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-03B](https://huggingface.co/riotu-lab/ArabianGPT-03B) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.4050
-- Bleu: 0.5608
-- Rouge1: 0.7770
-- Rouge2: 0.5994
-- Rougel: 0.7763
+- Loss: 0.3660
+- Bleu: 0.4889
+- Rouge1: 0.7442
+- Rouge2: 0.5302
+- Rougel: 0.7434
 
 ## Model description
 

all_results.json

@@ -0,0 +1,19 @@
+{
+    "epoch": 8.0,
+    "eval_bleu": 0.48887049804156,
+    "eval_loss": 0.3660341203212738,
+    "eval_rouge1": 0.7441728211754235,
+    "eval_rouge2": 0.5301574363836805,
+    "eval_rougeL": 0.743357806649209,
+    "eval_runtime": 74.2126,
+    "eval_samples": 10118,
+    "eval_samples_per_second": 136.338,
+    "eval_steps_per_second": 17.046,
+    "perplexity": 1.442004443470027,
+    "total_flos": 7.520432484148838e+16,
+    "train_loss": 0.23126965857739035,
+    "train_runtime": 11366.9663,
+    "train_samples": 40489,
+    "train_samples_per_second": 71.24,
+    "train_steps_per_second": 8.907
+}

egy_training_log.txt

@@ -156,3 +156,5 @@ INFO:root:Epoch 7.0: Train Loss = 0.1253, Eval Loss = 0.3939472734928131
 INFO:absl:Using default tokenizer.
 INFO:root:Epoch 8.0: Train Loss = 0.1166, Eval Loss = 0.3990260660648346
 INFO:absl:Using default tokenizer.
+INFO:__main__:*** Evaluate ***
+INFO:absl:Using default tokenizer.

eval_results.json

@@ -0,0 +1,13 @@
+{
+    "epoch": 8.0,
+    "eval_bleu": 0.48887049804156,
+    "eval_loss": 0.3660341203212738,
+    "eval_rouge1": 0.7441728211754235,
+    "eval_rouge2": 0.5301574363836805,
+    "eval_rougeL": 0.743357806649209,
+    "eval_runtime": 74.2126,
+    "eval_samples": 10118,
+    "eval_samples_per_second": 136.338,
+    "eval_steps_per_second": 17.046,
+    "perplexity": 1.442004443470027
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 8.0,
+    "total_flos": 7.520432484148838e+16,
+    "train_loss": 0.23126965857739035,
+    "train_runtime": 11366.9663,
+    "train_samples": 40489,
+    "train_samples_per_second": 71.24,
+    "train_steps_per_second": 8.907
+}

