End of training

README.md

@@ -18,11 +18,11 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [riotu-lab/ArabianGPT-01B](https://huggingface.co/riotu-lab/ArabianGPT-01B) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.8827
-- Bleu: 0.2463
-- Rouge1: 0.6087
-- Rouge2: 0.3631
-- Rougel: 0.6049
+- Loss: 0.7316
+- Bleu: 0.2459
+- Rouge1: 0.6017
+- Rouge2: 0.3519
+- Rougel: 0.5983
 
 ## Model description
 

all_results.json

@@ -1,19 +1,19 @@
 {
-    "epoch": 9.0,
+    "epoch": 14.0,
     "eval_bleu": 0.24587227576979195,
     "eval_loss": 0.7315686941146851,
     "eval_rouge1": 0.6017197427075045,
     "eval_rouge2": 0.3518746485163118,
     "eval_rougeL": 0.5982542515796094,
-    "eval_runtime": 47.1591,
+    "eval_runtime": 47.0307,
     "eval_samples": 14212,
-    "eval_samples_per_second": 301.363,
-    "eval_steps_per_second": 37.681,
+    "eval_samples_per_second": 302.185,
+    "eval_steps_per_second": 37.784,
     "perplexity": 2.078338328926906,
-    "total_flos": 3.3423104950272e+16,
-    "train_loss": 0.5779510789562912,
-    "train_runtime": 8660.6869,
+    "total_flos": 5.1991496589312e+16,
+    "train_loss": 0.09352104669935263,
+    "train_runtime": 3380.5822,
     "train_samples": 56851,
-    "train_samples_per_second": 131.285,
-    "train_steps_per_second": 16.412
+    "train_samples_per_second": 336.339,
+    "train_steps_per_second": 42.046
 }

egy_training_log.txt

@@ -317,3 +317,5 @@ INFO:root:Epoch 13.0: Train Loss = 0.2586, Eval Loss = 0.8498404026031494
 INFO:absl:Using default tokenizer.
 INFO:root:Epoch 14.0: Train Loss = 0.2392, Eval Loss = 0.867546021938324
 INFO:absl:Using default tokenizer.
+INFO:__main__:*** Evaluate ***
+INFO:absl:Using default tokenizer.

eval_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 9.0,
+    "epoch": 14.0,
     "eval_bleu": 0.24587227576979195,
     "eval_loss": 0.7315686941146851,
     "eval_rouge1": 0.6017197427075045,
     "eval_rouge2": 0.3518746485163118,
     "eval_rougeL": 0.5982542515796094,
-    "eval_runtime": 47.1591,
+    "eval_runtime": 47.0307,
     "eval_samples": 14212,
-    "eval_samples_per_second": 301.363,
-    "eval_steps_per_second": 37.681,
+    "eval_samples_per_second": 302.185,
+    "eval_steps_per_second": 37.784,
     "perplexity": 2.078338328926906
 }

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 9.0,
-    "total_flos": 3.3423104950272e+16,
-    "train_loss": 0.5779510789562912,
-    "train_runtime": 8660.6869,
+    "epoch": 14.0,
+    "total_flos": 5.1991496589312e+16,
+    "train_loss": 0.09352104669935263,
+    "train_runtime": 3380.5822,
     "train_samples": 56851,
-    "train_samples_per_second": 131.285,
-    "train_steps_per_second": 16.412
+    "train_samples_per_second": 336.339,
+    "train_steps_per_second": 42.046
 }

