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: 0.4789
-- Bleu: 0.4179
-- Rouge1: 0.6343
-- Rouge2: 0.3810
-- Rougel: 0.6307
+- Loss: 0.4655
+- Bleu: 0.3947
+- Rouge1: 0.5909
+- Rouge2: 0.3232
+- Rougel: 0.5882
 
 ## Model description
 

all_results.json

@@ -1,19 +1,19 @@
 {
-    "epoch": 16.0,
-    "eval_bleu": 0.22970619705356748,
-    "eval_loss": 0.47202983498573303,
-    "eval_rouge1": 0.5777164933812552,
-    "eval_rouge2": 0.33405816844574837,
-    "eval_rougeL": 0.5758449342460217,
-    "eval_runtime": 3.3676,
+    "epoch": 11.0,
+    "eval_bleu": 0.39471976173966783,
+    "eval_loss": 0.465503990650177,
+    "eval_rouge1": 0.5909012601210781,
+    "eval_rouge2": 0.32323469643447167,
+    "eval_rougeL": 0.5882112866531675,
+    "eval_runtime": 1.2441,
     "eval_samples": 304,
-    "eval_samples_per_second": 90.273,
-    "eval_steps_per_second": 11.284,
-    "perplexity": 1.603245215811176,
-    "total_flos": 1275105116160000.0,
-    "train_loss": 0.024901744976542354,
-    "train_runtime": 209.6658,
-    "train_samples": 1220,
-    "train_samples_per_second": 116.376,
-    "train_steps_per_second": 14.595
+    "eval_samples_per_second": 244.35,
+    "eval_steps_per_second": 30.544,
+    "perplexity": 1.5928167513782572,
+    "total_flos": 875916214272000.0,
+    "train_loss": 0.9488435417714746,
+    "train_runtime": 1368.0254,
+    "train_samples": 1219,
+    "train_samples_per_second": 17.821,
+    "train_steps_per_second": 2.237
 }

egy_training_log.txt

@@ -168,3 +168,5 @@ INFO:root:Epoch 10.0: Train Loss = 0.1206, Eval Loss = 0.47646617889404297
 INFO:absl:Using default tokenizer.
 INFO:root:Epoch 11.0: Train Loss = 0.1052, Eval Loss = 0.47586962580680847
 INFO:absl:Using default tokenizer.
+INFO:__main__:*** Evaluate ***
+INFO:absl:Using default tokenizer.

eval_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 16.0,
-    "eval_bleu": 0.22970619705356748,
-    "eval_loss": 0.47202983498573303,
-    "eval_rouge1": 0.5777164933812552,
-    "eval_rouge2": 0.33405816844574837,
-    "eval_rougeL": 0.5758449342460217,
-    "eval_runtime": 3.3676,
+    "epoch": 11.0,
+    "eval_bleu": 0.39471976173966783,
+    "eval_loss": 0.465503990650177,
+    "eval_rouge1": 0.5909012601210781,
+    "eval_rouge2": 0.32323469643447167,
+    "eval_rougeL": 0.5882112866531675,
+    "eval_runtime": 1.2441,
     "eval_samples": 304,
-    "eval_samples_per_second": 90.273,
-    "eval_steps_per_second": 11.284,
-    "perplexity": 1.603245215811176
+    "eval_samples_per_second": 244.35,
+    "eval_steps_per_second": 30.544,
+    "perplexity": 1.5928167513782572
 }

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 16.0,
-    "total_flos": 1275105116160000.0,
-    "train_loss": 0.024901744976542354,
-    "train_runtime": 209.6658,
-    "train_samples": 1220,
-    "train_samples_per_second": 116.376,
-    "train_steps_per_second": 14.595
+    "epoch": 11.0,
+    "total_flos": 875916214272000.0,
+    "train_loss": 0.9488435417714746,
+    "train_runtime": 1368.0254,
+    "train_samples": 1219,
+    "train_samples_per_second": 17.821,
+    "train_steps_per_second": 2.237
 }

