End of training

README.md

@@ -2,11 +2,23 @@
 library_name: transformers
 tags:
 - generated_from_trainer
+datasets:
+- kanishka/babylm2-rewritten-clean-spacy
 metrics:
 - accuracy
 model-index:
 - name: opt-babylm2-rewritten-clean-spacy-32k-earlystop_seed-42_3e-4
-  results: []
+  results:
+  - task:
+      name: Causal Language Modeling
+      type: text-generation
+    dataset:
+      name: kanishka/babylm2-rewritten-clean-spacy
+      type: kanishka/babylm2-rewritten-clean-spacy
+    metrics:
+    - name: Accuracy
+      type: accuracy
+      value: 0.4242773576186558
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -14,7 +26,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # opt-babylm2-rewritten-clean-spacy-32k-earlystop_seed-42_3e-4
 
-This model was trained from scratch on an unknown dataset.
+This model was trained from scratch on the kanishka/babylm2-rewritten-clean-spacy dataset.
 It achieves the following results on the evaluation set:
 - Loss: 2.9642
 - Accuracy: 0.4243

all_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 19.992235020059532,
+    "eval_accuracy": 0.4242773576186558,
+    "eval_loss": 2.9642465114593506,
+    "eval_runtime": 113.9001,
+    "eval_samples": 52190,
+    "eval_samples_per_second": 458.209,
+    "eval_steps_per_second": 7.164,
+    "perplexity": 19.380095059069767,
+    "total_flos": 1.291633274290176e+18,
+    "train_loss": 3.167741677725272,
+    "train_runtime": 41034.7396,
+    "train_samples": 494517,
+    "train_samples_per_second": 241.024,
+    "train_steps_per_second": 0.941
+}

eval_results.json

@@ -0,0 +1,10 @@
+{
+    "epoch": 19.992235020059532,
+    "eval_accuracy": 0.4242773576186558,
+    "eval_loss": 2.9642465114593506,
+    "eval_runtime": 113.9001,
+    "eval_samples": 52190,
+    "eval_samples_per_second": 458.209,
+    "eval_steps_per_second": 7.164,
+    "perplexity": 19.380095059069767
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 19.992235020059532,
+    "total_flos": 1.291633274290176e+18,
+    "train_loss": 3.167741677725272,
+    "train_runtime": 41034.7396,
+    "train_samples": 494517,
+    "train_samples_per_second": 241.024,
+    "train_steps_per_second": 0.941
+}

