Training in progress, step 56000, checkpoint

checkpoint-56000/config.json

@@ -0,0 +1,28 @@
+{
+  "architectures": [
+    "RobertaForMaskedLM"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 3,
+  "classifier_dropout": null,
+  "eos_token_id": 4,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 256,
+  "initializer_range": 0.02,
+  "intermediate_size": 1024,
+  "layer_norm_eps": 1e-05,
+  "mask_token_id": 6,
+  "max_position_embeddings": 512,
+  "model_type": "roberta",
+  "num_attention_heads": 8,
+  "num_hidden_layers": 8,
+  "pad_token_id": 1,
+  "position_embedding_type": "absolute",
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.45.2",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 8192
+}

checkpoint-56000/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f173016db0eef2953dc7e709c745b7dfb10ef04f6e7d49036bbbccf2f991e16e
+size 42892200

checkpoint-56000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:761b58d0907ed20f7c49cc52964469f64415d13da070144b72133fc812b4d165
+size 85868090

checkpoint-56000/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1cd1bb115edf141f93d4d564bf211752222ff359cf4384d2b9fc05caf744dd59
+size 14308

checkpoint-56000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:070aba5b9a4e1b60a7b5b726681c1fc286f0fe04da24001e57c946bfb79a2eef
+size 1000

checkpoint-56000/special_tokens_map.json

@@ -0,0 +1,6 @@
+{
+  "bos_token": "[PAD]",
+  "eos_token": "[EOS]",
+  "mask_token": "[MASK]",
+  "pad_token": "[CLS]"
+}

checkpoint-56000/tokenizer_config.json

@@ -0,0 +1,67 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "[EOS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "5": {
+      "content": "[BOS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "6": {
+      "content": "[MASK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "[PAD]",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "[EOS]",
+  "mask_token": "[MASK]",
+  "model_max_length": 128,
+  "pad_token": "[CLS]",
+  "tokenizer_class": "PreTrainedTokenizerFast"
+}

