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-1.5B](https://huggingface.co/riotu-lab/ArabianGPT-1.5B) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.7326
-- Bleu: 0.3760
-- Rouge1: 0.5807
-- Rouge2: 0.3234
-- Rougel: 0.5784
+- Loss: 0.6726
+- Bleu: 0.3710
+- Rouge1: 0.5676
+- Rouge2: 0.3031
+- Rougel: 0.5652
 
 ## Model description
 

all_results.json

@@ -0,0 +1,19 @@
+{
+    "epoch": 10.0,
+    "eval_bleu": 0.3710035049139421,
+    "eval_loss": 0.6725558042526245,
+    "eval_rouge1": 0.5675803798053409,
+    "eval_rouge2": 0.3030603907649001,
+    "eval_rougeL": 0.5652070951016016,
+    "eval_runtime": 145.1346,
+    "eval_samples": 5380,
+    "eval_samples_per_second": 37.069,
+    "eval_steps_per_second": 4.637,
+    "perplexity": 1.9592383568031233,
+    "total_flos": 2.42840137162752e+17,
+    "train_loss": 0.5466835172908793,
+    "train_runtime": 39682.6675,
+    "train_samples": 21429,
+    "train_samples_per_second": 10.8,
+    "train_steps_per_second": 1.35
+}

egy_training_log.txt

@@ -160,3 +160,5 @@ INFO:root:Epoch 9.0: Train Loss = 0.3353, Eval Loss = 0.6988218426704407
 INFO:absl:Using default tokenizer.
 INFO:root:Epoch 10.0: Train Loss = 0.2942, Eval Loss = 0.7147605419158936
 INFO:absl:Using default tokenizer.
+INFO:__main__:*** Evaluate ***
+INFO:absl:Using default tokenizer.

eval_results.json

@@ -0,0 +1,13 @@
+{
+    "epoch": 10.0,
+    "eval_bleu": 0.3710035049139421,
+    "eval_loss": 0.6725558042526245,
+    "eval_rouge1": 0.5675803798053409,
+    "eval_rouge2": 0.3030603907649001,
+    "eval_rougeL": 0.5652070951016016,
+    "eval_runtime": 145.1346,
+    "eval_samples": 5380,
+    "eval_samples_per_second": 37.069,
+    "eval_steps_per_second": 4.637,
+    "perplexity": 1.9592383568031233
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 10.0,
+    "total_flos": 2.42840137162752e+17,
+    "train_loss": 0.5466835172908793,
+    "train_runtime": 39682.6675,
+    "train_samples": 21429,
+    "train_samples_per_second": 10.8,
+    "train_steps_per_second": 1.35
+}

