Upload batch 24/24: vsfc/mBERT/mha_attention/seed_44 ... vsfc/mBERT/multi_branch_average/seed_44

mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/mha_attention/seed_44/all_results.json

@@ -0,0 +1,15 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.9336702463676564,
+    "eval_loss": 0.2265038937330246,
+    "eval_runtime": 5.4076,
+    "eval_samples": 1583,
+    "eval_samples_per_second": 292.736,
+    "eval_steps_per_second": 3.144,
+    "total_flos": 1029397778831520.0,
+    "train_loss": 0.21097227918553052,
+    "train_runtime": 97.3427,
+    "train_samples": 11426,
+    "train_samples_per_second": 352.137,
+    "train_steps_per_second": 7.366
+}

mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/mha_attention/seed_44/eval_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.9336702463676564,
+    "eval_loss": 0.2265038937330246,
+    "eval_runtime": 5.4076,
+    "eval_samples": 1583,
+    "eval_samples_per_second": 292.736,
+    "eval_steps_per_second": 3.144
+}

mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/mha_attention/seed_44/train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "total_flos": 1029397778831520.0,
+    "train_loss": 0.21097227918553052,
+    "train_runtime": 97.3427,
+    "train_samples": 11426,
+    "train_samples_per_second": 352.137,
+    "train_steps_per_second": 7.366
+}

mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/mha_attention/seed_44/trainer_state.json

