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: 1.9809
-- Bleu: 0.3149
-- Rouge1: 0.5865
-- Rouge2: 0.3495
-- Rougel: 0.5486
+- Loss: 1.9793
+- Bleu: 0.3130
+- Rouge1: 0.5832
+- Rouge2: 0.3461
+- Rougel: 0.5454
 
 ## Model description
 

all_results.json

@@ -0,0 +1,19 @@
+{
+    "epoch": 18.0,
+    "eval_bleu": 0.31303204778954513,
+    "eval_loss": 1.9793106317520142,
+    "eval_rouge1": 0.5832240471437244,
+    "eval_rouge2": 0.34612838297153165,
+    "eval_rougeL": 0.545438606875621,
+    "eval_runtime": 29.6378,
+    "eval_samples": 925,
+    "eval_samples_per_second": 31.21,
+    "eval_steps_per_second": 3.914,
+    "perplexity": 7.237751788702562,
+    "total_flos": 2.5969031184384e+16,
+    "train_loss": 1.6592627337077177,
+    "train_runtime": 3469.9556,
+    "train_samples": 3681,
+    "train_samples_per_second": 21.216,
+    "train_steps_per_second": 2.657
+}

egy_training_log.txt

@@ -319,3 +319,5 @@ INFO:root:Epoch 17.0: Train Loss = 1.4361, Eval Loss = 1.9800282716751099
 INFO:absl:Using default tokenizer.
 INFO:root:Epoch 18.0: Train Loss = 1.4159, Eval Loss = 1.979936957359314
 INFO:absl:Using default tokenizer.
+INFO:__main__:*** Evaluate ***
+INFO:absl:Using default tokenizer.

eval_results.json

@@ -0,0 +1,13 @@
+{
+    "epoch": 18.0,
+    "eval_bleu": 0.31303204778954513,
+    "eval_loss": 1.9793106317520142,
+    "eval_rouge1": 0.5832240471437244,
+    "eval_rouge2": 0.34612838297153165,
+    "eval_rougeL": 0.545438606875621,
+    "eval_runtime": 29.6378,
+    "eval_samples": 925,
+    "eval_samples_per_second": 31.21,
+    "eval_steps_per_second": 3.914,
+    "perplexity": 7.237751788702562
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 18.0,
+    "total_flos": 2.5969031184384e+16,
+    "train_loss": 1.6592627337077177,
+    "train_runtime": 3469.9556,
+    "train_samples": 3681,
+    "train_samples_per_second": 21.216,
+    "train_steps_per_second": 2.657
+}

