Training in progress, step 52000, checkpoint

checkpoint-52000/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-52000/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ce22f132f73cfe4cedc768dd734d0633e265ffd63f0df03ce7feb2eed77232d
+size 42892200

checkpoint-52000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0bb8146004fb7ea7ef55b8c77a2df91429a65308810252ec84c1014d2b8c3f0
+size 85868090

checkpoint-52000/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1144ab3a241f61df12f8b4af50abb0b44635992f383991f3543c42bcf315db2d
+size 14244

checkpoint-52000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7dd71325baf910ace3e1a6acf075a6e070eae334610c6e534376719bec843a3c
+size 1000

checkpoint-52000/special_tokens_map.json

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

checkpoint-52000/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-52000/trainer_state.json

@@ -0,0 +1,341 @@
+{
+  "best_metric": 2.153470039367676,
+  "best_model_checkpoint": "/home/p318482/babyLM_controlled/models_trained_recent/mlm_unmasking/fr_mlm_new/childes_mlm_unmasking_context_42/checkpoint-52000",
+  "epoch": 91.3884007029877,
+  "eval_steps": 2000,
+  "global_step": 52000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 3.51493848857645,
+      "eval_loss": 5.561293601989746,
+      "eval_runtime": 1.7076,
+      "eval_samples_per_second": 1305.334,
+      "eval_steps_per_second": 81.986,
+      "step": 2000
+    },
+    {
+      "epoch": 7.0298769771529,
+      "grad_norm": 1.3643518686294556,
+      "learning_rate": 4e-05,
+      "loss": 6.2537,
+      "step": 4000
+    },
+    {
+      "epoch": 7.0298769771529,
+      "eval_loss": 5.49988317489624,
+      "eval_runtime": 1.7023,
+      "eval_samples_per_second": 1309.415,
+      "eval_steps_per_second": 82.242,
+      "step": 4000
+    },
+    {
+      "epoch": 10.54481546572935,
+      "eval_loss": 5.408681392669678,
+      "eval_runtime": 1.7195,
+      "eval_samples_per_second": 1296.321,
+      "eval_steps_per_second": 81.42,
+      "step": 6000
+    },
+    {
+      "epoch": 14.0597539543058,
+      "grad_norm": 3.336256265640259,
+      "learning_rate": 8e-05,
+      "loss": 5.329,
+      "step": 8000
+    },
+    {
+      "epoch": 14.0597539543058,
+      "eval_loss": 4.505647659301758,
+      "eval_runtime": 1.6952,
+      "eval_samples_per_second": 1314.863,
+      "eval_steps_per_second": 82.584,
+      "step": 8000
+    },
+    {
+      "epoch": 17.57469244288225,
+      "eval_loss": 3.6001980304718018,
+      "eval_runtime": 1.6952,
+      "eval_samples_per_second": 1314.862,
+      "eval_steps_per_second": 82.584,
+      "step": 10000
+    },
+    {
+      "epoch": 21.0896309314587,
+      "grad_norm": 2.967787504196167,
+      "learning_rate": 0.00012,
+      "loss": 3.7571,
+      "step": 12000
+    },
+    {
+      "epoch": 21.0896309314587,
+      "eval_loss": 3.1521711349487305,
+      "eval_runtime": 1.6965,
+      "eval_samples_per_second": 1313.906,
+      "eval_steps_per_second": 82.524,
+      "step": 12000
+    },
+    {
+      "epoch": 24.604569420035148,
+      "eval_loss": 2.8882946968078613,
+      "eval_runtime": 1.7031,
+      "eval_samples_per_second": 1308.779,
+      "eval_steps_per_second": 82.202,
+      "step": 14000
+    },
+    {
+      "epoch": 28.1195079086116,
+      "grad_norm": 2.9542622566223145,
+      "learning_rate": 0.00016,
+      "loss": 2.9513,
+      "step": 16000
+    },
+    {
+      "epoch": 28.1195079086116,
+      "eval_loss": 2.7470672130584717,
+      "eval_runtime": 1.7009,
+      "eval_samples_per_second": 1310.516,
+      "eval_steps_per_second": 82.311,
+      "step": 16000
+    },
+    {
+      "epoch": 31.63444639718805,
+      "eval_loss": 2.6601271629333496,
+      "eval_runtime": 1.7005,
+      "eval_samples_per_second": 1310.816,
+      "eval_steps_per_second": 82.33,
+      "step": 18000
+    },
+    {
+      "epoch": 35.1493848857645,
+      "grad_norm": 2.4665582180023193,
+      "learning_rate": 0.0002,
+      "loss": 2.6047,
+      "step": 20000
+    },
+    {
+      "epoch": 35.1493848857645,
+      "eval_loss": 2.523987293243408,
+      "eval_runtime": 1.707,
+      "eval_samples_per_second": 1305.835,
+      "eval_steps_per_second": 82.017,
+      "step": 20000
+    },
+    {
+      "epoch": 38.66432337434095,
+      "eval_loss": 2.4842116832733154,
+      "eval_runtime": 1.6956,
+      "eval_samples_per_second": 1314.554,
+      "eval_steps_per_second": 82.565,
+      "step": 22000
+    },
+    {
+      "epoch": 42.1792618629174,
+      "grad_norm": 2.4973504543304443,
+      "learning_rate": 0.00024,
+      "loss": 2.3993,
+      "step": 24000
+    },
+    {
+      "epoch": 42.1792618629174,
+      "eval_loss": 2.3980119228363037,
+      "eval_runtime": 1.6977,
+      "eval_samples_per_second": 1312.977,
