End of training

all_results.json

@@ -1,12 +1,12 @@
 {
     "epoch": 100.0,
-    "eval_loss": 5.409937381744385,
-    "eval_runtime": 95.2682,
-    "eval_samples_per_second": 317.031,
-    "eval_steps_per_second": 7.211,
-    "total_flos": 6.859464121840128e+17,
-    "train_loss": 0.11642176906541846,
-    "train_runtime": 15360.6557,
-    "train_samples_per_second": 113.036,
-    "train_steps_per_second": 2.832
+    "eval_loss": 5.574370384216309,
+    "eval_runtime": 30.8156,
+    "eval_samples_per_second": 315.262,
+    "eval_steps_per_second": 7.172,
+    "total_flos": 2.3359629906138624e+18,
+    "train_loss": 0.060914613879798955,
+    "train_runtime": 49943.5819,
+    "train_samples_per_second": 118.392,
+    "train_steps_per_second": 2.961
 }

eval_results.json

@@ -1,6 +1,7 @@
 {
-    "eval_loss": 5.409937381744385,
-    "eval_runtime": 95.2682,
-    "eval_samples_per_second": 317.031,
-    "eval_steps_per_second": 7.211
+    "epoch": 100.0,
+    "eval_loss": 5.574370384216309,
+    "eval_runtime": 30.8156,
+    "eval_samples_per_second": 315.262,
+    "eval_steps_per_second": 7.172
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 100.0,
-    "total_flos": 6.859464121840128e+17,
-    "train_loss": 0.11642176906541846,
-    "train_runtime": 15360.6557,
-    "train_samples_per_second": 113.036,
-    "train_steps_per_second": 2.832
+    "total_flos": 2.3359629906138624e+18,
+    "train_loss": 0.060914613879798955,
+    "train_runtime": 49943.5819,
+    "train_samples_per_second": 118.392,
+    "train_steps_per_second": 2.961
 }

