End of training

README.md

@@ -3,6 +3,8 @@ license: apache-2.0
 base_model: riotu-lab/ArabianGPT-01B
 tags:
 - generated_from_trainer
+metrics:
+- accuracy
 model-index:
 - name: test-aragpt
   results: []
@@ -14,6 +16,9 @@ should probably proofread and complete it, then remove this comment. -->
 # test-aragpt
 
 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: 2.8715
+- Accuracy: 0.4496
 
 ## Model description
 

all_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 3.0,
-    "eval_accuracy": 0.2540569865270625,
-    "eval_loss": 4.338694095611572,
-    "eval_runtime": 27.7402,
-    "eval_samples": 1627,
-    "eval_samples_per_second": 58.651,
-    "eval_steps_per_second": 7.354,
-    "perplexity": 76.60743200557758,
-    "total_flos": 534995435520000.0,
-    "train_loss": 4.0715499788697,
-    "train_runtime": 64.1161,
-    "train_samples": 455,
-    "train_samples_per_second": 21.29,
-    "train_steps_per_second": 2.667
+    "eval_accuracy": 0.4495550674287192,
+    "eval_loss": 2.871492624282837,
+    "eval_runtime": 137.8202,
+    "eval_samples": 847,
+    "eval_samples_per_second": 6.146,
+    "eval_steps_per_second": 0.769,
+    "perplexity": 17.663363298254804,
+    "total_flos": 3000677695488000.0,
+    "train_loss": 3.2309236117914053,
+    "train_runtime": 357.8599,
+    "train_samples": 2552,
+    "train_samples_per_second": 21.394,
+    "train_steps_per_second": 2.674
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 3.0,
-    "eval_accuracy": 0.2540569865270625,
-    "eval_loss": 4.338694095611572,
-    "eval_runtime": 27.7402,
-    "eval_samples": 1627,
-    "eval_samples_per_second": 58.651,
-    "eval_steps_per_second": 7.354,
-    "perplexity": 76.60743200557758
+    "eval_accuracy": 0.4495550674287192,
+    "eval_loss": 2.871492624282837,
+    "eval_runtime": 137.8202,
+    "eval_samples": 847,
+    "eval_samples_per_second": 6.146,
+    "eval_steps_per_second": 0.769,
+    "perplexity": 17.663363298254804
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 3.0,
-    "total_flos": 534995435520000.0,
-    "train_loss": 4.0715499788697,
-    "train_runtime": 64.1161,
-    "train_samples": 455,
-    "train_samples_per_second": 21.29,
-    "train_steps_per_second": 2.667
+    "total_flos": 3000677695488000.0,
+    "train_loss": 3.2309236117914053,
+    "train_runtime": 357.8599,
+    "train_samples": 2552,
+    "train_samples_per_second": 21.394,
+    "train_steps_per_second": 2.674
 }

trainer_state.json

@@ -3,23 +3,30 @@
   "best_model_checkpoint": null,
   "epoch": 3.0,
   "eval_steps": 500,
-  "global_step": 171,
+  "global_step": 957,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
+    {
+      "epoch": 1.567398119122257,
+      "grad_norm": 1.1585382223129272,
+      "learning_rate": 2.387669801462905e-05,
+      "loss": 3.5029,
+      "step": 500
+    },
     {
       "epoch": 3.0,
-      "step": 171,
-      "total_flos": 534995435520000.0,
-      "train_loss": 4.0715499788697,
-      "train_runtime": 64.1161,
-      "train_samples_per_second": 21.29,
-      "train_steps_per_second": 2.667
+      "step": 957,
+      "total_flos": 3000677695488000.0,
+      "train_loss": 3.2309236117914053,
+      "train_runtime": 357.8599,
+      "train_samples_per_second": 21.394,
+      "train_steps_per_second": 2.674
     }
   ],
   "logging_steps": 500,
-  "max_steps": 171,
+  "max_steps": 957,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 3,
   "save_steps": 500,
@@ -35,7 +42,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 534995435520000.0,
+  "total_flos": 3000677695488000.0,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null