End of training

README.md

@@ -3,6 +3,9 @@ license: apache-2.0
 base_model: riotu-lab/ArabianGPT-01B
 tags:
 - generated_from_trainer
+metrics:
+- bleu
+- rouge
 model-index:
 - name: results
   results: []
@@ -14,6 +17,12 @@ should probably proofread and complete it, then remove this comment. -->
 # results
 
 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: 3.4630
+- Bleu: 0.0984
+- Rouge1: 0.3093
+- Rouge2: 0.0718
+- Rougel: 0.2296
 
 ## Model description
 

all_results.json

@@ -5,15 +5,15 @@
     "eval_rouge1": 0.3093411104418377,
     "eval_rouge2": 0.07177511381538038,
     "eval_rougeL": 0.22957881226204224,
-    "eval_runtime": 27.1725,
+    "eval_runtime": 27.2297,
     "eval_samples": 847,
-    "eval_samples_per_second": 31.171,
-    "eval_steps_per_second": 3.901,
+    "eval_samples_per_second": 31.106,
+    "eval_steps_per_second": 3.893,
     "perplexity": 31.912922646039927,
     "total_flos": 1000225898496000.0,
-    "train_loss": 4.835940561324452,
-    "train_runtime": 121.5981,
+    "train_loss": 0.0,
+    "train_runtime": 0.0037,
     "train_samples": 2552,
-    "train_samples_per_second": 20.987,
-    "train_steps_per_second": 2.623
+    "train_samples_per_second": 699005.016,
+    "train_steps_per_second": 87375.627
 }

egy_training_log.txt

@@ -287,3 +287,6 @@ WARNING:__main__:The tokenizer picked seems to have a very large `model_max_leng
 INFO:datasets.arrow_dataset:Loading cached processed dataset at /home/iais_marenpielka/.cache/huggingface/datasets/text/default-0637777c38512acf/0.0.0/96636a050ef51804b84abbfd4f4ad440e01153c24b86293eb5c3b300a41f9101/cache-607ae57e4b4160b3.arrow
 INFO:datasets.arrow_dataset:Loading cached processed dataset at /home/iais_marenpielka/.cache/huggingface/datasets/text/default-0637777c38512acf/0.0.0/96636a050ef51804b84abbfd4f4ad440e01153c24b86293eb5c3b300a41f9101/cache-4ddbb6e08bb37d3f.arrow
 WARNING:accelerate.utils.other:Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+INFO:__main__:*** Evaluate ***
+INFO:absl:Using default tokenizer.
+WARNING:root:No losses were recorded, so the loss graph was not generated.

eval_results.json

@@ -5,9 +5,9 @@
     "eval_rouge1": 0.3093411104418377,
     "eval_rouge2": 0.07177511381538038,
     "eval_rougeL": 0.22957881226204224,
-    "eval_runtime": 27.1725,
+    "eval_runtime": 27.2297,
     "eval_samples": 847,
-    "eval_samples_per_second": 31.171,
-    "eval_steps_per_second": 3.901,
+    "eval_samples_per_second": 31.106,
+    "eval_steps_per_second": 3.893,
     "perplexity": 31.912922646039927
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 1.0,
     "total_flos": 1000225898496000.0,
-    "train_loss": 4.835940561324452,
-    "train_runtime": 121.5981,
+    "train_loss": 0.0,
+    "train_runtime": 0.0037,
     "train_samples": 2552,
-    "train_samples_per_second": 20.987,
-    "train_steps_per_second": 2.623
+    "train_samples_per_second": 699005.016,
+    "train_steps_per_second": 87375.627
 }

trainer_state.json

@@ -12,10 +12,10 @@
       "epoch": 1.0,
       "step": 319,
       "total_flos": 1000225898496000.0,
-      "train_loss": 4.835940561324452,
-      "train_runtime": 121.5981,
-      "train_samples_per_second": 20.987,
-      "train_steps_per_second": 2.623
+      "train_loss": 0.0,
+      "train_runtime": 0.0037,
+      "train_samples_per_second": 699005.016,
+      "train_steps_per_second": 87375.627
     }
   ],
   "logging_steps": 500,