End of training

README.md

@@ -3,6 +3,8 @@ license: apache-2.0
 base_model: riotu-lab/ArabianGPT-01B
 tags:
 - generated_from_trainer
+metrics:
+- bleu
 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: 2.2835
+- Bleu: 0.2080
 
 ## Model description
 

all_results.json

@@ -1,16 +1,16 @@
 {
-    "epoch": 1.0,
-    "eval_bleu": 0.10603800831832183,
-    "eval_loss": 3.326219081878662,
-    "eval_runtime": 18.595,
+    "epoch": 15.0,
+    "eval_bleu": 0.20804732558419864,
+    "eval_loss": 2.2835311889648438,
+    "eval_runtime": 19.725,
     "eval_samples": 847,
-    "eval_samples_per_second": 45.55,
-    "eval_steps_per_second": 5.7,
-    "perplexity": 27.832908563445223,
-    "total_flos": 1000225898496000.0,
-    "train_loss": 3.8128880228742164,
-    "train_runtime": 119.3474,
+    "eval_samples_per_second": 42.94,
+    "eval_steps_per_second": 5.374,
+    "perplexity": 9.811264741472648,
+    "total_flos": 1.500338847744e+16,
+    "train_loss": 2.280068150184496,
+    "train_runtime": 1782.6608,
     "train_samples": 2552,
-    "train_samples_per_second": 21.383,
-    "train_steps_per_second": 2.673
+    "train_samples_per_second": 21.474,
+    "train_steps_per_second": 2.684
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
-    "epoch": 1.0,
-    "eval_bleu": 0.10603800831832183,
-    "eval_loss": 3.326219081878662,
-    "eval_runtime": 18.595,
+    "epoch": 15.0,
+    "eval_bleu": 0.20804732558419864,
+    "eval_loss": 2.2835311889648438,
+    "eval_runtime": 19.725,
     "eval_samples": 847,
-    "eval_samples_per_second": 45.55,
-    "eval_steps_per_second": 5.7,
-    "perplexity": 27.832908563445223
+    "eval_samples_per_second": 42.94,
+    "eval_steps_per_second": 5.374,
+    "perplexity": 9.811264741472648
 }

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 1.0,
-    "total_flos": 1000225898496000.0,
-    "train_loss": 3.8128880228742164,
-    "train_runtime": 119.3474,
+    "epoch": 15.0,
+    "total_flos": 1.500338847744e+16,
+    "train_loss": 2.280068150184496,
+    "train_runtime": 1782.6608,
     "train_samples": 2552,
-    "train_samples_per_second": 21.383,
-    "train_steps_per_second": 2.673
+    "train_samples_per_second": 21.474,
+    "train_steps_per_second": 2.684
 }

trainer_state.json

@@ -1,27 +1,90 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0,
+  "epoch": 15.0,
   "eval_steps": 500,
-  "global_step": 319,
+  "global_step": 4785,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 1.0,
-      "step": 319,
-      "total_flos": 1000225898496000.0,
-      "train_loss": 3.8128880228742164,
-      "train_runtime": 119.3474,
-      "train_samples_per_second": 21.383,
-      "train_steps_per_second": 2.673
+      "epoch": 1.567398119122257,
+      "grad_norm": 1.1618669033050537,
+      "learning_rate": 4.4775339602925815e-05,
+      "loss": 3.4578,
+      "step": 500
+    },
+    {
+      "epoch": 3.134796238244514,
+      "grad_norm": 1.134594440460205,
+      "learning_rate": 3.955067920585162e-05,
+      "loss": 2.7316,
+      "step": 1000
+    },
+    {
+      "epoch": 4.702194357366771,
+      "grad_norm": 1.16652250289917,
+      "learning_rate": 3.4326018808777435e-05,
+      "loss": 2.4393,
+      "step": 1500
+    },
+    {
+      "epoch": 6.269592476489028,
+      "grad_norm": 1.1664645671844482,
+      "learning_rate": 2.9101358411703238e-05,
+      "loss": 2.2559,
+      "step": 2000
+    },
+    {
+      "epoch": 7.836990595611285,
+      "grad_norm": 1.1747838258743286,
+      "learning_rate": 2.387669801462905e-05,
+      "loss": 2.127,
+      "step": 2500
+    },
+    {
+      "epoch": 9.404388714733543,
+      "grad_norm": 1.2233911752700806,
+      "learning_rate": 1.865203761755486e-05,
+      "loss": 2.0242,
+      "step": 3000
+    },
+    {
+      "epoch": 10.971786833855798,
+      "grad_norm": 1.149554967880249,
+      "learning_rate": 1.3427377220480669e-05,
+      "loss": 1.9585,
+      "step": 3500
+    },
+    {
+      "epoch": 12.539184952978056,
+      "grad_norm": 1.1873332262039185,
+      "learning_rate": 8.202716823406478e-06,
+      "loss": 1.9004,
+      "step": 4000
+    },
+    {
+      "epoch": 14.106583072100314,
+      "grad_norm": 1.1350129842758179,
+      "learning_rate": 2.9780564263322885e-06,
+      "loss": 1.8692,
+      "step": 4500
+    },
+    {
+      "epoch": 15.0,
+      "step": 4785,
+      "total_flos": 1.500338847744e+16,
+      "train_loss": 2.280068150184496,
+      "train_runtime": 1782.6608,
+      "train_samples_per_second": 21.474,
+      "train_steps_per_second": 2.684
     }
   ],
   "logging_steps": 500,
-  "max_steps": 319,
+  "max_steps": 4785,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
+  "num_train_epochs": 15,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -35,7 +98,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1000225898496000.0,
+  "total_flos": 1.500338847744e+16,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null