trainer_state.json

@@ -0,0 +1,241 @@
+{
+  "best_metric": 0.6725558042526245,
+  "best_model_checkpoint": "/home/iais_marenpielka/Bouthaina/res_nw_dj_1.5b/checkpoint-13395",
+  "epoch": 10.0,
+  "eval_steps": 500,
+  "global_step": 26790,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.771394670009613,
+      "learning_rate": 4.7947437829691034e-05,
+      "loss": 1.4014,
+      "step": 2679
+    },
+    {
+      "epoch": 1.0,
+      "eval_bleu": 0.333905242851474,
+      "eval_loss": 0.7688375115394592,
+      "eval_rouge1": 0.48423971066981253,
+      "eval_rouge2": 0.21153498148413064,
+      "eval_rougeL": 0.4815428367948251,
+      "eval_runtime": 207.2379,
+      "eval_samples_per_second": 25.961,
+      "eval_steps_per_second": 3.247,
+      "step": 2679
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.9780656695365906,
+      "learning_rate": 4.542388847023361e-05,
+      "loss": 0.706,
+      "step": 5358
+    },
+    {
+      "epoch": 2.0,
+      "eval_bleu": 0.35260875951359005,
+      "eval_loss": 0.7103222608566284,
+      "eval_rouge1": 0.5204323597665172,
+      "eval_rouge2": 0.24873477681283626,
+      "eval_rougeL": 0.5179629154924683,
+      "eval_runtime": 150.5501,
+      "eval_samples_per_second": 35.736,
+      "eval_steps_per_second": 4.47,
+      "step": 5358
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.8800147771835327,
+      "learning_rate": 4.290033911077619e-05,
+      "loss": 0.6195,
+      "step": 8037
+    },
+    {
+      "epoch": 3.0,
+      "eval_bleu": 0.36352997505195117,
+      "eval_loss": 0.6853417158126831,
+      "eval_rouge1": 0.5434024863238025,
+      "eval_rouge2": 0.2747842276086267,
+      "eval_rougeL": 0.5410553184881033,
+      "eval_runtime": 145.7188,
+      "eval_samples_per_second": 36.92,
+      "eval_steps_per_second": 4.618,
+      "step": 8037
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 1.0725535154342651,
+      "learning_rate": 4.0376789751318766e-05,
+      "loss": 0.5498,
+      "step": 10716
+    },
+    {
+      "epoch": 4.0,
+      "eval_bleu": 0.3675234824247013,
+      "eval_loss": 0.6749088168144226,
+      "eval_rouge1": 0.5582749012549353,
+      "eval_rouge2": 0.289810829038981,
+      "eval_rougeL": 0.5557677336771762,
+      "eval_runtime": 146.4465,
+      "eval_samples_per_second": 36.737,
+      "eval_steps_per_second": 4.596,
+      "step": 10716
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.9458489418029785,
+      "learning_rate": 3.785324039186134e-05,
+      "loss": 0.4884,
+      "step": 13395
+    },
+    {
+      "epoch": 5.0,
+      "eval_bleu": 0.3710035049139421,
+      "eval_loss": 0.6725558042526245,
+      "eval_rouge1": 0.5675803798053409,
+      "eval_rouge2": 0.3030603907649001,
+      "eval_rougeL": 0.5652070951016016,
+      "eval_runtime": 145.5306,
+      "eval_samples_per_second": 36.968,
+      "eval_steps_per_second": 4.624,
+      "step": 13395
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 1.10252046585083,
+      "learning_rate": 3.532969103240392e-05,
+      "loss": 0.4327,
+      "step": 16074
+    },
+    {
+      "epoch": 6.0,
+      "eval_bleu": 0.37667057605203397,
+      "eval_loss": 0.6755143404006958,
+      "eval_rouge1": 0.5737053663827637,
+      "eval_rouge2": 0.3128433873357377,
+      "eval_rougeL": 0.5714111425642474,
+      "eval_runtime": 145.5321,
+      "eval_samples_per_second": 36.968,
+      "eval_steps_per_second": 4.624,
+      "step": 16074
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 1.2120763063430786,
+      "learning_rate": 3.28061416729465e-05,
+      "loss": 0.3816,
+      "step": 18753
+    },
+    {
+      "epoch": 7.0,
+      "eval_bleu": 0.3766396732779585,
+      "eval_loss": 0.685249388217926,
+      "eval_rouge1": 0.5767112819015094,
+      "eval_rouge2": 0.3159316111397813,
+      "eval_rougeL": 0.5745753191765783,
+      "eval_runtime": 157.4622,
+      "eval_samples_per_second": 34.167,
+      "eval_steps_per_second": 4.274,
+      "step": 18753
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 1.0003564357757568,
+      "learning_rate": 3.028259231348907e-05,
+      "loss": 0.3353,
+      "step": 21432
+    },
+    {
+      "epoch": 8.0,
+      "eval_bleu": 0.3720908799990584,
+      "eval_loss": 0.6988218426704407,
+      "eval_rouge1": 0.5774281057590944,
+      "eval_rouge2": 0.3197785460431908,
+      "eval_rougeL": 0.5754197829494876,
+      "eval_runtime": 149.1468,
+      "eval_samples_per_second": 36.072,
+      "eval_steps_per_second": 4.512,
+      "step": 21432
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 1.2703362703323364,
+      "learning_rate": 2.775904295403165e-05,
+      "loss": 0.2942,
+      "step": 24111
+    },
+    {
+      "epoch": 9.0,
+      "eval_bleu": 0.37678208073339087,
+      "eval_loss": 0.7147605419158936,
+      "eval_rouge1": 0.5784888328390186,
+      "eval_rouge2": 0.32068030075147713,
+      "eval_rougeL": 0.5762720217407303,
+      "eval_runtime": 146.6837,
+      "eval_samples_per_second": 36.678,
+      "eval_steps_per_second": 4.588,
+      "step": 24111
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 1.0180821418762207,
+      "learning_rate": 2.523549359457423e-05,
+      "loss": 0.2579,
+      "step": 26790
+    },
+    {
+      "epoch": 10.0,
+      "eval_bleu": 0.37598768218760426,
+      "eval_loss": 0.7325860857963562,
+      "eval_rouge1": 0.5807141412670587,
+      "eval_rouge2": 0.32344450667178803,
+      "eval_rougeL": 0.5784030823868939,
+      "eval_runtime": 145.4936,
+      "eval_samples_per_second": 36.978,
+      "eval_steps_per_second": 4.626,
+      "step": 26790
+    },
+    {
+      "epoch": 10.0,
+      "step": 26790,
+      "total_flos": 2.42840137162752e+17,
+      "train_loss": 0.5466835172908793,
+      "train_runtime": 39682.6675,
+      "train_samples_per_second": 10.8,
+      "train_steps_per_second": 1.35
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 53580,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 20,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 5,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 0
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.42840137162752e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}