trainer_state.json

@@ -1,325 +1,230 @@
 {
-  "best_metric": 0.47202983498573303,
+  "best_metric": 0.465503990650177,
   "best_model_checkpoint": "/home/iais_marenpielka/Bouthaina/res_nw_yem/checkpoint-918",
-  "epoch": 16.0,
+  "epoch": 11.0,
   "eval_steps": 500,
-  "global_step": 2448,
+  "global_step": 1683,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 1.0,
-      "grad_norm": 43.000205993652344,
+      "grad_norm": 41.373477935791016,
       "learning_rate": 1.53e-05,
-      "loss": 7.8314,
+      "loss": 7.4125,
       "step": 153
     },
     {
       "epoch": 1.0,
-      "eval_bleu": 0.010356123410675664,
-      "eval_loss": 2.359743595123291,
-      "eval_rouge1": 0.09828786255795381,
-      "eval_rouge2": 0.0018177296310952823,
-      "eval_rougeL": 0.09639405964210626,
-      "eval_runtime": 1.1837,
-      "eval_samples_per_second": 256.818,
-      "eval_steps_per_second": 32.102,
+      "eval_bleu": 0.013470228801119116,
+      "eval_loss": 2.3318939208984375,
+      "eval_rouge1": 0.4042810146176442,
+      "eval_rouge2": 0.1765421862019638,
+      "eval_rougeL": 0.40385731535628305,
+      "eval_runtime": 1.2315,
+      "eval_samples_per_second": 246.862,
+      "eval_steps_per_second": 30.858,
       "step": 153
     },
     {
       "epoch": 2.0,
-      "grad_norm": 1.3453600406646729,
+      "grad_norm": 1.7725179195404053,
       "learning_rate": 3.06e-05,
-      "loss": 0.9995,
+      "loss": 1.0077,
       "step": 306
     },
     {
       "epoch": 2.0,
-      "eval_bleu": 0.11609954154539694,
-      "eval_loss": 0.5537932515144348,
-      "eval_rouge1": 0.47363801342590006,
-      "eval_rouge2": 0.18538347478382655,
-      "eval_rougeL": 0.4727595492611425,
-      "eval_runtime": 9.2767,
-      "eval_samples_per_second": 32.77,
-      "eval_steps_per_second": 4.096,
+      "eval_bleu": 0.3375849927940314,
+      "eval_loss": 0.5519431233406067,
+      "eval_rouge1": 0.47471630348624405,
+      "eval_rouge2": 0.18392227454023974,
+      "eval_rougeL": 0.4733153598700648,
+      "eval_runtime": 1.1576,
+      "eval_samples_per_second": 262.603,
+      "eval_steps_per_second": 32.825,
       "step": 306
     },
     {
       "epoch": 3.0,
-      "grad_norm": 1.2846729755401611,
+      "grad_norm": 1.682711124420166,
       "learning_rate": 4.5900000000000004e-05,
-      "loss": 0.4848,
+      "loss": 0.483,
       "step": 459
     },
     {
       "epoch": 3.0,
-      "eval_bleu": 0.14721751205352876,
-      "eval_loss": 0.5034472942352295,
-      "eval_rouge1": 0.500839694175635,
-      "eval_rouge2": 0.21735831386751384,
-      "eval_rougeL": 0.5001625449149274,
-      "eval_runtime": 10.2581,
-      "eval_samples_per_second": 29.635,
-      "eval_steps_per_second": 3.704,
+      "eval_bleu": 0.35824220855964284,
+      "eval_loss": 0.4972667694091797,
+      "eval_rouge1": 0.5145197705108664,
+      "eval_rouge2": 0.2200386914249186,
+      "eval_rougeL": 0.5134621462419481,
+      "eval_runtime": 1.2102,
+      "eval_samples_per_second": 251.205,
+      "eval_steps_per_second": 31.401,
       "step": 459
     },
     {
       "epoch": 4.0,
-      "grad_norm": 1.2576143741607666,
+      "grad_norm": 1.890738606452942,
       "learning_rate": 4.7812500000000003e-05,
-      "loss": 0.3823,
+      "loss": 0.3795,
       "step": 612
     },
     {
       "epoch": 4.0,
-      "eval_bleu": 0.1911807573581778,
-      "eval_loss": 0.4827471375465393,
-      "eval_rouge1": 0.5330693017126298,
