added finetuned categorical classification model + more evaluated dataset modifications

data_categories/Computers_and_Electronics.csv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:23ba5a3666e7d67ef57a7fcf301685ae7b66ffc1ed4f33ee6d7b01d11af136ed
-size 27167
+oid sha256:951be0765dfdba915851bf70958f057fa8e57ee652368599652c09134597cdb5
+size 32224

data_categories/Final_Category_Data_With_Labels.csv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8c2ba96d90a437a017a25af64364a58c7e2954ca1519a5fce27d0e55addae8da
-size 1810529
+oid sha256:1cc9c2f33eb98166acba40de7c124801b57c19ececb8d24900980fb9d5cbabe6
+size 1671966

data_categories/Food_and_Drink.csv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:317f8d87736b3523b83ec3fe0c8bcd2d36df06fce8076a4a414125bfc1ca249b
-size 42073
+oid sha256:f25d7848aac215feb6ac57d256005b70d1e39eb0f46bd8cbdc60b94469c3a368
+size 46822

data_categories/Online Communities.csv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:903899c0689fab2c6648a715e5e488429b847054af33c1d26b252eb36352cf29
-size 46352
+oid sha256:65ae0ed633eae668652cee2a7aab4e48a873a1815edb094521257af71360d5d3
+size 55212

finetuned_entity_categorical_classification/checkpoint-11032/added_tokens.json

@@ -0,0 +1,7 @@
+{
+  "[CLS]": 101,
+  "[MASK]": 103,
+  "[PAD]": 0,
+  "[SEP]": 102,
+  "[UNK]": 100
+}

finetuned_entity_categorical_classification/checkpoint-11032/config.json

@@ -0,0 +1,83 @@
+{
+  "_name_or_path": "distilbert-base-uncased",
+  "activation": "gelu",
+  "architectures": [
+    "DistilBertForSequenceClassification"
+  ],
+  "attention_dropout": 0.1,
+  "dim": 768,
+  "dropout": 0.1,
+  "hidden_dim": 3072,
+  "id2label": {
+    "0": "Beauty_and_Fitness",
+    "1": "People_and_Society",
+    "2": "Travel_and_Transportation",
+    "3": "Shopping",
+    "4": "Adult",
+    "5": "Sports",
+    "6": "Science",
+    "7": "Food_and_Drink",
+    "8": "News",
+    "9": "Sensitive Subjects",
+    "10": "Autos_and_Vehicles",
+    "11": "Law_and_Government",
+    "12": "Business_and_Industrial",
+    "13": "Health",
+    "14": "Real Estate",
+    "15": "Books_and_Literature",
+    "16": "Computers_and_Electronics",
+    "17": "Internet_and_Telecom",
+    "18": "Home_and_Garden",
+    "19": "Jobs_and_Education",
+    "20": "Online Communities",
+    "21": "Finance",
+    "22": "Arts_and_Entertainment",
+    "23": "Games",
+    "24": "Hobbies_and_Leisure",
+    "25": "Reference",
+    "26": "Pets_and_Animals"
+  },
+  "initializer_range": 0.02,
+  "label2id": {
+    "Adult": 4,
+    "Arts_and_Entertainment": 22,
+    "Autos_and_Vehicles": 10,
+    "Beauty_and_Fitness": 0,
+    "Books_and_Literature": 15,
+    "Business_and_Industrial": 12,
+    "Computers_and_Electronics": 16,
+    "Finance": 21,
+    "Food_and_Drink": 7,
+    "Games": 23,
+    "Health": 13,
+    "Hobbies_and_Leisure": 24,
+    "Home_and_Garden": 18,
+    "Internet_and_Telecom": 17,
+    "Jobs_and_Education": 19,
+    "Law_and_Government": 11,
+    "News": 8,
+    "Online Communities": 20,
+    "People_and_Society": 1,
+    "Pets_and_Animals": 26,
+    "Real Estate": 14,
+    "Reference": 25,
+    "Science": 6,
+    "Sensitive Subjects": 9,
+    "Shopping": 3,
+    "Sports": 5,
+    "Travel_and_Transportation": 2
+  },
+  "max_position_embeddings": 512,
+  "model_type": "distilbert",
+  "n_heads": 12,
+  "n_layers": 6,
+  "pad_token_id": 0,
+  "problem_type": "single_label_classification",
+  "qa_dropout": 0.1,
+  "seq_classif_dropout": 0.2,
+  "sinusoidal_pos_embds": false,
+  "tie_weights_": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.34.0",
+  "vocab_size": 30522
+}