@@ -0,0 +1,265 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 717,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.10460251046025104,
+      "grad_norm": 6.291289806365967,
+      "learning_rate": 2.8953974895397488e-05,
+      "loss": 0.5299,
+      "step": 25
+    },
+    {
+      "epoch": 0.20920502092050208,
+      "grad_norm": 3.1578359603881836,
+      "learning_rate": 2.7907949790794982e-05,
+      "loss": 0.3754,
+      "step": 50
+    },
+    {
+      "epoch": 0.3138075313807531,
+      "grad_norm": 3.3247146606445312,
+      "learning_rate": 2.686192468619247e-05,
+      "loss": 0.3676,
+      "step": 75
+    },
+    {
+      "epoch": 0.41841004184100417,
+      "grad_norm": 3.995304584503174,
+      "learning_rate": 2.5815899581589957e-05,
+      "loss": 0.3007,
+      "step": 100
+    },
+    {
+      "epoch": 0.5230125523012552,
+      "grad_norm": 5.295639991760254,
+      "learning_rate": 2.4769874476987448e-05,
+      "loss": 0.2934,
+      "step": 125
+    },
+    {
+      "epoch": 0.6276150627615062,
+      "grad_norm": 4.045957565307617,
+      "learning_rate": 2.372384937238494e-05,
+      "loss": 0.2569,
+      "step": 150
+    },
+    {
+      "epoch": 0.7322175732217573,
+      "grad_norm": 5.039194107055664,
+      "learning_rate": 2.2677824267782426e-05,
+      "loss": 0.2958,
+      "step": 175
+    },
+    {
+      "epoch": 0.8368200836820083,
+      "grad_norm": 3.61696195602417,
+      "learning_rate": 2.1631799163179917e-05,
+      "loss": 0.2204,
+      "step": 200
+    },
+    {
+      "epoch": 0.9414225941422594,
+      "grad_norm": 5.950832843780518,
+      "learning_rate": 2.0585774058577408e-05,
+      "loss": 0.2223,
+      "step": 225
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.9317751105495894,
+      "eval_loss": 0.22243711352348328,
+      "eval_runtime": 6.5128,
+      "eval_samples_per_second": 243.06,
+      "eval_steps_per_second": 2.61,
+      "step": 239
+    },
+    {
+      "epoch": 1.0460251046025104,
+      "grad_norm": 3.6595256328582764,
+      "learning_rate": 1.9539748953974895e-05,
+      "loss": 0.2059,
+      "step": 250
+    },
+    {
+      "epoch": 1.1506276150627615,
+      "grad_norm": 6.140735149383545,
+      "learning_rate": 1.8493723849372386e-05,
+      "loss": 0.2136,
+      "step": 275
+    },
+    {
+      "epoch": 1.2552301255230125,
+      "grad_norm": 3.1940929889678955,
+      "learning_rate": 1.7447698744769877e-05,
+      "loss": 0.1764,
+      "step": 300
+    },
+    {
+      "epoch": 1.3598326359832635,
+      "grad_norm": 2.013075351715088,
+      "learning_rate": 1.6401673640167365e-05,
+      "loss": 0.1971,
+      "step": 325
+    },
+    {
+      "epoch": 1.4644351464435146,
+      "grad_norm": 3.6433889865875244,
+      "learning_rate": 1.5355648535564852e-05,
+      "loss": 0.2052,
+      "step": 350
+    },
+    {
+      "epoch": 1.5690376569037658,
+      "grad_norm": 8.410216331481934,
+      "learning_rate": 1.4309623430962344e-05,
+      "loss": 0.189,
+      "step": 375
+    },
+    {
+      "epoch": 1.6736401673640167,
+      "grad_norm": 6.743198394775391,
+      "learning_rate": 1.3263598326359832e-05,
+      "loss": 0.1916,
+      "step": 400
+    },
+    {
+      "epoch": 1.778242677824268,
+      "grad_norm": 4.715135097503662,
+      "learning_rate": 1.2217573221757323e-05,
+      "loss": 0.1952,
+      "step": 425
+    },
+    {
+      "epoch": 1.8828451882845187,
+      "grad_norm": 6.165963172912598,
+      "learning_rate": 1.1171548117154812e-05,
+      "loss": 0.1896,
+      "step": 450
+    },
+    {
+      "epoch": 1.98744769874477,
+      "grad_norm": 4.81492280960083,
+      "learning_rate": 1.0125523012552301e-05,
+      "loss": 0.157,
+      "step": 475
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.9330385344283006,
+      "eval_loss": 0.22808004915714264,
+      "eval_runtime": 4.5791,
+      "eval_samples_per_second": 345.702,
+      "eval_steps_per_second": 3.713,
+      "step": 478
+    },
+    {
+      "epoch": 2.092050209205021,
+      "grad_norm": 8.199837684631348,
+      "learning_rate": 9.079497907949792e-06,
+      "loss": 0.1407,
+      "step": 500
+    },
+    {
+      "epoch": 2.196652719665272,
+      "grad_norm": 1.9374539852142334,
+      "learning_rate": 8.03347280334728e-06,
+      "loss": 0.1212,
+      "step": 525
+    },
+    {
+      "epoch": 2.301255230125523,
+      "grad_norm": 2.3150672912597656,
+      "learning_rate": 6.98744769874477e-06,
+      "loss": 0.143,
+      "step": 550
+    },
+    {
+      "epoch": 2.405857740585774,
+      "grad_norm": 3.678471326828003,
+      "learning_rate": 5.941422594142259e-06,
+      "loss": 0.13,
+      "step": 575
+    },
+    {
+      "epoch": 2.510460251046025,
+      "grad_norm": 2.900938034057617,
+      "learning_rate": 4.895397489539749e-06,
+      "loss": 0.1314,
+      "step": 600
+    },
+    {
+      "epoch": 2.6150627615062763,
+      "grad_norm": 6.6299357414245605,
+      "learning_rate": 3.849372384937238e-06,
+      "loss": 0.1179,
+      "step": 625
+    },
+    {
+      "epoch": 2.719665271966527,
+      "grad_norm": 3.290058135986328,
+      "learning_rate": 2.803347280334728e-06,
+      "loss": 0.1465,
+      "step": 650
+    },
+    {
+      "epoch": 2.8242677824267783,
+      "grad_norm": 2.124258041381836,
+      "learning_rate": 1.7573221757322176e-06,
+      "loss": 0.1127,
+      "step": 675
+    },
+    {
+      "epoch": 2.928870292887029,
+      "grad_norm": 4.11226224899292,
+      "learning_rate": 7.112970711297071e-07,
+      "loss": 0.1313,
+      "step": 700
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.9336702463676564,
+      "eval_loss": 0.2265038937330246,
+      "eval_runtime": 5.3242,
+      "eval_samples_per_second": 297.324,
+      "eval_steps_per_second": 3.193,
+      "step": 717
+    },
+    {
+      "epoch": 3.0,
+      "step": 717,
+      "total_flos": 1029397778831520.0,
+      "train_loss": 0.21097227918553052,
+      "train_runtime": 97.3427,
+      "train_samples_per_second": 352.137,
+      "train_steps_per_second": 7.366
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 717,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1029397778831520.0,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}