trainer_state.json

@@ -1,9 +1,9 @@
 {
   "best_metric": 0.7315686941146851,
   "best_model_checkpoint": "/home/iais_marenpielka/Bouthaina/res_nw_eg/checkpoint-28428",
-  "epoch": 9.0,
+  "epoch": 14.0,
   "eval_steps": 500,
-  "global_step": 63963,
+  "global_step": 99498,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -180,13 +180,108 @@
       "step": 63963
     },
     {
-      "epoch": 9.0,
-      "step": 63963,
-      "total_flos": 3.3423104950272e+16,
-      "train_loss": 0.5779510789562912,
-      "train_runtime": 8660.6869,
-      "train_samples_per_second": 131.285,
-      "train_steps_per_second": 16.412
+      "epoch": 10.0,
+      "grad_norm": 2.328486442565918,
+      "learning_rate": 2.5088251906241178e-05,
+      "loss": 0.3074,
+      "step": 71070
+    },
+    {
+      "epoch": 10.0,
+      "eval_bleu": 0.24784142265151649,
+      "eval_loss": 0.8155524134635925,
+      "eval_rouge1": 0.6101301030440347,
+      "eval_rouge2": 0.364415924488246,
+      "eval_rougeL": 0.606258270290786,
+      "eval_runtime": 82.2965,
+      "eval_samples_per_second": 172.693,
+      "eval_steps_per_second": 21.593,
+      "step": 71070
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 2.1735293865203857,
+      "learning_rate": 2.2579426715617057e-05,
+      "loss": 0.2813,
+      "step": 78177
+    },
+    {
+      "epoch": 11.0,
+      "eval_bleu": 0.24877766407334076,
+      "eval_loss": 0.8325821757316589,
+      "eval_rouge1": 0.6108646154589945,
+      "eval_rouge2": 0.36567755520636375,
+      "eval_rougeL": 0.6071552064184781,
+      "eval_runtime": 46.9697,
+      "eval_samples_per_second": 302.578,
+      "eval_steps_per_second": 37.833,
+      "step": 78177
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 2.380600929260254,
+      "learning_rate": 2.0070601524992942e-05,
+      "loss": 0.2586,
+      "step": 85284
+    },
+    {
+      "epoch": 12.0,
+      "eval_bleu": 0.24613125712387748,
+      "eval_loss": 0.8498404026031494,
+      "eval_rouge1": 0.6082478770786404,
+      "eval_rouge2": 0.3630077370066766,
+      "eval_rougeL": 0.604462114793365,
+      "eval_runtime": 46.8897,
+      "eval_samples_per_second": 303.094,
+      "eval_steps_per_second": 37.897,
+      "step": 85284
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 2.3239755630493164,
+      "learning_rate": 1.7561776334368824e-05,
+      "loss": 0.2392,
+      "step": 92391
+    },
+    {
+      "epoch": 13.0,
+      "eval_bleu": 0.24701514132558042,
+      "eval_loss": 0.867546021938324,
+      "eval_rouge1": 0.6094521129523778,
+      "eval_rouge2": 0.364158573223837,
+      "eval_rougeL": 0.6054673467586391,
+      "eval_runtime": 169.8152,
+      "eval_samples_per_second": 83.691,
+      "eval_steps_per_second": 10.464,
+      "step": 92391
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 2.4565744400024414,
+      "learning_rate": 1.5052951143744705e-05,
+      "loss": 0.2227,
+      "step": 99498
+    },
+    {
+      "epoch": 14.0,
+      "eval_bleu": 0.24626598003060837,
+      "eval_loss": 0.8826896548271179,
+      "eval_rouge1": 0.6087178921955774,
+      "eval_rouge2": 0.36314063359901005,
+      "eval_rougeL": 0.6048623164787159,
+      "eval_runtime": 170.1544,
+      "eval_samples_per_second": 83.524,
+      "eval_steps_per_second": 10.443,
+      "step": 99498
+    },
+    {
+      "epoch": 14.0,
+      "step": 99498,
+      "total_flos": 5.1991496589312e+16,
+      "train_loss": 0.09352104669935263,
+      "train_runtime": 3380.5822,
+      "train_samples_per_second": 336.339,
+      "train_steps_per_second": 42.046
     }
   ],
   "logging_steps": 500,
@@ -215,7 +310,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 3.3423104950272e+16,
+  "total_flos": 5.1991496589312e+16,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null