finetuned_entity_categorical_classification/checkpoint-11032/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f7ab679648567dedd5ee16883a5b327f5addcd457801f36d1cf2fde78ddd114
+size 535881018

finetuned_entity_categorical_classification/checkpoint-11032/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eaedbcf0c17b7424c96b7c789c62d9d7a5b1e7ec2f611accbc326a6346dfc71b
+size 267932842

finetuned_entity_categorical_classification/checkpoint-11032/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d5094212461ae02915e8b4efcecec1cdff5f9e5d485a0996bdaf106d8d66aa6
+size 1064

finetuned_entity_categorical_classification/checkpoint-11032/special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "cls_token": "[CLS]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
+}

finetuned_entity_categorical_classification/checkpoint-11032/tokenizer_config.json

@@ -0,0 +1,56 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "101": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "102": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "103": {
+      "content": "[MASK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [],
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "do_lower_case": true,
+  "mask_token": "[MASK]",
+  "model_max_length": 512,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "DistilBertTokenizer",
+  "unk_token": "[UNK]"
+}

finetuned_entity_categorical_classification/checkpoint-11032/trainer_state.json

@@ -0,0 +1,214 @@
+{
+  "best_metric": 0.21237443387508392,
+  "best_model_checkpoint": "finetuned_entity_categorical_classification/checkpoint-4728",
+  "epoch": 7.0,
+  "eval_steps": 500,
+  "global_step": 11032,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.957698815566836e-05,
+      "loss": 1.5567,
+      "step": 500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.915397631133672e-05,
+      "loss": 0.3944,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.873096446700508e-05,
+      "loss": 0.2773,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.9374900840869427,
+      "eval_loss": 0.2187376469373703,
+      "eval_runtime": 2.2114,
+      "eval_samples_per_second": 2850.256,
+      "eval_steps_per_second": 178.169,
+      "step": 1576
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.830795262267344e-05,
+      "loss": 0.1997,
+      "step": 2000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.7884940778341796e-05,
+      "loss": 0.202,
+      "step": 2500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7461928934010152e-05,
+      "loss": 0.1797,
+      "step": 3000
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.9452641599238458,
+      "eval_loss": 0.22844311594963074,
+      "eval_runtime": 2.2403,
+      "eval_samples_per_second": 2813.437,
+      "eval_steps_per_second": 175.868,
+      "step": 3152
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7038917089678512e-05,
+      "loss": 0.148,
+      "step": 3500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.661590524534687e-05,
+      "loss": 0.1357,
+      "step": 4000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.619289340101523e-05,
+      "loss": 0.1525,
+      "step": 4500
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.9481199428843408,
+      "eval_loss": 0.21237443387508392,
+      "eval_runtime": 2.2281,
+      "eval_samples_per_second": 2828.834,
+      "eval_steps_per_second": 176.83,
+      "step": 4728
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.576988155668359e-05,
+      "loss": 0.1218,
+      "step": 5000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5346869712351946e-05,
+      "loss": 0.1147,
+      "step": 5500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4923857868020306e-05,
+      "loss": 0.1195,
+      "step": 6000
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.94859590671109,
+      "eval_loss": 0.2216227501630783,
+      "eval_runtime": 2.2255,
+      "eval_samples_per_second": 2832.161,
+      "eval_steps_per_second": 177.038,
+      "step": 6304
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.4500846023688663e-05,
+      "loss": 0.114,
+      "step": 6500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.4077834179357023e-05,
+      "loss": 0.104,
+      "step": 7000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.3654822335025382e-05,
+      "loss": 0.0936,
+      "step": 7500
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9462160875773441,
+      "eval_loss": 0.24313929677009583,
+      "eval_runtime": 2.1752,
+      "eval_samples_per_second": 2897.6,
+      "eval_steps_per_second": 181.129,
+      "step": 7880
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.323181049069374e-05,
+      "loss": 0.099,
+      "step": 8000
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.28087986463621e-05,
+      "loss": 0.0929,
+      "step": 8500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2385786802030457e-05,
+      "loss": 0.0872,
+      "step": 9000
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.9465333967951769,
+      "eval_loss": 0.25265371799468994,
+      "eval_runtime": 2.2222,
+      "eval_samples_per_second": 2836.351,
+      "eval_steps_per_second": 177.3,
+      "step": 9456
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.1962774957698817e-05,
+      "loss": 0.0963,
+      "step": 9500
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.1539763113367176e-05,
+      "loss": 0.0733,
+      "step": 10000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.1116751269035532e-05,
+      "loss": 0.0812,
+      "step": 10500
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.0693739424703892e-05,
+      "loss": 0.0929,
+      "step": 11000
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.9490718705378391,
+      "eval_loss": 0.23417465388774872,
+      "eval_runtime": 2.1699,
+      "eval_samples_per_second": 2904.721,
+      "eval_steps_per_second": 181.574,
+      "step": 11032
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 23640,
+  "num_train_epochs": 15,
+  "save_steps": 500,
+  "total_flos": 703983650219928.0,
+  "trial_name": null,
+  "trial_params": null
+}

