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 [aubmindlab/aragpt2-large](https://huggingface.co/aubmindlab/aragpt2-large) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.0583
-- Bleu: 0.0918
-- Rouge1: 0.4656
-- Rouge2: 0.2308
-- Rougel: 0.4636
+- Loss: 0.0472
+- Bleu: 0.0632
+- Rouge1: 0.4039
+- Rouge2: 0.1633
+- Rougel: 0.4013
 
 ## Model description
 

all_results.json

@@ -0,0 +1,19 @@
+{
+    "epoch": 7.0,
+    "eval_bleu": 0.06321503399346401,
+    "eval_loss": 0.047215357422828674,
+    "eval_rouge1": 0.40393751877732786,
+    "eval_rouge2": 0.16331196085274022,
+    "eval_rougeL": 0.4013347547558428,
+    "eval_runtime": 342.1869,
+    "eval_samples": 1672,
+    "eval_samples_per_second": 4.886,
+    "eval_steps_per_second": 1.222,
+    "perplexity": 1.0483477542414168,
+    "total_flos": 2.03668247150592e+17,
+    "train_loss": 0.04967995416767697,
+    "train_runtime": 22443.4901,
+    "train_samples": 6685,
+    "train_samples_per_second": 5.957,
+    "train_steps_per_second": 1.49
+}

egy_training_log.txt

@@ -568,3 +568,5 @@ INFO:root:Epoch 6.0: Train Loss = 0.0189, Eval Loss = 0.053812965750694275
 INFO:absl:Using default tokenizer.
 INFO:root:Epoch 7.0: Train Loss = 0.0164, Eval Loss = 0.05724157765507698
 INFO:absl:Using default tokenizer.
+INFO:__main__:*** Evaluate ***
+INFO:absl:Using default tokenizer.

eval_results.json

@@ -0,0 +1,13 @@
+{
+    "epoch": 7.0,
+    "eval_bleu": 0.06321503399346401,
+    "eval_loss": 0.047215357422828674,
+    "eval_rouge1": 0.40393751877732786,
+    "eval_rouge2": 0.16331196085274022,
+    "eval_rougeL": 0.4013347547558428,
+    "eval_runtime": 342.1869,
+    "eval_samples": 1672,
+    "eval_samples_per_second": 4.886,
+    "eval_steps_per_second": 1.222,
+    "perplexity": 1.0483477542414168
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 7.0,
+    "total_flos": 2.03668247150592e+17,
+    "train_loss": 0.04967995416767697,
+    "train_runtime": 22443.4901,
+    "train_samples": 6685,
+    "train_samples_per_second": 5.957,
+    "train_steps_per_second": 1.49
+}

trainer_state.json

@@ -0,0 +1,184 @@
+{
+  "best_metric": 0.047215357422828674,
+  "best_model_checkpoint": "/home/iais_marenpielka/Bouthaina/res_nw_gulf_aragpt2-large/checkpoint-3344",
+  "epoch": 7.0,
+  "eval_steps": 500,
+  "global_step": 11704,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7535182237625122,
+      "learning_rate": 4.822100789313904e-05,
+      "loss": 0.2041,
+      "step": 1672
+    },
+    {
+      "epoch": 1.0,
+      "eval_bleu": 0.04446881026937374,
+      "eval_loss": 0.04873450845479965,
+      "eval_rouge1": 0.3745425104768441,
+      "eval_rouge2": 0.13019286214073578,
+      "eval_rougeL": 0.37175423318625556,
+      "eval_runtime": 280.7972,
+      "eval_samples_per_second": 5.954,
+      "eval_steps_per_second": 1.489,
+      "step": 1672
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5986363887786865,
+      "learning_rate": 4.568306010928962e-05,
+      "loss": 0.0404,
+      "step": 3344
+    },
+    {
+      "epoch": 2.0,
+      "eval_bleu": 0.06321503399346401,
+      "eval_loss": 0.047215357422828674,
+      "eval_rouge1": 0.40393751877732786,
+      "eval_rouge2": 0.16331196085274022,
+      "eval_rougeL": 0.4013347547558428,
+      "eval_runtime": 280.6911,
+      "eval_samples_per_second": 5.957,
+      "eval_steps_per_second": 1.489,
+      "step": 3344
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.6692299246788025,
+      "learning_rate": 4.3145112325440196e-05,
+      "loss": 0.0301,
+      "step": 5016
+    },
+    {
+      "epoch": 3.0,
+      "eval_bleu": 0.07634204530814317,
+      "eval_loss": 0.047973889857530594,
+      "eval_rouge1": 0.43387069380649074,
+      "eval_rouge2": 0.20015750349405118,
+      "eval_rougeL": 0.4322337100043073,
+      "eval_runtime": 342.1899,
+      "eval_samples_per_second": 4.886,
+      "eval_steps_per_second": 1.222,
+      "step": 5016
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.39187440276145935,
+      "learning_rate": 4.0607164541590774e-05,
+      "loss": 0.0232,
+      "step": 6688
+    },
+    {
+      "epoch": 4.0,
+      "eval_bleu": 0.08429412502024626,
+      "eval_loss": 0.05150514841079712,
+      "eval_rouge1": 0.45350332758712053,
+      "eval_rouge2": 0.2191636179696206,
+      "eval_rougeL": 0.45170908410677657,
+      "eval_runtime": 220.2929,
+      "eval_samples_per_second": 7.59,
+      "eval_steps_per_second": 1.897,
+      "step": 6688
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.7217269539833069,
+      "learning_rate": 3.806921675774135e-05,
+      "loss": 0.0189,
+      "step": 8360
+    },
+    {
+      "epoch": 5.0,
+      "eval_bleu": 0.08759878040919578,
+      "eval_loss": 0.053812965750694275,
+      "eval_rouge1": 0.4654488236955554,
+      "eval_rouge2": 0.22994914252764548,
+      "eval_rougeL": 0.46382599849734485,
+      "eval_runtime": 342.2351,
+      "eval_samples_per_second": 4.886,
+      "eval_steps_per_second": 1.221,
+      "step": 8360
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 0.5110601186752319,
+      "learning_rate": 3.553126897389193e-05,
+      "loss": 0.0164,
+      "step": 10032
+    },
+    {
+      "epoch": 6.0,
+      "eval_bleu": 0.09295393672158372,
+      "eval_loss": 0.05724157765507698,
+      "eval_rouge1": 0.46747472033375215,
+      "eval_rouge2": 0.2370003193385344,
+      "eval_rougeL": 0.46529496417190075,
+      "eval_runtime": 342.1804,
+      "eval_samples_per_second": 4.886,
+      "eval_steps_per_second": 1.222,
+      "step": 10032
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 0.23091594874858856,
+      "learning_rate": 3.2993321190042506e-05,
+      "loss": 0.0148,
+      "step": 11704
+    },
+    {
+      "epoch": 7.0,
+      "eval_bleu": 0.0918093965340221,
+      "eval_loss": 0.05825383961200714,
+      "eval_rouge1": 0.46559708964007684,
+      "eval_rouge2": 0.2307789620256464,
+      "eval_rougeL": 0.46361556118175484,
+      "eval_runtime": 280.7351,
+      "eval_samples_per_second": 5.956,
+      "eval_steps_per_second": 1.489,
+      "step": 11704
+    },
+    {
+      "epoch": 7.0,
+      "step": 11704,
+      "total_flos": 2.03668247150592e+17,
+      "train_loss": 0.04967995416767697,
+      "train_runtime": 22443.4901,
+      "train_samples_per_second": 5.957,
+      "train_steps_per_second": 1.49
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 33440,
+  "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": 2.03668247150592e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}