-      "eval_rouge2": 0.27439720992270494,
-      "eval_rougeL": 0.531424425624079,
-      "eval_runtime": 5.4191,
-      "eval_samples_per_second": 56.098,
-      "eval_steps_per_second": 7.012,
+      "eval_bleu": 0.3717354832918087,
+      "eval_loss": 0.4739198088645935,
+      "eval_rouge1": 0.5397330658204715,
+      "eval_rouge2": 0.2524237751117604,
+      "eval_rougeL": 0.5377765269792509,
+      "eval_runtime": 1.2779,
+      "eval_samples_per_second": 237.89,
+      "eval_steps_per_second": 29.736,
       "step": 612
     },
     {
       "epoch": 5.0,
-      "grad_norm": 1.7599295377731323,
+      "grad_norm": 1.3704463243484497,
       "learning_rate": 4.482421875e-05,
-      "loss": 0.293,
+      "loss": 0.2901,
       "step": 765
     },
     {
       "epoch": 5.0,
-      "eval_bleu": 0.20959557277319146,
-      "eval_loss": 0.4732062816619873,
-      "eval_rouge1": 0.5619100572084703,
-      "eval_rouge2": 0.30991934212678557,
-      "eval_rougeL": 0.5587085347101918,
-      "eval_runtime": 2.8763,
-      "eval_samples_per_second": 105.69,
-      "eval_steps_per_second": 13.211,
+      "eval_bleu": 0.37954739911448565,
+      "eval_loss": 0.46911630034446716,
+      "eval_rouge1": 0.57101655669895,
+      "eval_rouge2": 0.29268127822000256,
+      "eval_rougeL": 0.567665717772471,
+      "eval_runtime": 1.1189,
+      "eval_samples_per_second": 271.684,
+      "eval_steps_per_second": 33.96,
       "step": 765
     },
     {
       "epoch": 6.0,
-      "grad_norm": 1.4758834838867188,
+      "grad_norm": 2.911377429962158,
       "learning_rate": 4.18359375e-05,
       "loss": 0.2239,
       "step": 918
     },
     {
       "epoch": 6.0,
-      "eval_bleu": 0.22970619705356748,
-      "eval_loss": 0.47202983498573303,
-      "eval_rouge1": 0.5777164933812552,
-      "eval_rouge2": 0.33405816844574837,
-      "eval_rougeL": 0.5758449342460217,
-      "eval_runtime": 1.7018,
-      "eval_samples_per_second": 178.629,
-      "eval_steps_per_second": 22.329,
+      "eval_bleu": 0.39471976173966783,
+      "eval_loss": 0.465503990650177,
+      "eval_rouge1": 0.5909012601210781,
+      "eval_rouge2": 0.32323469643447167,
+      "eval_rougeL": 0.5882112866531675,
+      "eval_runtime": 5.248,
+      "eval_samples_per_second": 57.927,
+      "eval_steps_per_second": 7.241,
       "step": 918
     },
     {
       "epoch": 7.0,
-      "grad_norm": 1.1557574272155762,
+      "grad_norm": 1.639866590499878,
       "learning_rate": 3.884765625e-05,
-      "loss": 0.1766,
+      "loss": 0.1763,
       "step": 1071
     },
     {
       "epoch": 7.0,
-      "eval_bleu": 0.23015149161791434,
-      "eval_loss": 0.47373390197753906,
-      "eval_rouge1": 0.588487328464713,
-      "eval_rouge2": 0.3467647604638442,
-      "eval_rougeL": 0.5872191002290259,
-      "eval_runtime": 19.3271,
-      "eval_samples_per_second": 15.729,
-      "eval_steps_per_second": 1.966,
+      "eval_bleu": 0.4011122295711346,
+      "eval_loss": 0.46726807951927185,
+      "eval_rouge1": 0.6046871760857919,
+      "eval_rouge2": 0.3378049073324742,
+      "eval_rougeL": 0.6018660127815343,
+      "eval_runtime": 14.6215,
+      "eval_samples_per_second": 20.791,
+      "eval_steps_per_second": 2.599,
       "step": 1071
     },
     {
       "epoch": 8.0,
-      "grad_norm": 1.236377477645874,
+      "grad_norm": 1.1884260177612305,
       "learning_rate": 3.5859375e-05,
-      "loss": 0.1434,