finetuned_entity_categorical_classification/checkpoint-11032/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:665da3b0732e752c339d7e2fda57582d0f87bae18392c4edd4fe1327453b2e44
+size 4600

finetuned_entity_categorical_classification/checkpoint-12608/added_tokens.json

@@ -0,0 +1,7 @@
+{
+  "[CLS]": 101,
+  "[MASK]": 103,
+  "[PAD]": 0,
+  "[SEP]": 102,
+  "[UNK]": 100
+}

finetuned_entity_categorical_classification/checkpoint-12608/config.json

@@ -0,0 +1,83 @@
+{
+  "_name_or_path": "distilbert-base-uncased",
+  "activation": "gelu",
+  "architectures": [
+    "DistilBertForSequenceClassification"
+  ],
+  "attention_dropout": 0.1,
+  "dim": 768,
+  "dropout": 0.1,
+  "hidden_dim": 3072,
+  "id2label": {
+    "0": "Beauty_and_Fitness",
+    "1": "People_and_Society",
+    "2": "Travel_and_Transportation",
+    "3": "Shopping",
+    "4": "Adult",
+    "5": "Sports",
+    "6": "Science",
+    "7": "Food_and_Drink",
+    "8": "News",
+    "9": "Sensitive Subjects",
+    "10": "Autos_and_Vehicles",
+    "11": "Law_and_Government",
+    "12": "Business_and_Industrial",
+    "13": "Health",
+    "14": "Real Estate",
+    "15": "Books_and_Literature",
+    "16": "Computers_and_Electronics",
+    "17": "Internet_and_Telecom",
+    "18": "Home_and_Garden",
+    "19": "Jobs_and_Education",
+    "20": "Online Communities",
+    "21": "Finance",
+    "22": "Arts_and_Entertainment",
+    "23": "Games",
+    "24": "Hobbies_and_Leisure",
+    "25": "Reference",
+    "26": "Pets_and_Animals"
+  },
+  "initializer_range": 0.02,
+  "label2id": {
+    "Adult": 4,
+    "Arts_and_Entertainment": 22,
+    "Autos_and_Vehicles": 10,
+    "Beauty_and_Fitness": 0,
+    "Books_and_Literature": 15,
+    "Business_and_Industrial": 12,
+    "Computers_and_Electronics": 16,
+    "Finance": 21,
+    "Food_and_Drink": 7,
+    "Games": 23,
+    "Health": 13,
+    "Hobbies_and_Leisure": 24,
+    "Home_and_Garden": 18,
+    "Internet_and_Telecom": 17,
+    "Jobs_and_Education": 19,
+    "Law_and_Government": 11,
+    "News": 8,
+    "Online Communities": 20,
+    "People_and_Society": 1,
+    "Pets_and_Animals": 26,
+    "Real Estate": 14,
+    "Reference": 25,
+    "Science": 6,
+    "Sensitive Subjects": 9,
+    "Shopping": 3,
+    "Sports": 5,
+    "Travel_and_Transportation": 2
+  },
+  "max_position_embeddings": 512,
+  "model_type": "distilbert",
+  "n_heads": 12,
+  "n_layers": 6,
+  "pad_token_id": 0,
+  "problem_type": "single_label_classification",
+  "qa_dropout": 0.1,
+  "seq_classif_dropout": 0.2,
+  "sinusoidal_pos_embds": false,
+  "tie_weights_": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.34.0",
+  "vocab_size": 30522
+}