trainer_state.json

@@ -0,0 +1,379 @@
+{
+  "best_metric": 1.9793106317520142,
+  "best_model_checkpoint": "/home/iais_marenpielka/Bouthaina/results_fixed/checkpoint-6915",
+  "epoch": 18.0,
+  "eval_steps": 500,
+  "global_step": 8298,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.0845986984815619,
+      "grad_norm": 1.1301698684692383,
+      "learning_rate": 5e-05,
+      "loss": 3.9816,
+      "step": 500
+    },
+    {
+      "epoch": 1.0845986984815619,
+      "eval_bleu": 0.22704779726518312,
+      "eval_loss": 2.8213491439819336,
+      "eval_rouge1": 0.3850822060355352,
+      "eval_rouge2": 0.1625206163604567,
+      "eval_rougeL": 0.31976194879885783,
+      "eval_runtime": 89.8358,
+      "eval_samples_per_second": 10.297,
+      "eval_steps_per_second": 1.291,
+      "step": 500
+    },
+    {
+      "epoch": 2.0,
+      "eval_bleu": 0.2480969312402941,
+      "eval_loss": 2.4793343544006348,
+      "eval_rouge1": 0.4405430075781575,
+      "eval_rouge2": 0.20302695232764206,
+      "eval_rougeL": 0.3844289432538539,
+      "eval_runtime": 29.4641,
+      "eval_samples_per_second": 31.394,
+      "eval_steps_per_second": 3.937,
+      "step": 922
+    },
+    {
+      "epoch": 2.1691973969631237,
+      "grad_norm": 1.0731440782546997,
+      "learning_rate": 4.713302752293578e-05,
+      "loss": 2.6514,
+      "step": 1000
+    },
+    {
+      "epoch": 3.0,
+      "eval_bleu": 0.2658003323042654,
+      "eval_loss": 2.2930543422698975,
+      "eval_rouge1": 0.48284806373571365,
+      "eval_rouge2": 0.2384861670767343,
+      "eval_rougeL": 0.43274525400088326,
+      "eval_runtime": 91.1451,
+      "eval_samples_per_second": 10.149,
+      "eval_steps_per_second": 1.273,
+      "step": 1383
+    },
+    {
+      "epoch": 3.2537960954446854,
+      "grad_norm": 1.1006942987442017,
+      "learning_rate": 4.426605504587156e-05,
+      "loss": 2.3308,
+      "step": 1500
+    },
+    {
+      "epoch": 4.0,
+      "eval_bleu": 0.27970338794806143,
+      "eval_loss": 2.180091619491577,
+      "eval_rouge1": 0.5114337163243292,
+      "eval_rouge2": 0.267870161729887,
+      "eval_rougeL": 0.4658699841467887,
+      "eval_runtime": 29.4867,
+      "eval_samples_per_second": 31.37,
+      "eval_steps_per_second": 3.934,
+      "step": 1844
+    },
+    {
+      "epoch": 4.3383947939262475,
+      "grad_norm": 1.0734232664108276,
+      "learning_rate": 4.139908256880734e-05,
+      "loss": 2.1322,
+      "step": 2000
+    },
+    {
+      "epoch": 5.0,
+      "eval_bleu": 0.2886420174009109,
+      "eval_loss": 2.1132538318634033,
+      "eval_rouge1": 0.5263880712269413,
+      "eval_rouge2": 0.28674622405060524,
+      "eval_rougeL": 0.4851858911276635,
+      "eval_runtime": 29.5338,
+      "eval_samples_per_second": 31.32,
+      "eval_steps_per_second": 3.928,
+      "step": 2305
+    },
+    {
+      "epoch": 5.422993492407809,
+      "grad_norm": 1.058827519416809,
+      "learning_rate": 3.8532110091743125e-05,
+      "loss": 1.9942,
+      "step": 2500
+    },
+    {
+      "epoch": 6.0,
+      "eval_bleu": 0.2925900760553107,
+      "eval_loss": 2.064920663833618,
+      "eval_rouge1": 0.5417489504784114,
+      "eval_rouge2": 0.2993448660857305,
+      "eval_rougeL": 0.5002749085104436,
+      "eval_runtime": 29.6386,
+      "eval_samples_per_second": 31.209,
+      "eval_steps_per_second": 3.914,
+      "step": 2766
+    },
+    {
+      "epoch": 6.507592190889371,
+      "grad_norm": 1.0469533205032349,