+      "loss": 0.1429,
       "step": 1224
     },
     {
       "epoch": 8.0,
-      "eval_bleu": 0.24506232565608202,
-      "eval_loss": 0.47583380341529846,
-      "eval_rouge1": 0.5938395263674476,
-      "eval_rouge2": 0.3647516376509473,
-      "eval_rougeL": 0.5903133777624114,
-      "eval_runtime": 12.6597,
-      "eval_samples_per_second": 24.013,
-      "eval_steps_per_second": 3.002,
+      "eval_bleu": 0.4084847317717048,
+      "eval_loss": 0.46664053201675415,
+      "eval_rouge1": 0.619977821151152,
+      "eval_rouge2": 0.3652452050559121,
+      "eval_rougeL": 0.6168287532556173,
+      "eval_runtime": 13.3071,
+      "eval_samples_per_second": 22.845,
+      "eval_steps_per_second": 2.856,
       "step": 1224
     },
     {
       "epoch": 9.0,
-      "grad_norm": 1.2275965213775635,
+      "grad_norm": 1.133341908454895,
       "learning_rate": 3.287109375e-05,
-      "loss": 0.1202,
+      "loss": 0.1206,
       "step": 1377
     },
     {
       "epoch": 9.0,
-      "eval_bleu": 0.2508515461433435,
-      "eval_loss": 0.4827924072742462,
-      "eval_rouge1": 0.6046956358537867,
-      "eval_rouge2": 0.3682856780453533,
-      "eval_rougeL": 0.6015063720236549,
-      "eval_runtime": 30.4628,
-      "eval_samples_per_second": 9.979,
-      "eval_steps_per_second": 1.247,
+      "eval_bleu": 0.412982833438841,
+      "eval_loss": 0.47646617889404297,
+      "eval_rouge1": 0.6230350047425831,
+      "eval_rouge2": 0.37065368099082896,
+      "eval_rougeL": 0.6207890023241163,
+      "eval_runtime": 4.1195,
+      "eval_samples_per_second": 73.796,
+      "eval_steps_per_second": 9.225,
       "step": 1377
     },
     {
       "epoch": 10.0,
-      "grad_norm": 0.9903411865234375,
+      "grad_norm": 1.0680757761001587,
       "learning_rate": 2.9882812500000002e-05,
-      "loss": 0.1042,
+      "loss": 0.1052,
       "step": 1530
     },
     {
       "epoch": 10.0,
-      "eval_bleu": 0.24386094085391116,
-      "eval_loss": 0.4852388799190521,
-      "eval_rouge1": 0.5980129828644593,
-      "eval_rouge2": 0.3703776134768108,
-      "eval_rougeL": 0.5951273362785527,
-      "eval_runtime": 23.4553,
-      "eval_samples_per_second": 12.961,
-      "eval_steps_per_second": 1.62,
+      "eval_bleu": 0.41481558555871684,
+      "eval_loss": 0.47586962580680847,
+      "eval_rouge1": 0.6232860618818599,
+      "eval_rouge2": 0.3762342947707073,
+      "eval_rougeL": 0.6204668025380538,
+      "eval_runtime": 1.7191,
+      "eval_samples_per_second": 176.836,
+      "eval_steps_per_second": 22.105,
       "step": 1530
     },
     {
       "epoch": 11.0,
-      "grad_norm": 1.104041337966919,
+      "grad_norm": 1.2003881931304932,
       "learning_rate": 2.689453125e-05,
       "loss": 0.0955,
       "step": 1683
     },
     {
       "epoch": 11.0,
-      "eval_bleu": 0.25597221159625966,
-      "eval_loss": 0.4884573221206665,
-      "eval_rouge1": 0.6115922153776887,
-      "eval_rouge2": 0.37952408975600893,
-      "eval_rougeL": 0.6085537919616412,
-      "eval_runtime": 1.297,
-      "eval_samples_per_second": 234.387,
-      "eval_steps_per_second": 29.298,
+      "eval_bleu": 0.4179443502693027,
+      "eval_loss": 0.47885996103286743,
+      "eval_rouge1": 0.6342804965319174,
+      "eval_rouge2": 0.3809817493272587,
+      "eval_rougeL": 0.6306668222386218,
+      "eval_runtime": 1.9634,
+      "eval_samples_per_second": 154.831,
+      "eval_steps_per_second": 19.354,