finetuned_entity_categorical_classification/checkpoint-12608/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5067b387ccee9053d789251b1a5c5acf414f391a88bf14a4a7c607312ff4f8f8
+size 535881018

finetuned_entity_categorical_classification/checkpoint-12608/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6b0aec1e67b9a8171d7b46316b23bbf9fee0086b5973a30240e840a0b0f91d6
+size 267932842

finetuned_entity_categorical_classification/checkpoint-12608/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5110991592dae080d60ce2bceb93f31532f0a386693c0950a2c500d7d21275c
+size 1064

finetuned_entity_categorical_classification/checkpoint-12608/special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "cls_token": "[CLS]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
+}

finetuned_entity_categorical_classification/checkpoint-12608/tokenizer_config.json

@@ -0,0 +1,56 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "101": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "102": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "103": {
+      "content": "[MASK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [],
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "do_lower_case": true,
+  "mask_token": "[MASK]",
+  "model_max_length": 512,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "DistilBertTokenizer",
+  "unk_token": "[UNK]"
+}

finetuned_entity_categorical_classification/checkpoint-12608/trainer_state.json

@@ -0,0 +1,241 @@
+{
+  "best_metric": 0.21237443387508392,
+  "best_model_checkpoint": "finetuned_entity_categorical_classification/checkpoint-4728",
+  "epoch": 8.0,
+  "eval_steps": 500,
+  "global_step": 12608,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.957698815566836e-05,
+      "loss": 1.5567,
+      "step": 500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.915397631133672e-05,
+      "loss": 0.3944,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.873096446700508e-05,
+      "loss": 0.2773,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.9374900840869427,
+      "eval_loss": 0.2187376469373703,
+      "eval_runtime": 2.2114,
+      "eval_samples_per_second": 2850.256,
+      "eval_steps_per_second": 178.169,
+      "step": 1576
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.830795262267344e-05,
+      "loss": 0.1997,
+      "step": 2000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.7884940778341796e-05,
+      "loss": 0.202,
+      "step": 2500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7461928934010152e-05,
+      "loss": 0.1797,
+      "step": 3000
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.9452641599238458,
+      "eval_loss": 0.22844311594963074,
+      "eval_runtime": 2.2403,
+      "eval_samples_per_second": 2813.437,
+      "eval_steps_per_second": 175.868,
+      "step": 3152
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7038917089678512e-05,
+      "loss": 0.148,
+      "step": 3500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.661590524534687e-05,
+      "loss": 0.1357,
+      "step": 4000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.619289340101523e-05,
+      "loss": 0.1525,
+      "step": 4500
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.9481199428843408,
+      "eval_loss": 0.21237443387508392,
+      "eval_runtime": 2.2281,
+      "eval_samples_per_second": 2828.834,
+      "eval_steps_per_second": 176.83,
+      "step": 4728
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.576988155668359e-05,