trainer_state.json

@@ -0,0 +1,497 @@
+{
+  "best_metric": 2.9642465114593506,
+  "best_model_checkpoint": "models/opt-babylm2-rewritten-clean-spacy-32k-earlystop_seed-42_3e-4/checkpoint-38620",
+  "epoch": 19.992235020059532,
+  "eval_steps": 500,
+  "global_step": 38620,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.5176653293645658,
+      "grad_norm": 0.485008180141449,
+      "learning_rate": 9.375e-06,
+      "loss": 6.8783,
+      "step": 1000
+    },
+    {
+      "epoch": 0.9996117510029766,
+      "eval_accuracy": 0.289060183567583,
+      "eval_loss": 4.46536922454834,
+      "eval_runtime": 113.2643,
+      "eval_samples_per_second": 460.781,
+      "eval_steps_per_second": 7.204,
+      "step": 1931
+    },
+    {
+      "epoch": 1.0353306587291315,
+      "grad_norm": 0.6161150932312012,
+      "learning_rate": 1.875e-05,
+      "loss": 4.6542,
+      "step": 2000
+    },
+    {
+      "epoch": 1.5529959880936974,
+      "grad_norm": 0.6660575270652771,
+      "learning_rate": 2.8125e-05,
+      "loss": 4.2263,
+      "step": 3000
+    },
+    {
+      "epoch": 1.9997411673353178,
+      "eval_accuracy": 0.3337067802786951,
+      "eval_loss": 3.917879343032837,
+      "eval_runtime": 113.7402,
+      "eval_samples_per_second": 458.853,
+      "eval_steps_per_second": 7.174,
+      "step": 3863
+    },
+    {
+      "epoch": 2.070661317458263,
+      "grad_norm": 0.6615888476371765,
+      "learning_rate": 3.75e-05,
+      "loss": 3.9582,
+      "step": 4000
+    },
+    {
+      "epoch": 2.588326646822829,
+      "grad_norm": 0.7149969339370728,
+      "learning_rate": 4.6874999999999994e-05,
+      "loss": 3.7724,
+      "step": 5000
+    },
+    {
+      "epoch": 2.9998705836676587,
+      "eval_accuracy": 0.3562454681048507,
+      "eval_loss": 3.6396915912628174,
+      "eval_runtime": 113.9103,
+      "eval_samples_per_second": 458.168,
+      "eval_steps_per_second": 7.164,
+      "step": 5795
+    },
+    {
+      "epoch": 3.105991976187395,
+      "grad_norm": 0.6370624899864197,
+      "learning_rate": 5.625e-05,
+      "loss": 3.6294,
+      "step": 6000
+    },
+    {
+      "epoch": 3.6236573055519608,
+      "grad_norm": 0.6175299286842346,
+      "learning_rate": 6.5625e-05,
+      "loss": 3.5091,
+      "step": 7000
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.3723592153857136,
+      "eval_loss": 3.456868886947632,
+      "eval_runtime": 113.9063,
+      "eval_samples_per_second": 458.184,
+      "eval_steps_per_second": 7.164,
+      "step": 7727
+    },
+    {
+      "epoch": 4.141322634916526,
+      "grad_norm": 0.5864672064781189,
+      "learning_rate": 7.5e-05,
+      "loss": 3.4133,
+      "step": 8000
+    },
+    {
+      "epoch": 4.658987964281092,
+      "grad_norm": 0.5497373342514038,
+      "learning_rate": 8.437499999999999e-05,
+      "loss": 3.3306,
+      "step": 9000
+    },
+    {
+      "epoch": 4.999611751002977,
+      "eval_accuracy": 0.3838290710037608,
+      "eval_loss": 3.3310000896453857,
+      "eval_runtime": 113.7441,
+      "eval_samples_per_second": 458.837,
+      "eval_steps_per_second": 7.174,
+      "step": 9658
+    },
+    {
+      "epoch": 5.176653293645658,
+      "grad_norm": 0.5599066019058228,
+      "learning_rate": 9.374999999999999e-05,
+      "loss": 3.2608,
+      "step": 10000
+    },
+    {
+      "epoch": 5.694318623010224,
+      "grad_norm": 0.5453025698661804,
+      "learning_rate": 0.00010312499999999999,
+      "loss": 3.2012,
+      "step": 11000
+    },
+    {
+      "epoch": 5.999741167335317,
+      "eval_accuracy": 0.3918066341309469,
+      "eval_loss": 3.2469069957733154,
+      "eval_runtime": 113.6865,
+      "eval_samples_per_second": 459.07,
+      "eval_steps_per_second": 7.178,
+      "step": 11590
+    },
+    {
+      "epoch": 6.21198395237479,
+      "grad_norm": 0.5039867162704468,
+      "learning_rate": 0.0001125,
+      "loss": 3.152,
+      "step": 12000
+    },
+    {
+      "epoch": 6.729649281739356,
+      "grad_norm": 0.4990510642528534,
+      "learning_rate": 0.000121875,
+      "loss": 3.1088,
+      "step": 13000
+    },
+    {
+      "epoch": 6.999870583667659,
+      "eval_accuracy": 0.39817732342985096,
+      "eval_loss": 3.1827573776245117,
+      "eval_runtime": 114.0161,
+      "eval_samples_per_second": 457.742,