+      "learning_rate": 3.56651376146789e-05,
+      "loss": 1.8884,
+      "step": 3000
+    },
+    {
+      "epoch": 7.0,
+      "eval_bleu": 0.2966637771816971,
+      "eval_loss": 2.03544545173645,
+      "eval_rouge1": 0.5528641921007253,
+      "eval_rouge2": 0.3107950360794697,
+      "eval_rougeL": 0.5117019175789275,
+      "eval_runtime": 29.7995,
+      "eval_samples_per_second": 31.041,
+      "eval_steps_per_second": 3.893,
+      "step": 3227
+    },
+    {
+      "epoch": 7.592190889370933,
+      "grad_norm": 1.1290621757507324,
+      "learning_rate": 3.2798165137614676e-05,
+      "loss": 1.8003,
+      "step": 3500
+    },
+    {
+      "epoch": 8.0,
+      "eval_bleu": 0.297270373541541,
+      "eval_loss": 2.016411304473877,
+      "eval_rouge1": 0.5596603132781475,
+      "eval_rouge2": 0.3199340438705295,
+      "eval_rougeL": 0.520436776425443,
+      "eval_runtime": 29.6516,
+      "eval_samples_per_second": 31.196,
+      "eval_steps_per_second": 3.912,
+      "step": 3688
+    },
+    {
+      "epoch": 8.676789587852495,
+      "grad_norm": 1.0918660163879395,
+      "learning_rate": 2.9931192660550462e-05,
+      "loss": 1.7305,
+      "step": 4000
+    },
+    {
+      "epoch": 9.0,
+      "eval_bleu": 0.30530297795149647,
+      "eval_loss": 2.0036492347717285,
+      "eval_rouge1": 0.5659695387091741,
+      "eval_rouge2": 0.3261535326554196,
+      "eval_rougeL": 0.5265625742800235,
+      "eval_runtime": 29.7893,
+      "eval_samples_per_second": 31.051,
+      "eval_steps_per_second": 3.894,
+      "step": 4149
+    },
+    {
+      "epoch": 9.761388286334057,
+      "grad_norm": 1.1562738418579102,
+      "learning_rate": 2.7064220183486238e-05,
+      "loss": 1.6672,
+      "step": 4500
+    },
+    {
+      "epoch": 10.0,
+      "eval_bleu": 0.3071835871794814,
+      "eval_loss": 1.993295669555664,
+      "eval_rouge1": 0.5704281178813693,
+      "eval_rouge2": 0.3318532927873396,
+      "eval_rougeL": 0.5325234479218677,
+      "eval_runtime": 29.7553,
+      "eval_samples_per_second": 31.087,
+      "eval_steps_per_second": 3.898,
+      "step": 4610
+    },
+    {
+      "epoch": 10.845986984815617,
+      "grad_norm": 1.1015173196792603,
+      "learning_rate": 2.419724770642202e-05,
+      "loss": 1.6132,
+      "step": 5000
+    },
+    {
+      "epoch": 11.0,
+      "eval_bleu": 0.3093132521096642,
+      "eval_loss": 1.9886106252670288,
+      "eval_rouge1": 0.5737300860653806,
+      "eval_rouge2": 0.33663827794301615,
+      "eval_rougeL": 0.536346207031389,
+      "eval_runtime": 29.743,
+      "eval_samples_per_second": 31.1,
+      "eval_steps_per_second": 3.9,
+      "step": 5071
+    },
+    {
+      "epoch": 11.93058568329718,
+      "grad_norm": 1.0905733108520508,
+      "learning_rate": 2.13302752293578e-05,
+      "loss": 1.5659,
+      "step": 5500
+    },
+    {
+      "epoch": 12.0,
+      "eval_bleu": 0.3098881020021761,
+      "eval_loss": 1.9834030866622925,
+      "eval_rouge1": 0.5777003074459022,
+      "eval_rouge2": 0.3397382155739319,
+      "eval_rougeL": 0.5395519937552808,
+      "eval_runtime": 29.7562,
+      "eval_samples_per_second": 31.086,
+      "eval_steps_per_second": 3.898,
+      "step": 5532
+    },
+    {
+      "epoch": 13.0,
+      "eval_bleu": 0.3117110664590574,
+      "eval_loss": 1.981902003288269,
+      "eval_rouge1": 0.5796236336727056,
+      "eval_rouge2": 0.34230164752840053,
+      "eval_rougeL": 0.5417627454602093,