+      "loss": 0.1218,
+      "step": 5000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5346869712351946e-05,
+      "loss": 0.1147,
+      "step": 5500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4923857868020306e-05,
+      "loss": 0.1195,
+      "step": 6000
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.94859590671109,
+      "eval_loss": 0.2216227501630783,
+      "eval_runtime": 2.2255,
+      "eval_samples_per_second": 2832.161,
+      "eval_steps_per_second": 177.038,
+      "step": 6304
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.4500846023688663e-05,
+      "loss": 0.114,
+      "step": 6500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.4077834179357023e-05,
+      "loss": 0.104,
+      "step": 7000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.3654822335025382e-05,
+      "loss": 0.0936,
+      "step": 7500
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9462160875773441,
+      "eval_loss": 0.24313929677009583,
+      "eval_runtime": 2.1752,
+      "eval_samples_per_second": 2897.6,
+      "eval_steps_per_second": 181.129,
+      "step": 7880
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.323181049069374e-05,
+      "loss": 0.099,
+      "step": 8000
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.28087986463621e-05,
+      "loss": 0.0929,
+      "step": 8500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2385786802030457e-05,
+      "loss": 0.0872,
+      "step": 9000
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.9465333967951769,
+      "eval_loss": 0.25265371799468994,
+      "eval_runtime": 2.2222,
+      "eval_samples_per_second": 2836.351,
+      "eval_steps_per_second": 177.3,
+      "step": 9456
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.1962774957698817e-05,
+      "loss": 0.0963,
+      "step": 9500
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.1539763113367176e-05,
+      "loss": 0.0733,
+      "step": 10000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.1116751269035532e-05,
+      "loss": 0.0812,
+      "step": 10500
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.0693739424703892e-05,
+      "loss": 0.0929,
+      "step": 11000
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.9490718705378391,
+      "eval_loss": 0.23417465388774872,
+      "eval_runtime": 2.1699,
+      "eval_samples_per_second": 2904.721,
+      "eval_steps_per_second": 181.574,
+      "step": 11032
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.0270727580372251e-05,
+      "loss": 0.0629,
+      "step": 11500
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 9.84771573604061e-06,
+      "loss": 0.0802,
+      "step": 12000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.424703891708968e-06,
+      "loss": 0.0757,
+      "step": 12500
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.9490718705378391,
+      "eval_loss": 0.24814845621585846,
+      "eval_runtime": 2.241,
+      "eval_samples_per_second": 2812.616,
+      "eval_steps_per_second": 175.816,
+      "step": 12608
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 23640,
+  "num_train_epochs": 15,
+  "save_steps": 500,
+  "total_flos": 804434563822968.0,
+  "trial_name": null,
+  "trial_params": null
+}

finetuned_entity_categorical_classification/checkpoint-12608/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:665da3b0732e752c339d7e2fda57582d0f87bae18392c4edd4fe1327453b2e44
+size 4600