+      "eval_steps_per_second": 7.157,
+      "step": 13522
+    },
+    {
+      "epoch": 7.2473146111039215,
+      "grad_norm": 0.46827396750450134,
+      "learning_rate": 0.00013125,
+      "loss": 3.0691,
+      "step": 14000
+    },
+    {
+      "epoch": 7.764979940468487,
+      "grad_norm": 0.4815771281719208,
+      "learning_rate": 0.000140625,
+      "loss": 3.0364,
+      "step": 15000
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.40226096953439355,
+      "eval_loss": 3.1404154300689697,
+      "eval_runtime": 113.6142,
+      "eval_samples_per_second": 459.361,
+      "eval_steps_per_second": 7.182,
+      "step": 15454
+    },
+    {
+      "epoch": 8.282645269833052,
+      "grad_norm": 0.4806288778781891,
+      "learning_rate": 0.000149990625,
+      "loss": 3.0022,
+      "step": 16000
+    },
+    {
+      "epoch": 8.80031059919762,
+      "grad_norm": 0.4368716776371002,
+      "learning_rate": 0.00015936562499999999,
+      "loss": 2.9837,
+      "step": 17000
+    },
+    {
+      "epoch": 8.999611751002977,
+      "eval_accuracy": 0.4056956294684956,
+      "eval_loss": 3.1080353260040283,
+      "eval_runtime": 113.6637,
+      "eval_samples_per_second": 459.162,
+      "eval_steps_per_second": 7.179,
+      "step": 17385
+    },
+    {
+      "epoch": 9.317975928562184,
+      "grad_norm": 0.44254282116889954,
+      "learning_rate": 0.00016873124999999998,
+      "loss": 2.9495,
+      "step": 18000
+    },
+    {
+      "epoch": 9.835641257926751,
+      "grad_norm": 0.4331772029399872,
+      "learning_rate": 0.00017809687499999998,
+      "loss": 2.9377,
+      "step": 19000
+    },
+    {
+      "epoch": 9.999741167335317,
+      "eval_accuracy": 0.40769541156182726,
+      "eval_loss": 3.083951711654663,
+      "eval_runtime": 113.9484,
+      "eval_samples_per_second": 458.015,
+      "eval_steps_per_second": 7.161,
+      "step": 19317
+    },
+    {
+      "epoch": 10.353306587291316,
+      "grad_norm": 0.425889790058136,
+      "learning_rate": 0.00018747187499999998,
+      "loss": 2.9051,
+      "step": 20000
+    },
+    {
+      "epoch": 10.870971916655883,
+      "grad_norm": 0.40670710802078247,
+      "learning_rate": 0.00019684687499999998,
+      "loss": 2.9019,
+      "step": 21000
+    },
+    {
+      "epoch": 10.99987058366766,
+      "eval_accuracy": 0.41011139539164965,
+      "eval_loss": 3.063291311264038,
+      "eval_runtime": 113.7948,
+      "eval_samples_per_second": 458.633,
+      "eval_steps_per_second": 7.171,
+      "step": 21249
+    },
+    {
+      "epoch": 11.388637246020448,
+      "grad_norm": 0.4176114499568939,
+      "learning_rate": 0.00020622187499999998,
+      "loss": 2.8676,
+      "step": 22000
+    },
+    {
+      "epoch": 11.906302575385013,
+      "grad_norm": 0.400343120098114,
+      "learning_rate": 0.00021559687499999997,
+      "loss": 2.8713,
+      "step": 23000
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.4117278120291995,
+      "eval_loss": 3.0504517555236816,
+      "eval_runtime": 113.5309,
+      "eval_samples_per_second": 459.699,
+      "eval_steps_per_second": 7.187,
+      "step": 23181
+    },
+    {
+      "epoch": 12.42396790474958,
+      "grad_norm": 0.3873593509197235,
+      "learning_rate": 0.00022496249999999997,
+      "loss": 2.8353,
+      "step": 24000
+    },
+    {
+      "epoch": 12.941633234114144,
+      "grad_norm": 0.3830123543739319,
+      "learning_rate": 0.00023433749999999997,
+      "loss": 2.8449,
+      "step": 25000
+    },
+    {
+      "epoch": 12.999611751002977,
+      "eval_accuracy": 0.41297589125705847,
+      "eval_loss": 3.037565231323242,
+      "eval_runtime": 113.7043,
+      "eval_samples_per_second": 458.998,
+      "eval_steps_per_second": 7.177,
+      "step": 25112
+    },
+    {
+      "epoch": 13.459298563478711,
+      "grad_norm": 0.390222430229187,
+      "learning_rate": 0.00024369374999999997,
+      "loss": 2.8068,
+      "step": 26000
+    },
+    {
+      "epoch": 13.976963892843276,
+      "grad_norm": 0.3650892376899719,
+      "learning_rate": 0.00025306875,
+      "loss": 2.8231,
+      "step": 27000
+    },
+    {
+      "epoch": 13.999741167335317,
+      "eval_accuracy": 0.41429497800269754,
+      "eval_loss": 3.026965856552124,
+      "eval_runtime": 113.7894,