+      "eval_runtime": 29.7814,
+      "eval_samples_per_second": 31.06,
+      "eval_steps_per_second": 3.895,
+      "step": 5993
+    },
+    {
+      "epoch": 13.015184381778742,
+      "grad_norm": 1.0530316829681396,
+      "learning_rate": 1.8463302752293578e-05,
+      "loss": 1.5244,
+      "step": 6000
+    },
+    {
+      "epoch": 14.0,
+      "eval_bleu": 0.3126312196492616,
+      "eval_loss": 1.9798119068145752,
+      "eval_rouge1": 0.5833425546826478,
+      "eval_rouge2": 0.34524587884448443,
+      "eval_rougeL": 0.5450520234727538,
+      "eval_runtime": 29.8075,
+      "eval_samples_per_second": 31.032,
+      "eval_steps_per_second": 3.892,
+      "step": 6454
+    },
+    {
+      "epoch": 14.099783080260304,
+      "grad_norm": 1.1455042362213135,
+      "learning_rate": 1.559633027522936e-05,
+      "loss": 1.4884,
+      "step": 6500
+    },
+    {
+      "epoch": 15.0,
+      "eval_bleu": 0.31303204778954513,
+      "eval_loss": 1.9793106317520142,
+      "eval_rouge1": 0.5832240471437244,
+      "eval_rouge2": 0.34612838297153165,
+      "eval_rougeL": 0.545438606875621,
+      "eval_runtime": 29.7921,
+      "eval_samples_per_second": 31.048,
+      "eval_steps_per_second": 3.894,
+      "step": 6915
+    },
+    {
+      "epoch": 15.184381778741866,
+      "grad_norm": 1.1202692985534668,
+      "learning_rate": 1.2729357798165138e-05,
+      "loss": 1.4594,
+      "step": 7000
+    },
+    {
+      "epoch": 16.0,
+      "eval_bleu": 0.3133487171249794,
+      "eval_loss": 1.9800282716751099,
+      "eval_rouge1": 0.584603146586737,
+      "eval_rouge2": 0.34685127859250275,
+      "eval_rougeL": 0.5465551086322522,
+      "eval_runtime": 30.1539,
+      "eval_samples_per_second": 30.676,
+      "eval_steps_per_second": 3.847,
+      "step": 7376
+    },
+    {
+      "epoch": 16.268980477223426,
+      "grad_norm": 1.1474549770355225,
+      "learning_rate": 9.862385321100918e-06,
+      "loss": 1.4361,
+      "step": 7500
+    },
+    {
+      "epoch": 17.0,
+      "eval_bleu": 0.3150747205411821,
+      "eval_loss": 1.979936957359314,
+      "eval_rouge1": 0.5865159051716312,
+      "eval_rouge2": 0.34931764934488574,
+      "eval_rougeL": 0.5485201965669062,
+      "eval_runtime": 29.8082,
+      "eval_samples_per_second": 31.032,
+      "eval_steps_per_second": 3.892,
+      "step": 7837
+    },
+    {
+      "epoch": 17.35357917570499,
+      "grad_norm": 1.0869196653366089,
+      "learning_rate": 6.995412844036697e-06,
+      "loss": 1.4159,
+      "step": 8000
+    },
+    {
+      "epoch": 18.0,
+      "eval_bleu": 0.31486980402021325,
+      "eval_loss": 1.9808851480484009,
+      "eval_rouge1": 0.5865213969000711,
+      "eval_rouge2": 0.3495173970652049,
+      "eval_rougeL": 0.5486068806911688,
+      "eval_runtime": 29.8468,
+      "eval_samples_per_second": 30.992,
+      "eval_steps_per_second": 3.887,
+      "step": 8298
+    },
+    {
+      "epoch": 18.0,
+      "step": 8298,
+      "total_flos": 2.5969031184384e+16,
+      "train_loss": 1.6592627337077177,
+      "train_runtime": 3469.9556,
+      "train_samples_per_second": 21.216,
+      "train_steps_per_second": 2.657
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 9220,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 20,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 3,
+        "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": 2.5969031184384e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}