checkpoint-56000/trainer_state.json

@@ -0,0 +1,364 @@
+{
+  "best_metric": 3.1443490982055664,
+  "best_model_checkpoint": "/home/p318482/babyLM_controlled/models_trained_recent/mlm_unmasking/en_mlm_new/childes_mlm_unmasking_sent_13/checkpoint-56000",
+  "epoch": 2.073958854137733,
+  "eval_steps": 2000,
+  "global_step": 56000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.07406995907634761,
+      "eval_loss": 5.4637370109558105,
+      "eval_runtime": 24.2102,
+      "eval_samples_per_second": 1237.041,
+      "eval_steps_per_second": 77.323,
+      "step": 2000
+    },
+    {
+      "epoch": 0.14813991815269523,
+      "grad_norm": 4.785717487335205,
+      "learning_rate": 4e-05,
+      "loss": 6.1476,
+      "step": 4000
+    },
+    {
+      "epoch": 0.14813991815269523,
+      "eval_loss": 4.718623161315918,
+      "eval_runtime": 24.2123,
+      "eval_samples_per_second": 1236.934,
+      "eval_steps_per_second": 77.316,
+      "step": 4000
+    },
+    {
+      "epoch": 0.22220987722904284,
+      "eval_loss": 4.352034091949463,
+      "eval_runtime": 24.1788,
+      "eval_samples_per_second": 1238.646,
+      "eval_steps_per_second": 77.423,
+      "step": 6000
+    },
+    {
+      "epoch": 0.29627983630539045,
+      "grad_norm": 7.057693004608154,
+      "learning_rate": 8e-05,
+      "loss": 4.4471,
+      "step": 8000
+    },
+    {
+      "epoch": 0.29627983630539045,
+      "eval_loss": 3.99312686920166,
+      "eval_runtime": 24.1035,
+      "eval_samples_per_second": 1242.518,
+      "eval_steps_per_second": 77.665,
+      "step": 8000
+    },
+    {
+      "epoch": 0.37034979538173807,
+      "eval_loss": 3.763000965118408,
+      "eval_runtime": 24.2844,
+      "eval_samples_per_second": 1233.262,
+      "eval_steps_per_second": 77.087,
+      "step": 10000
+    },
+    {
+      "epoch": 0.4444197544580857,
+      "grad_norm": 8.325577735900879,
+      "learning_rate": 0.00012,
+      "loss": 3.8939,
+      "step": 12000
+    },
+    {
+      "epoch": 0.4444197544580857,
+      "eval_loss": 3.6284842491149902,
+      "eval_runtime": 24.245,
+      "eval_samples_per_second": 1235.265,
+      "eval_steps_per_second": 77.212,
+      "step": 12000
+    },
+    {
+      "epoch": 0.5184897135344333,
+      "eval_loss": 3.4887945652008057,
+      "eval_runtime": 24.2373,
+      "eval_samples_per_second": 1235.655,
+      "eval_steps_per_second": 77.236,
+      "step": 14000
+    },
+    {
+      "epoch": 0.5925596726107809,
+      "grad_norm": 10.0059232711792,
+      "learning_rate": 0.00016,
+      "loss": 3.6013,
+      "step": 16000
+    },
+    {
+      "epoch": 0.5925596726107809,
+      "eval_loss": 3.443763256072998,
+      "eval_runtime": 24.5656,
+      "eval_samples_per_second": 1219.141,
+      "eval_steps_per_second": 76.204,
+      "step": 16000
+    },
+    {
+      "epoch": 0.6666296316871285,
+      "eval_loss": 3.3501055240631104,
+      "eval_runtime": 24.1675,
+      "eval_samples_per_second": 1239.227,
+      "eval_steps_per_second": 77.459,
+      "step": 18000
+    },
+    {
+      "epoch": 0.7406995907634761,
+      "grad_norm": 5.149896621704102,
+      "learning_rate": 0.0002,
+      "loss": 3.4806,
+      "step": 20000
+    },
+    {
+      "epoch": 0.7406995907634761,
+      "eval_loss": 3.362109899520874,
+      "eval_runtime": 24.3668,
+      "eval_samples_per_second": 1229.089,
+      "eval_steps_per_second": 76.826,
+      "step": 20000
+    },
+    {
+      "epoch": 0.8147695498398237,
+      "eval_loss": 3.317254066467285,
+      "eval_runtime": 24.2052,
+      "eval_samples_per_second": 1237.297,
+      "eval_steps_per_second": 77.339,
+      "step": 22000
+    },
+    {
+      "epoch": 0.8888395089161714,
+      "grad_norm": 6.9324188232421875,
+      "learning_rate": 0.00024,
+      "loss": 3.4164,
+      "step": 24000
+    },
+    {
+      "epoch": 0.8888395089161714,
+      "eval_loss": 3.2997493743896484,
+      "eval_runtime": 24.1632,
+      "eval_samples_per_second": 1239.447,
+      "eval_steps_per_second": 77.473,
+      "step": 24000
+    },
+    {
+      "epoch": 0.962909467992519,
+      "eval_loss": 3.279545545578003,
+      "eval_runtime": 24.2189,
+      "eval_samples_per_second": 1236.597,
+      "eval_steps_per_second": 77.295,
+      "step": 26000
+    },
+    {
+      "epoch": 1.0369794270688666,
+      "grad_norm": 6.834834098815918,
+      "learning_rate": 0.00028000000000000003,
+      "loss": 3.3837,
+      "step": 28000
+    },
+    {
+      "epoch": 1.0369794270688666,
+      "eval_loss": 3.295776844024658,
+      "eval_runtime": 24.2221,
+      "eval_samples_per_second": 1236.433,
+      "eval_steps_per_second": 77.285,
+      "step": 28000
+    },
+    {
+      "epoch": 1.1110493861452142,
+      "eval_loss": 3.2695140838623047,
