End of training

all_results.json

@@ -0,0 +1,15 @@
+{
+    "epoch": 5.0,
+    "eval_exact_match": 62.46875,
+    "eval_f1": 80.33013168664009,
+    "eval_runtime": 76.7171,
+    "eval_samples": 3359,
+    "eval_samples_per_second": 43.784,
+    "eval_steps_per_second": 1.564,
+    "total_flos": 5.432587638826752e+16,
+    "train_loss": 0.7275064119728663,
+    "train_runtime": 5090.5681,
+    "train_samples": 15599,
+    "train_samples_per_second": 15.321,
+    "train_steps_per_second": 0.548
+}

eval_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 5.0,
+    "eval_exact_match": 62.46875,
+    "eval_f1": 80.33013168664009,
+    "eval_runtime": 76.7171,
+    "eval_samples": 3359,
+    "eval_samples_per_second": 43.784,
+    "eval_steps_per_second": 1.564
+}

runs/May10_18-36-59_ml-machine/events.out.tfevents.1715371545.ml-machine.109069.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b89a6b3d58928db93abb0acd9fb379b823183eee285e0b931e591a259c105afe
+size 412

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 5.0,
+    "total_flos": 5.432587638826752e+16,
+    "train_loss": 0.7275064119728663,
+    "train_runtime": 5090.5681,
+    "train_samples": 15599,
+    "train_samples_per_second": 15.321,
+    "train_steps_per_second": 0.548
+}

trainer_state.json

@@ -0,0 +1,155 @@
+{
+  "best_metric": 80.33013168664009,
+  "best_model_checkpoint": "/root/turkic_qa/ru_uzn_models/ru_uzn_xlm_roberta_large_squad_model/checkpoint-2790",
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 2790,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.0,
+      "step": 558,
+      "train_exact_match": 64.83516483516483,
+      "train_f1": 83.08204401574274,
+      "train_runtime": 24.0955,
+      "train_samples_per_second": 43.826,
+      "train_steps_per_second": 1.577
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 80.57772827148438,
+      "learning_rate": 1e-05,
+      "loss": 1.2698,
+      "step": 558
+    },
+    {
+      "epoch": 1.0,
+      "eval_exact_match": 61.25,
+      "eval_f1": 79.38877529103148,
+      "eval_runtime": 76.3721,
+      "eval_samples_per_second": 43.982,
+      "eval_steps_per_second": 1.571,
+      "step": 558
+    },
+    {
+      "epoch": 2.0,
+      "step": 1116,
+      "train_exact_match": 74.82517482517483,
+      "train_f1": 89.2404305639639,
+      "train_runtime": 24.2914,
+      "train_samples_per_second": 42.855,
+      "train_steps_per_second": 1.564
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 80.02055358886719,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.8886,
+      "step": 1116
+    },
+    {
+      "epoch": 2.0,
+      "eval_exact_match": 62.3125,
+      "eval_f1": 80.2126480616015,
+      "eval_runtime": 77.9968,
+      "eval_samples_per_second": 43.066,
+      "eval_steps_per_second": 1.539,
+      "step": 1116
+    },
+    {
+      "epoch": 3.0,
+      "step": 1674,
+      "train_exact_match": 78.52147852147853,
+      "train_f1": 92.40489073606565,
+      "train_runtime": 25.6503,
+      "train_samples_per_second": 41.013,
+      "train_steps_per_second": 1.481
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 28.389789581298828,
+      "learning_rate": 5e-06,
+      "loss": 0.6286,
+      "step": 1674
+    },
+    {
+      "epoch": 3.0,
+      "eval_exact_match": 61.9375,
+      "eval_f1": 80.06485926873198,
+      "eval_runtime": 83.2186,
+      "eval_samples_per_second": 40.364,
+      "eval_steps_per_second": 1.442,
+      "step": 1674
+    },
+    {
+      "epoch": 4.0,
+      "step": 2232,
+      "train_exact_match": 82.31768231768231,
+      "train_f1": 93.82612819903538,
+      "train_runtime": 23.9529,
+      "train_samples_per_second": 43.669,
+      "train_steps_per_second": 1.586
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 21.804250717163086,
+      "learning_rate": 2.5e-06,
+      "loss": 0.4725,
+      "step": 2232
+    },
+    {
+      "epoch": 4.0,
+      "eval_exact_match": 62.625,
+      "eval_f1": 80.25175708657538,
+      "eval_runtime": 77.0148,
+      "eval_samples_per_second": 43.615,
+      "eval_steps_per_second": 1.558,
+      "step": 2232
+    },
+    {
+      "epoch": 5.0,
+      "step": 2790,
+      "train_exact_match": 85.21478521478521,
+      "train_f1": 94.39980772824222,
+      "train_runtime": 24.3057,
+      "train_samples_per_second": 42.871,
+      "train_steps_per_second": 1.563
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 48.04233932495117,
+      "learning_rate": 0.0,
+      "loss": 0.378,
+      "step": 2790
+    },
+    {
+      "epoch": 5.0,
+      "eval_exact_match": 62.46875,
+      "eval_f1": 80.33013168664009,
+      "eval_runtime": 78.9765,
+      "eval_samples_per_second": 42.532,
+      "eval_steps_per_second": 1.519,
+      "step": 2790
+    },
+    {
+      "epoch": 5.0,
+      "step": 2790,
+      "total_flos": 5.432587638826752e+16,
+      "train_loss": 0.7275064119728663,
+      "train_runtime": 5090.5681,
+      "train_samples_per_second": 15.321,
+      "train_steps_per_second": 0.548
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 2790,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "total_flos": 5.432587638826752e+16,
+  "train_batch_size": 28,
+  "trial_name": null,
+  "trial_params": null
+}