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: 0.4779
-- Bleu: 0.2606
-- Rouge1: 0.6492
-- Rouge2: 0.4062
-- Rougel: 0.6482
+- Loss: 0.4183
+- Bleu: 0.2374
+- Rouge1: 0.6213
+- Rouge2: 0.3630
+- Rougel: 0.6195
 
 ## Model description
 

all_results.json

@@ -1,19 +1,19 @@
 {
-    "epoch": 9.0,
+    "epoch": 14.0,
     "eval_bleu": 0.23736106874390867,
     "eval_loss": 0.4183219075202942,
     "eval_rouge1": 0.6213217050496613,
     "eval_rouge2": 0.3629928034660268,
     "eval_rougeL": 0.6194945230812778,
-    "eval_runtime": 7.4528,
+    "eval_runtime": 6.133,
     "eval_samples": 1675,
-    "eval_samples_per_second": 224.748,
-    "eval_steps_per_second": 28.177,
+    "eval_samples_per_second": 273.111,
+    "eval_steps_per_second": 34.241,
     "perplexity": 1.5194097050971664,
-    "total_flos": 3938977382400000.0,
-    "train_loss": 0.41718409883979146,
-    "train_runtime": 2958.6328,
+    "total_flos": 6127298150400000.0,
+    "train_loss": 0.03383859399638808,
+    "train_runtime": 1080.9024,
     "train_samples": 6700,
-    "train_samples_per_second": 45.291,
-    "train_steps_per_second": 5.665
+    "train_samples_per_second": 123.97,
+    "train_steps_per_second": 15.506
 }

egy_training_log.txt

@@ -317,3 +317,5 @@ INFO:root:Epoch 13.0: Train Loss = 0.0936, Eval Loss = 0.470233678817749
 INFO:absl:Using default tokenizer.
 INFO:root:Epoch 14.0: Train Loss = 0.0899, Eval Loss = 0.47367072105407715
 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": 0.23736106874390867,
     "eval_loss": 0.4183219075202942,
     "eval_rouge1": 0.6213217050496613,
     "eval_rouge2": 0.3629928034660268,
     "eval_rougeL": 0.6194945230812778,
-    "eval_runtime": 7.4528,
+    "eval_runtime": 6.133,
     "eval_samples": 1675,
-    "eval_samples_per_second": 224.748,
-    "eval_steps_per_second": 28.177,
+    "eval_samples_per_second": 273.111,
+    "eval_steps_per_second": 34.241,
     "perplexity": 1.5194097050971664
 }

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 9.0,
-    "total_flos": 3938977382400000.0,
-    "train_loss": 0.41718409883979146,
-    "train_runtime": 2958.6328,
+    "epoch": 14.0,
+    "total_flos": 6127298150400000.0,
+    "train_loss": 0.03383859399638808,
+    "train_runtime": 1080.9024,
     "train_samples": 6700,
-    "train_samples_per_second": 45.291,
-    "train_steps_per_second": 5.665
+    "train_samples_per_second": 123.97,
+    "train_steps_per_second": 15.506
 }

trainer_state.json

@@ -1,9 +1,9 @@
 {
   "best_metric": 0.4183219075202942,
   "best_model_checkpoint": "/home/iais_marenpielka/Bouthaina/res_nw_gulf/checkpoint-3352",
-  "epoch": 9.0,
+  "epoch": 14.0,
   "eval_steps": 500,
-  "global_step": 7542,
+  "global_step": 11732,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -180,13 +180,108 @@
       "step": 7542
     },
     {
-      "epoch": 9.0,
-      "step": 7542,
-      "total_flos": 3938977382400000.0,
-      "train_loss": 0.41718409883979146,
-      "train_runtime": 2958.6328,
-      "train_samples_per_second": 45.291,
-      "train_steps_per_second": 5.665
+      "epoch": 10.0,
+      "grad_norm": 1.159719705581665,
+      "learning_rate": 2.5768757687576876e-05,
+      "loss": 0.105,
+      "step": 8380
+    },
+    {
+      "epoch": 10.0,
+      "eval_bleu": 0.2555335148067654,
+      "eval_loss": 0.45896273851394653,
+      "eval_rouge1": 0.6440840106866277,
+      "eval_rouge2": 0.3998364161836553,
+      "eval_rougeL": 0.6425824451349154,
+      "eval_runtime": 5.4849,
+      "eval_samples_per_second": 305.385,
+      "eval_steps_per_second": 38.287,
+      "step": 8380
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 0.923600435256958,
+      "learning_rate": 2.3191881918819188e-05,
+      "loss": 0.0982,
+      "step": 9218
+    },
+    {
+      "epoch": 11.0,
+      "eval_bleu": 0.25802091175426106,
+      "eval_loss": 0.46457362174987793,
+      "eval_rouge1": 0.6455262594531862,
+      "eval_rouge2": 0.40187791063551404,
+      "eval_rougeL": 0.6445205006389869,
+      "eval_runtime": 6.4432,
+      "eval_samples_per_second": 259.965,
+      "eval_steps_per_second": 32.593,
+      "step": 9218
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 0.9120431542396545,
+      "learning_rate": 2.0615006150061504e-05,
+      "loss": 0.0936,
+      "step": 10056
+    },
+    {
+      "epoch": 12.0,
+      "eval_bleu": 0.257202175451904,
+      "eval_loss": 0.470233678817749,
+      "eval_rouge1": 0.6456936789584837,
+      "eval_rouge2": 0.4045772651416589,
+      "eval_rougeL": 0.6445820832270409,
+      "eval_runtime": 13.7399,
+      "eval_samples_per_second": 121.907,
+      "eval_steps_per_second": 15.284,
+      "step": 10056
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 0.6435267329216003,
+      "learning_rate": 1.8038130381303812e-05,
+      "loss": 0.0899,
+      "step": 10894
+    },
+    {
+      "epoch": 13.0,
+      "eval_bleu": 0.25765634141938715,
+      "eval_loss": 0.47367072105407715,
+      "eval_rouge1": 0.6488279078084271,
+      "eval_rouge2": 0.40534740561340493,
+      "eval_rougeL": 0.6478436883847489,
+      "eval_runtime": 17.2761,
+      "eval_samples_per_second": 96.955,
+      "eval_steps_per_second": 12.156,
+      "step": 10894
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 0.7371336221694946,
+      "learning_rate": 1.5461254612546124e-05,
+      "loss": 0.0871,
+      "step": 11732
+    },
+    {
+      "epoch": 14.0,
+      "eval_bleu": 0.2606034683461693,
+      "eval_loss": 0.4779162108898163,
+      "eval_rouge1": 0.6491650390141908,
+      "eval_rouge2": 0.40618585329089035,
+      "eval_rougeL": 0.6482221321240169,
+      "eval_runtime": 5.371,
+      "eval_samples_per_second": 311.859,
+      "eval_steps_per_second": 39.099,
+      "step": 11732
+    },
+    {
+      "epoch": 14.0,
+      "step": 11732,
+      "total_flos": 6127298150400000.0,
+      "train_loss": 0.03383859399638808,
+      "train_runtime": 1080.9024,
+      "train_samples_per_second": 123.97,
+      "train_steps_per_second": 15.506
     }
   ],
   "logging_steps": 500,
@@ -215,7 +310,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 3938977382400000.0,
+  "total_flos": 6127298150400000.0,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null