "eval_samples_per_second": 31.372,
+      "eval_steps_per_second": 3.939,
       "step": 884
     },
     {
       "epoch": 3.0,
-      "grad_norm": 1.5551185607910156,
+      "grad_norm": 1.5551155805587769,
       "learning_rate": 0.0,
       "loss": 2.5505,
       "step": 1326
@@ -56,23 +56,23 @@
     {
       "epoch": 3.0,
       "eval_bleu": 0.2510920700175718,
-      "eval_loss": 2.5200681686401367,
-      "eval_rouge1": 0.45339177611687703,
-      "eval_rouge2": 0.21369428759421394,
-      "eval_rougeL": 0.39838760416988706,
-      "eval_runtime": 29.0706,
-      "eval_samples_per_second": 30.409,
-      "eval_steps_per_second": 3.818,
+      "eval_loss": 2.5200693607330322,
+      "eval_rouge1": 0.45339120041307046,
+      "eval_rouge2": 0.21369562389429725,
+      "eval_rougeL": 0.3983722007696566,
+      "eval_runtime": 28.2169,
+      "eval_samples_per_second": 31.329,
+      "eval_steps_per_second": 3.934,
       "step": 1326
     },
     {
       "epoch": 3.0,
       "step": 1326,
       "total_flos": 4151799742464000.0,
-      "train_loss": 0.0,
-      "train_runtime": 0.0936,
-      "train_samples_per_second": 113198.697,
-      "train_steps_per_second": 14169.874
+      "train_loss": 3.14277273785115,
+      "train_runtime": 586.188,
+      "train_samples_per_second": 18.071,
+      "train_steps_per_second": 2.262
     }
   ],
   "logging_steps": 500,