trainer_state.json

@@ -0,0 +1,203 @@
+{
+  "best_metric": 0.3660341203212738,
+  "best_model_checkpoint": "/home/iais_marenpielka/Bouthaina/res_nw_lev_03/checkpoint-15186",
+  "epoch": 8.0,
+  "eval_steps": 500,
+  "global_step": 40496,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.426126956939697,
+      "learning_rate": 4.773575540996625e-05,
+      "loss": 0.6345,
+      "step": 5062
+    },
+    {
+      "epoch": 1.0,
+      "eval_bleu": 0.4054405178422978,
+      "eval_loss": 0.424884170293808,
+      "eval_rouge1": 0.6718054932020717,
+      "eval_rouge2": 0.4168304983667167,
+      "eval_rougeL": 0.6705720028130773,
+      "eval_runtime": 99.0407,
+      "eval_samples_per_second": 102.16,
+      "eval_steps_per_second": 12.773,
+      "step": 5062
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 3.7220001220703125,
+      "learning_rate": 4.522334723049435e-05,
+      "loss": 0.3258,
+      "step": 10124
+    },
+    {
+      "epoch": 2.0,
+      "eval_bleu": 0.4443048022041477,
+      "eval_loss": 0.38237395882606506,
+      "eval_rouge1": 0.7134969978151362,
+      "eval_rouge2": 0.4799759371169404,
+      "eval_rougeL": 0.7123642730734705,
+      "eval_runtime": 184.2876,
+      "eval_samples_per_second": 54.903,
+      "eval_steps_per_second": 6.864,
+      "step": 10124
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 4.069755554199219,
+      "learning_rate": 4.271093905102244e-05,
+      "loss": 0.2266,
+      "step": 15186
+    },
+    {
+      "epoch": 3.0,
+      "eval_bleu": 0.48887049804156,
+      "eval_loss": 0.3660341203212738,
+      "eval_rouge1": 0.7441728211754235,
+      "eval_rouge2": 0.5301574363836805,
+      "eval_rougeL": 0.743357806649209,
+      "eval_runtime": 197.651,
+      "eval_samples_per_second": 51.191,
+      "eval_steps_per_second": 6.4,
+      "step": 15186
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 2.417670488357544,
+      "learning_rate": 4.0198530871550526e-05,
+      "loss": 0.1702,
+      "step": 20248
+    },
+    {
+      "epoch": 4.0,
+      "eval_bleu": 0.5217066405916664,
+      "eval_loss": 0.37126386165618896,
+      "eval_rouge1": 0.7592034964866012,
+      "eval_rouge2": 0.5617725815136836,
+      "eval_rougeL": 0.7583714246633779,
+      "eval_runtime": 197.8347,
+      "eval_samples_per_second": 51.144,
+      "eval_steps_per_second": 6.394,
+      "step": 20248
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 2.2886266708374023,
+      "learning_rate": 3.7686122692078615e-05,
+      "loss": 0.1405,
+      "step": 25310
+    },
+    {
+      "epoch": 5.0,
+      "eval_bleu": 0.5410963843168277,
+      "eval_loss": 0.38092052936553955,
+      "eval_rouge1": 0.7670677626194377,
+      "eval_rouge2": 0.5785130119272781,
+      "eval_rougeL": 0.766336455280991,
+      "eval_runtime": 74.4752,
+      "eval_samples_per_second": 135.857,
+      "eval_steps_per_second": 16.986,
+      "step": 25310
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 1.4138972759246826,
+      "learning_rate": 3.517371451260672e-05,
+      "loss": 0.1253,
+      "step": 30372
+    },
+    {
+      "epoch": 6.0,
+      "eval_bleu": 0.5538973347056134,
+      "eval_loss": 0.3939472734928131,
+      "eval_rouge1": 0.7718106498807895,
+      "eval_rouge2": 0.5899343275061358,
+      "eval_rougeL": 0.7710407354471525,
+      "eval_runtime": 197.8643,
+      "eval_samples_per_second": 51.136,
+      "eval_steps_per_second": 6.393,
+      "step": 30372
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 1.0359286069869995,
+      "learning_rate": 3.266130633313481e-05,
+      "loss": 0.1166,
+      "step": 35434
+    },
+    {
+      "epoch": 7.0,
+      "eval_bleu": 0.5535836812783581,
+      "eval_loss": 0.3990260660648346,
+      "eval_rouge1": 0.7744583802216872,
+      "eval_rouge2": 0.5941788640581145,
+      "eval_rougeL": 0.7736991376335538,
+      "eval_runtime": 197.6914,
+      "eval_samples_per_second": 51.181,
+      "eval_steps_per_second": 6.399,
+      "step": 35434
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 3.439483165740967,
+      "learning_rate": 3.0148898153662896e-05,
+      "loss": 0.1107,
+      "step": 40496
+    },
+    {
+      "epoch": 8.0,
+      "eval_bleu": 0.5607932303098884,
+      "eval_loss": 0.4049707055091858,
+      "eval_rouge1": 0.7769922891885475,
+      "eval_rouge2": 0.5994002303095487,
+      "eval_rougeL": 0.7762651005907448,
+      "eval_runtime": 197.4325,
+      "eval_samples_per_second": 51.248,
+      "eval_steps_per_second": 6.407,
+      "step": 40496
+    },
+    {
+      "epoch": 8.0,
+      "step": 40496,
+      "total_flos": 7.520432484148838e+16,
+      "train_loss": 0.23126965857739035,
+      "train_runtime": 11366.9663,
+      "train_samples_per_second": 71.24,
+      "train_steps_per_second": 8.907
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 101240,
+  "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": 7.520432484148838e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}