       "step": 1683
     },
     {
-      "epoch": 12.0,
-      "grad_norm": 0.6220578551292419,
-      "learning_rate": 2.3906250000000002e-05,
-      "loss": 0.0875,
-      "step": 1836
-    },
-    {
-      "epoch": 12.0,
-      "eval_bleu": 0.25506926105311084,
-      "eval_loss": 0.4913596212863922,
-      "eval_rouge1": 0.6097696683692526,
-      "eval_rouge2": 0.38019766139150524,
-      "eval_rougeL": 0.6055686891522601,
-      "eval_runtime": 2.0886,
-      "eval_samples_per_second": 145.549,
-      "eval_steps_per_second": 18.194,
-      "step": 1836
-    },
-    {
-      "epoch": 13.0,
-      "grad_norm": 0.7318098545074463,
-      "learning_rate": 2.091796875e-05,
-      "loss": 0.0825,
-      "step": 1989
-    },
-    {
-      "epoch": 13.0,
-      "eval_bleu": 0.26272086017465046,
-      "eval_loss": 0.4981193542480469,
-      "eval_rouge1": 0.6144877321856528,
-      "eval_rouge2": 0.3872195934378641,
-      "eval_rougeL": 0.6108120100650027,
-      "eval_runtime": 1.1422,
-      "eval_samples_per_second": 266.148,
-      "eval_steps_per_second": 33.268,
-      "step": 1989
-    },
-    {
-      "epoch": 14.0,
-      "grad_norm": 0.794402539730072,
-      "learning_rate": 1.79296875e-05,
-      "loss": 0.0787,
-      "step": 2142
-    },
-    {
-      "epoch": 14.0,
-      "eval_bleu": 0.25895474712357536,
-      "eval_loss": 0.5004830360412598,
-      "eval_rouge1": 0.6138305476317043,
-      "eval_rouge2": 0.3796550798534031,
-      "eval_rougeL": 0.6100541920692137,
-      "eval_runtime": 1.1033,
-      "eval_samples_per_second": 275.529,
-      "eval_steps_per_second": 34.441,
-      "step": 2142
-    },
-    {
-      "epoch": 15.0,
-      "grad_norm": 0.7674385905265808,
-      "learning_rate": 1.4941406250000001e-05,
-      "loss": 0.0757,
-      "step": 2295
-    },
-    {
-      "epoch": 15.0,
-      "eval_bleu": 0.2632177488955002,
-      "eval_loss": 0.5019999146461487,
-      "eval_rouge1": 0.6170590027553351,
-      "eval_rouge2": 0.3850023586523127,
-      "eval_rougeL": 0.6140574753716048,
-      "eval_runtime": 4.8266,
-      "eval_samples_per_second": 62.985,
-      "eval_steps_per_second": 7.873,
-      "step": 2295
-    },
-    {
-      "epoch": 16.0,
-      "grad_norm": 0.7136771082878113,
-      "learning_rate": 1.1953125000000001e-05,
-      "loss": 0.074,
-      "step": 2448
-    },
-    {
-      "epoch": 16.0,
-      "eval_bleu": 0.25943804422988964,
-      "eval_loss": 0.5049206018447876,
-      "eval_rouge1": 0.6161012779292924,
-      "eval_rouge2": 0.3828890396119353,
-      "eval_rougeL": 0.6124852816095869,
-      "eval_runtime": 1.0975,
-      "eval_samples_per_second": 276.998,
-      "eval_steps_per_second": 34.625,
-      "step": 2448
-    },
-    {
-      "epoch": 16.0,
-      "step": 2448,
-      "total_flos": 1275105116160000.0,
-      "train_loss": 0.024901744976542354,
-      "train_runtime": 209.6658,
-      "train_samples_per_second": 116.376,
-      "train_steps_per_second": 14.595
+      "epoch": 11.0,
+      "step": 1683,
+      "total_flos": 875916214272000.0,
+      "train_loss": 0.9488435417714746,
+      "train_runtime": 1368.0254,
+      "train_samples_per_second": 17.821,
+      "train_steps_per_second": 2.237
     }
   ],
   "logging_steps": 500,
@@ -348,7 +253,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1275105116160000.0,
+  "total_flos": 875916214272000.0,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null