+      "eval_runtime": 24.1872,
+      "eval_samples_per_second": 1238.217,
+      "eval_steps_per_second": 77.396,
+      "step": 30000
+    },
+    {
+      "epoch": 1.1851193452215618,
+      "grad_norm": 5.360170841217041,
+      "learning_rate": 0.00032,
+      "loss": 3.3341,
+      "step": 32000
+    },
+    {
+      "epoch": 1.1851193452215618,
+      "eval_loss": 3.2437243461608887,
+      "eval_runtime": 24.189,
+      "eval_samples_per_second": 1238.122,
+      "eval_steps_per_second": 77.39,
+      "step": 32000
+    },
+    {
+      "epoch": 1.2591893042979094,
+      "eval_loss": 3.2345123291015625,
+      "eval_runtime": 24.206,
+      "eval_samples_per_second": 1237.255,
+      "eval_steps_per_second": 77.336,
+      "step": 34000
+    },
+    {
+      "epoch": 1.333259263374257,
+      "grad_norm": 5.8984575271606445,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 3.332,
+      "step": 36000
+    },
+    {
+      "epoch": 1.333259263374257,
+      "eval_loss": 3.215236186981201,
+      "eval_runtime": 24.2072,
+      "eval_samples_per_second": 1237.196,
+      "eval_steps_per_second": 77.332,
+      "step": 36000
+    },
+    {
+      "epoch": 1.4073292224506047,
+      "eval_loss": 3.2174911499023438,
+      "eval_runtime": 24.2689,
+      "eval_samples_per_second": 1234.051,
+      "eval_steps_per_second": 77.136,
+      "step": 38000
+    },
+    {
+      "epoch": 1.4813991815269523,
+      "grad_norm": 7.638304710388184,
+      "learning_rate": 0.0004,
+      "loss": 3.3094,
+      "step": 40000
+    },
+    {
+      "epoch": 1.4813991815269523,
+      "eval_loss": 3.2130022048950195,
+      "eval_runtime": 28.0919,
+      "eval_samples_per_second": 1066.108,
+      "eval_steps_per_second": 66.638,
+      "step": 40000
+    },
+    {
+      "epoch": 1.5554691406032997,
+      "eval_loss": 3.2210235595703125,
+      "eval_runtime": 27.2599,
+      "eval_samples_per_second": 1098.646,
+      "eval_steps_per_second": 68.672,
+      "step": 42000
+    },
+    {
+      "epoch": 1.6295390996796475,
+      "grad_norm": 4.747150421142578,
+      "learning_rate": 0.00044,
+      "loss": 3.3162,
+      "step": 44000
+    },
+    {
+      "epoch": 1.6295390996796475,
+      "eval_loss": NaN,
+      "eval_runtime": 28.3768,
+      "eval_samples_per_second": 1055.403,
+      "eval_steps_per_second": 65.969,
+      "step": 44000
+    },
+    {
+      "epoch": 1.7036090587559949,
+      "eval_loss": 3.1528797149658203,
+      "eval_runtime": 27.2801,
+      "eval_samples_per_second": 1097.834,
+      "eval_steps_per_second": 68.621,
+      "step": 46000
+    },
+    {
+      "epoch": 1.7776790178323427,
+      "grad_norm": 5.181939125061035,
+      "learning_rate": 0.00048,
+      "loss": 3.3268,
+      "step": 48000
+    },
+    {
+      "epoch": 1.7776790178323427,
+      "eval_loss": 3.1836674213409424,
+      "eval_runtime": 27.6868,
+      "eval_samples_per_second": 1081.708,
+      "eval_steps_per_second": 67.614,
+      "step": 48000
+    },
+    {
+      "epoch": 1.8517489769086901,
+      "eval_loss": 3.1855623722076416,
+      "eval_runtime": 31.1992,
+      "eval_samples_per_second": 959.929,
+      "eval_steps_per_second": 60.002,
+      "step": 50000
+    },
+    {
+      "epoch": 1.925818935985038,
+      "grad_norm": 4.922842025756836,
+      "learning_rate": 0.0005200000000000001,
+      "loss": 3.3103,
+      "step": 52000
+    },
+    {
+      "epoch": 1.925818935985038,
+      "eval_loss": 3.1568427085876465,
+      "eval_runtime": 30.1831,
+      "eval_samples_per_second": 992.244,
+      "eval_steps_per_second": 62.021,
+      "step": 52000
+    },
+    {
+      "epoch": 1.9998888950613853,
+      "eval_loss": 3.190311908721924,
+      "eval_runtime": 30.5184,
+      "eval_samples_per_second": 981.343,
+      "eval_steps_per_second": 61.34,
+      "step": 54000
+    },
+    {
+      "epoch": 2.073958854137733,
+      "grad_norm": 4.992619037628174,
+      "learning_rate": 0.0005600000000000001,
+      "loss": 3.2876,
+      "step": 56000
+    },
+    {
+      "epoch": 2.073958854137733,
+      "eval_loss": 3.1443490982055664,
+      "eval_runtime": 30.6733,
+      "eval_samples_per_second": 976.387,
+      "eval_steps_per_second": 61.03,
+      "step": 56000
+    }
+  ],
+  "logging_steps": 4000,
+  "max_steps": 400000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 15,
+  "save_steps": 4000,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 3,
+        "early_stopping_threshold": 0.001
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 0
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.168463941337088e+16,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-56000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9dd67d3cacfd6861c6144cee0155c16fcc5b69dd58f8d3c81877d5095c915a2b
+size 5432