trainer_state.json

@@ -2,174 +2,174 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 100.0,
-  "eval_steps": 4350,
-  "global_step": 43500,
+  "eval_steps": 14790,
+  "global_step": 147900,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 10.0,
-      "grad_norm": 0.00021520511654671282,
-      "learning_rate": 9.000919540229886e-06,
-      "loss": 0.48,
-      "step": 4350
+      "grad_norm": 0.06951820850372314,
+      "learning_rate": 9.00054090601758e-06,
+      "loss": 0.2989,
+      "step": 14790
     },
     {
       "epoch": 10.0,
-      "eval_loss": 3.651794195175171,
-      "eval_runtime": 55.7113,
-      "eval_samples_per_second": 311.66,
-      "eval_steps_per_second": 7.09,
-      "step": 4350
+      "eval_loss": 6.212554454803467,
+      "eval_runtime": 32.6367,
+      "eval_samples_per_second": 297.671,
+      "eval_steps_per_second": 6.772,
+      "step": 14790
     },
     {
       "epoch": 20.0,
-      "grad_norm": 0.00017535020015202463,
-      "learning_rate": 8.001379310344829e-06,
-      "loss": 0.1193,
-      "step": 8700
+      "grad_norm": 0.0010786975035443902,
+      "learning_rate": 8.001014198782963e-06,
+      "loss": 0.0626,
+      "step": 29580
     },
     {
       "epoch": 20.0,
-      "eval_loss": 3.453643798828125,
-      "eval_runtime": 55.7389,
-      "eval_samples_per_second": 311.506,
-      "eval_steps_per_second": 7.087,
-      "step": 8700
+      "eval_loss": 6.155207633972168,
+      "eval_runtime": 30.9179,
+      "eval_samples_per_second": 314.22,
+      "eval_steps_per_second": 7.148,
+      "step": 29580
     },
     {
       "epoch": 30.0,
-      "grad_norm": 0.18407990038394928,
-      "learning_rate": 7.00183908045977e-06,
-      "loss": 0.0925,
-      "step": 13050
+      "grad_norm": 0.027268769219517708,
+      "learning_rate": 7.001487491548344e-06,
+      "loss": 0.0467,
+      "step": 44370
     },
     {
       "epoch": 30.0,
-      "eval_loss": 3.4169740676879883,
-      "eval_runtime": 55.9578,
-      "eval_samples_per_second": 310.287,
-      "eval_steps_per_second": 7.059,
-      "step": 13050
+      "eval_loss": 6.024782180786133,
+      "eval_runtime": 30.7841,
+      "eval_samples_per_second": 315.585,
+      "eval_steps_per_second": 7.179,
+      "step": 44370
     },
     {
       "epoch": 40.0,
-      "grad_norm": 8.286705269711092e-05,
-      "learning_rate": 6.002298850574713e-06,
-      "loss": 0.0827,
-      "step": 17400
+      "grad_norm": 0.0007825565990060568,
+      "learning_rate": 6.001893171061528e-06,
+      "loss": 0.0383,
+      "step": 59160
     },
     {
       "epoch": 40.0,
-      "eval_loss": 3.3203234672546387,
-      "eval_runtime": 55.9217,
-      "eval_samples_per_second": 310.488,
-      "eval_steps_per_second": 7.063,
-      "step": 17400
+      "eval_loss": 6.02599573135376,
+      "eval_runtime": 30.7709,
+      "eval_samples_per_second": 315.721,
+      "eval_steps_per_second": 7.182,
+      "step": 59160
     },
     {
       "epoch": 50.0,
-      "grad_norm": 0.0003959204477723688,
-      "learning_rate": 5.002758620689656e-06,
-      "loss": 0.0755,
-      "step": 21750
+      "grad_norm": NaN,
+      "learning_rate": 5.002434077079108e-06,
+      "loss": 0.0333,
+      "step": 73950
     },
     {
       "epoch": 50.0,
-      "eval_loss": 3.196115732192993,
-      "eval_runtime": 55.4125,
-      "eval_samples_per_second": 313.341,
-      "eval_steps_per_second": 7.128,
-      "step": 21750
+      "eval_loss": 5.985586643218994,
+      "eval_runtime": 30.6402,
+      "eval_samples_per_second": 317.068,
+      "eval_steps_per_second": 7.213,
+      "step": 73950
     },
     {
       "epoch": 60.0,
-      "grad_norm": 7.149695011321455e-05,
-      "learning_rate": 4.003218390804598e-06,
-      "loss": 0.0685,
-      "step": 26100
+      "grad_norm": 8.283340866910294e-05,
+      "learning_rate": 4.002772143340095e-06,
+      "loss": 0.0301,
+      "step": 88740
     },
     {
       "epoch": 60.0,
-      "eval_loss": 3.110957145690918,
-      "eval_runtime": 55.9729,
-      "eval_samples_per_second": 310.204,
-      "eval_steps_per_second": 7.057,
-      "step": 26100
+      "eval_loss": 5.848947525024414,
+      "eval_runtime": 30.8176,
+      "eval_samples_per_second": 315.242,
+      "eval_steps_per_second": 7.171,
+      "step": 88740
     },
     {
       "epoch": 70.0,
-      "grad_norm": 0.013322222046554089,
-      "learning_rate": 3.0036781609195404e-06,
-      "loss": 0.0664,
-      "step": 30450
+      "grad_norm": 0.0011442904360592365,
+      "learning_rate": 3.0033130493576744e-06,
+      "loss": 0.0275,
+      "step": 103530
     },
     {
       "epoch": 70.0,
-      "eval_loss": 3.0240492820739746,
-      "eval_runtime": 56.1145,
-      "eval_samples_per_second": 309.421,
-      "eval_steps_per_second": 7.039,
-      "step": 30450
+      "eval_loss": 5.8452348709106445,
+      "eval_runtime": 32.9595,
+      "eval_samples_per_second": 294.756,
+      "eval_steps_per_second": 6.705,
+      "step": 103530
     },
     {
       "epoch": 80.0,
-      "grad_norm": 2.9004069801885635e-05,
-      "learning_rate": 2.004137931034483e-06,
-      "loss": 0.0621,
-      "step": 34800
+      "grad_norm": 0.061361562460660934,
+      "learning_rate": 2.0037863421230565e-06,
+      "loss": 0.0255,
+      "step": 118320
     },
     {
       "epoch": 80.0,
-      "eval_loss": 2.9804065227508545,
-      "eval_runtime": 55.7487,
-      "eval_samples_per_second": 311.451,
-      "eval_steps_per_second": 7.085,
-      "step": 34800
+      "eval_loss": 5.728600978851318,
+      "eval_runtime": 31.4319,
+      "eval_samples_per_second": 309.081,
+      "eval_steps_per_second": 7.031,
+      "step": 118320
     },
     {
       "epoch": 90.0,
-      "grad_norm": 0.00013569927250500768,
-      "learning_rate": 1.0045977011494254e-06,
-      "loss": 0.0596,
-      "step": 39150
+      "grad_norm": 0.0459863506257534,
+      "learning_rate": 1.0042596348884382e-06,
+      "loss": 0.0238,
+      "step": 133110
     },
     {
       "epoch": 90.0,
-      "eval_loss": 2.895709276199341,
-      "eval_runtime": 55.8728,
-      "eval_samples_per_second": 310.76,
-      "eval_steps_per_second": 7.07,
-      "step": 39150
+      "eval_loss": 5.677281856536865,
+      "eval_runtime": 30.8915,
+      "eval_samples_per_second": 314.488,
+      "eval_steps_per_second": 7.154,
+      "step": 133110
     },
     {
       "epoch": 100.0,
-      "grad_norm": 0.0004280584107618779,
-      "learning_rate": 4.827586206896552e-09,
-      "loss": 0.0576,
-      "step": 43500
+      "grad_norm": NaN,
+      "learning_rate": 4.732927653820149e-09,
+      "loss": 0.0225,
+      "step": 147900
     },
     {
       "epoch": 100.0,
-      "eval_loss": 2.854401111602783,
-      "eval_runtime": 55.466,
-      "eval_samples_per_second": 313.038,
-      "eval_steps_per_second": 7.121,
-      "step": 43500
+      "eval_loss": 5.574370384216309,
+      "eval_runtime": 30.8147,
+      "eval_samples_per_second": 315.271,
+      "eval_steps_per_second": 7.172,
+      "step": 147900
     },
     {
       "epoch": 100.0,
-      "step": 43500,
-      "total_flos": 6.859464121840128e+17,
-      "train_loss": 0.11642176906541846,
-      "train_runtime": 15360.6557,
-      "train_samples_per_second": 113.036,
-      "train_steps_per_second": 2.832
+      "step": 147900,
+      "total_flos": 2.3359629906138624e+18,
+      "train_loss": 0.060914613879798955,
+      "train_runtime": 49943.5819,
+      "train_samples_per_second": 118.392,
+      "train_steps_per_second": 2.961
     }
   ],
-  "logging_steps": 4350,
-  "max_steps": 43500,
+  "logging_steps": 14790,
+  "max_steps": 147900,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 100,
   "save_steps": 500,
@@ -185,7 +185,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 6.859464121840128e+17,
+  "total_flos": 2.3359629906138624e+18,
   "train_batch_size": 40,
   "trial_name": null,
   "trial_params": null