+      "eval_steps_per_second": 82.466,
+      "step": 24000
+    },
+    {
+      "epoch": 45.69420035149385,
+      "eval_loss": 2.346266269683838,
+      "eval_runtime": 1.698,
+      "eval_samples_per_second": 1312.731,
+      "eval_steps_per_second": 82.451,
+      "step": 26000
+    },
+    {
+      "epoch": 49.209138840070295,
+      "grad_norm": 2.5078186988830566,
+      "learning_rate": 0.00028000000000000003,
+      "loss": 2.2551,
+      "step": 28000
+    },
+    {
+      "epoch": 49.209138840070295,
+      "eval_loss": 2.325450897216797,
+      "eval_runtime": 1.708,
+      "eval_samples_per_second": 1305.065,
+      "eval_steps_per_second": 81.969,
+      "step": 28000
+    },
+    {
+      "epoch": 52.72407732864675,
+      "eval_loss": 2.3088743686676025,
+      "eval_runtime": 1.6993,
+      "eval_samples_per_second": 1311.726,
+      "eval_steps_per_second": 82.387,
+      "step": 30000
+    },
+    {
+      "epoch": 56.2390158172232,
+      "grad_norm": 2.2375969886779785,
+      "learning_rate": 0.00032,
+      "loss": 2.1506,
+      "step": 32000
+    },
+    {
+      "epoch": 56.2390158172232,
+      "eval_loss": 2.2676920890808105,
+      "eval_runtime": 1.7068,
+      "eval_samples_per_second": 1305.937,
+      "eval_steps_per_second": 82.024,
+      "step": 32000
+    },
+    {
+      "epoch": 59.753954305799645,
+      "eval_loss": 2.2623283863067627,
+      "eval_runtime": 1.6975,
+      "eval_samples_per_second": 1313.122,
+      "eval_steps_per_second": 82.475,
+      "step": 34000
+    },
+    {
+      "epoch": 63.2688927943761,
+      "grad_norm": 2.131579875946045,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 2.0777,
+      "step": 36000
+    },
+    {
+      "epoch": 63.2688927943761,
+      "eval_loss": 2.2328691482543945,
+      "eval_runtime": 1.705,
+      "eval_samples_per_second": 1307.34,
+      "eval_steps_per_second": 82.112,
+      "step": 36000
+    },
+    {
+      "epoch": 66.78383128295255,
+      "eval_loss": 2.205517530441284,
+      "eval_runtime": 1.714,
+      "eval_samples_per_second": 1300.481,
+      "eval_steps_per_second": 81.681,
+      "step": 38000
+    },
+    {
+      "epoch": 70.298769771529,
+      "grad_norm": 2.3363709449768066,
+      "learning_rate": 0.0004,
+      "loss": 2.0179,
+      "step": 40000
+    },
+    {
+      "epoch": 70.298769771529,
+      "eval_loss": 2.235269069671631,
+      "eval_runtime": 1.6984,
+      "eval_samples_per_second": 1312.396,
+      "eval_steps_per_second": 82.43,
+      "step": 40000
+    },
+    {
+      "epoch": 73.81370826010544,
+      "eval_loss": 2.191019058227539,
+      "eval_runtime": 1.6977,
+      "eval_samples_per_second": 1312.964,
+      "eval_steps_per_second": 82.465,
+      "step": 42000
+    },
+    {
+      "epoch": 77.3286467486819,
+      "grad_norm": 2.0552377700805664,
+      "learning_rate": 0.00044,
+      "loss": 1.9801,
+      "step": 44000
+    },
+    {
+      "epoch": 77.3286467486819,
+      "eval_loss": 2.2010703086853027,
+      "eval_runtime": 1.7107,
+      "eval_samples_per_second": 1302.947,
+      "eval_steps_per_second": 81.836,
+      "step": 44000
+    },
+    {
+      "epoch": 80.84358523725835,
+      "eval_loss": 2.1846556663513184,
+      "eval_runtime": 1.6975,
+      "eval_samples_per_second": 1313.128,
+      "eval_steps_per_second": 82.476,
+      "step": 46000
+    },
+    {
+      "epoch": 84.3585237258348,
+      "grad_norm": 2.805938720703125,
+      "learning_rate": 0.00048,
+      "loss": 1.9489,
+      "step": 48000
+    },
+    {
+      "epoch": 84.3585237258348,
+      "eval_loss": 2.173431873321533,
+      "eval_runtime": 1.7002,
+      "eval_samples_per_second": 1310.995,
+      "eval_steps_per_second": 82.342,
+      "step": 48000
+    },
+    {
+      "epoch": 87.87346221441125,
+      "eval_loss": 2.1883482933044434,
+      "eval_runtime": 1.7077,
+      "eval_samples_per_second": 1305.249,
+      "eval_steps_per_second": 81.981,
+      "step": 50000
+    },
+    {
+      "epoch": 91.3884007029877,
+      "grad_norm": 1.8798856735229492,
+      "learning_rate": 0.0005200000000000001,
+      "loss": 1.9218,
+      "step": 52000
+    },
+    {
+      "epoch": 91.3884007029877,
+      "eval_loss": 2.153470039367676,
+      "eval_runtime": 1.7038,
+      "eval_samples_per_second": 1308.221,
+      "eval_steps_per_second": 82.167,
+      "step": 52000
+    }
+  ],
+  "logging_steps": 4000,
+  "max_steps": 400000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 703,
+  "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": 1.084477889544192e+16,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-52000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:864f26ae3ed21d609fcbef4afeaac5049d9afc52c206fb700dc34dd3284525ec
+size 5496