+      "eval_samples_per_second": 458.654,
+      "eval_steps_per_second": 7.171,
+      "step": 27044
+    },
+    {
+      "epoch": 14.494629222207843,
+      "grad_norm": 0.3750116527080536,
+      "learning_rate": 0.00026244375,
+      "loss": 2.7828,
+      "step": 28000
+    },
+    {
+      "epoch": 14.99987058366766,
+      "eval_accuracy": 0.4150191044035932,
+      "eval_loss": 3.0222034454345703,
+      "eval_runtime": 113.786,
+      "eval_samples_per_second": 458.668,
+      "eval_steps_per_second": 7.171,
+      "step": 28976
+    },
+    {
+      "epoch": 15.012294551572408,
+      "grad_norm": 0.39160436391830444,
+      "learning_rate": 0.00027181875,
+      "loss": 2.804,
+      "step": 29000
+    },
+    {
+      "epoch": 15.529959880936975,
+      "grad_norm": 0.36566755175590515,
+      "learning_rate": 0.00028118437499999993,
+      "loss": 2.7644,
+      "step": 30000
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.4156012157689288,
+      "eval_loss": 3.0159823894500732,
+      "eval_runtime": 113.6022,
+      "eval_samples_per_second": 459.41,
+      "eval_steps_per_second": 7.183,
+      "step": 30908
+    },
+    {
+      "epoch": 16.04762521030154,
+      "grad_norm": 0.37446266412734985,
+      "learning_rate": 0.000290559375,
+      "loss": 2.7817,
+      "step": 31000
+    },
+    {
+      "epoch": 16.565290539666105,
+      "grad_norm": 0.3600151836872101,
+      "learning_rate": 0.000299925,
+      "loss": 2.7508,
+      "step": 32000
+    },
+    {
+      "epoch": 16.999611751002977,
+      "eval_accuracy": 0.4174987320086111,
+      "eval_loss": 3.003650188446045,
+      "eval_runtime": 113.5557,
+      "eval_samples_per_second": 459.598,
+      "eval_steps_per_second": 7.186,
+      "step": 32839
+    },
+    {
+      "epoch": 17.082955869030673,
+      "grad_norm": 0.3546937108039856,
+      "learning_rate": 0.0002550453172205438,
+      "loss": 2.7556,
+      "step": 33000
+    },
+    {
+      "epoch": 17.60062119839524,
+      "grad_norm": 0.3554268777370453,
+      "learning_rate": 0.00020977341389728096,
+      "loss": 2.7036,
+      "step": 34000
+    },
+    {
+      "epoch": 17.99974116733532,
+      "eval_accuracy": 0.4205405588178939,
+      "eval_loss": 2.9801993370056152,
+      "eval_runtime": 113.7922,
+      "eval_samples_per_second": 458.643,
+      "eval_steps_per_second": 7.171,
+      "step": 34771
+    },
+    {
+      "epoch": 18.118286527759803,
+      "grad_norm": 0.36419546604156494,
+      "learning_rate": 0.00016450151057401813,
+      "loss": 2.6855,
+      "step": 35000
+    },
+    {
+      "epoch": 18.63595185712437,
+      "grad_norm": 0.3601652979850769,
+      "learning_rate": 0.00011918429003021148,
+      "loss": 2.6333,
+      "step": 36000
+    },
+    {
+      "epoch": 18.999870583667658,
+      "eval_accuracy": 0.42307188290146486,
+      "eval_loss": 2.9676599502563477,
+      "eval_runtime": 114.5137,
+      "eval_samples_per_second": 455.753,
+      "eval_steps_per_second": 7.126,
+      "step": 36703
+    },
+    {
+      "epoch": 19.153617186488933,
+      "grad_norm": 0.3770598769187927,
+      "learning_rate": 7.386706948640483e-05,
+      "loss": 2.608,
+      "step": 37000
+    },
+    {
+      "epoch": 19.671282515853502,
+      "grad_norm": 0.37394067645072937,
+      "learning_rate": 2.8595166163141992e-05,
+      "loss": 2.557,
+      "step": 38000
+    },
+    {
+      "epoch": 19.992235020059532,
+      "eval_accuracy": 0.4242773576186558,
+      "eval_loss": 2.9642465114593506,
+      "eval_runtime": 114.3789,
+      "eval_samples_per_second": 456.291,
+      "eval_steps_per_second": 7.134,
+      "step": 38620
+    },
+    {
+      "epoch": 19.992235020059532,
+      "step": 38620,
+      "total_flos": 1.291633274290176e+18,
+      "train_loss": 3.167741677725272,
+      "train_runtime": 41034.7396,
+      "train_samples_per_second": 241.024,
+      "train_steps_per_second": 0.941
+    }
+  ],
+  "logging_steps": 1000,
+  "max_steps": 38620,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 20,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 3,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 0
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.291633274290176e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}