finetuned_entity_categorical_classification/checkpoint-14184/added_tokens.json

@@ -0,0 +1,7 @@
+{
+  "[CLS]": 101,
+  "[MASK]": 103,
+  "[PAD]": 0,
+  "[SEP]": 102,
+  "[UNK]": 100
+}

finetuned_entity_categorical_classification/checkpoint-14184/config.json

@@ -0,0 +1,83 @@
+{
+  "_name_or_path": "distilbert-base-uncased",
+  "activation": "gelu",
+  "architectures": [
+    "DistilBertForSequenceClassification"
+  ],
+  "attention_dropout": 0.1,
+  "dim": 768,
+  "dropout": 0.1,
+  "hidden_dim": 3072,
+  "id2label": {
+    "0": "Beauty_and_Fitness",
+    "1": "People_and_Society",
+    "2": "Travel_and_Transportation",
+    "3": "Shopping",
+    "4": "Adult",
+    "5": "Sports",
+    "6": "Science",
+    "7": "Food_and_Drink",
+    "8": "News",
+    "9": "Sensitive Subjects",
+    "10": "Autos_and_Vehicles",
+    "11": "Law_and_Government",
+    "12": "Business_and_Industrial",
+    "13": "Health",
+    "14": "Real Estate",
+    "15": "Books_and_Literature",
+    "16": "Computers_and_Electronics",
+    "17": "Internet_and_Telecom",
+    "18": "Home_and_Garden",
+    "19": "Jobs_and_Education",
+    "20": "Online Communities",
+    "21": "Finance",
+    "22": "Arts_and_Entertainment",
+    "23": "Games",
+    "24": "Hobbies_and_Leisure",
+    "25": "Reference",
+    "26": "Pets_and_Animals"
+  },
+  "initializer_range": 0.02,
+  "label2id": {
+    "Adult": 4,
+    "Arts_and_Entertainment": 22,
+    "Autos_and_Vehicles": 10,
+    "Beauty_and_Fitness": 0,
+    "Books_and_Literature": 15,
+    "Business_and_Industrial": 12,
+    "Computers_and_Electronics": 16,
+    "Finance": 21,
+    "Food_and_Drink": 7,
+    "Games": 23,
+    "Health": 13,
+    "Hobbies_and_Leisure": 24,
+    "Home_and_Garden": 18,
+    "Internet_and_Telecom": 17,
+    "Jobs_and_Education": 19,
+    "Law_and_Government": 11,
+    "News": 8,
+    "Online Communities": 20,
+    "People_and_Society": 1,
+    "Pets_and_Animals": 26,
+    "Real Estate": 14,
+    "Reference": 25,
+    "Science": 6,
+    "Sensitive Subjects": 9,
+    "Shopping": 3,
+    "Sports": 5,
+    "Travel_and_Transportation": 2
+  },
+  "max_position_embeddings": 512,
+  "model_type": "distilbert",
+  "n_heads": 12,
+  "n_layers": 6,
+  "pad_token_id": 0,
+  "problem_type": "single_label_classification",
+  "qa_dropout": 0.1,
+  "seq_classif_dropout": 0.2,
+  "sinusoidal_pos_embds": false,
+  "tie_weights_": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.34.0",
+  "vocab_size": 30522
+}

finetuned_entity_categorical_classification/checkpoint-14184/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7aa590b9129b35b41a2d1ce971a668ce5e1163066050e58b78e70e810d954b14
+size 535881018

finetuned_entity_categorical_classification/checkpoint-14184/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc27a7b4430993ed96dc4ca6a093c1d56f91af41b7dae7656f4123bafb7cc351
+size 267932842

finetuned_entity_categorical_classification/checkpoint-14184/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91556a0da86661baaedcd0c391f9995fe50cace2508f811078e409ba7f709686
+size 1064

finetuned_entity_categorical_classification/checkpoint-14184/special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "cls_token": "[CLS]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
+}

finetuned_entity_categorical_classification/checkpoint-14184/tokenizer_config.json

@@ -0,0 +1,56 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "101": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "102": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "103": {
+      "content": "[MASK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [],
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "do_lower_case": true,
+  "mask_token": "[MASK]",
+  "model_max_length": 512,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "DistilBertTokenizer",
+  "unk_token": "[UNK]"
+}

finetuned_entity_categorical_classification/checkpoint-14184/trainer_state.json

@@ -0,0 +1,268 @@
+{
+  "best_metric": 0.21237443387508392,
+  "best_model_checkpoint": "finetuned_entity_categorical_classification/checkpoint-4728",
+  "epoch": 9.0,
+  "eval_steps": 500,
+  "global_step": 14184,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.957698815566836e-05,
+      "loss": 1.5567,
+      "step": 500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.915397631133672e-05,
+      "loss": 0.3944,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.873096446700508e-05,
+      "loss": 0.2773,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.9374900840869427,
+      "eval_loss": 0.2187376469373703,
+      "eval_runtime": 2.2114,
+      "eval_samples_per_second": 2850.256,
+      "eval_steps_per_second": 178.169,
+      "step": 1576
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.830795262267344e-05,
+      "loss": 0.1997,
+      "step": 2000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.7884940778341796e-05,
+      "loss": 0.202,
+      "step": 2500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7461928934010152e-05,
+      "loss": 0.1797,
+      "step": 3000
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.9452641599238458,
+      "eval_loss": 0.22844311594963074,
+      "eval_runtime": 2.2403,
+      "eval_samples_per_second": 2813.437,
+      "eval_steps_per_second": 175.868,
+      "step": 3152
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7038917089678512e-05,
+      "loss": 0.148,
+      "step": 3500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.661590524534687e-05,
+      "loss": 0.1357,
+      "step": 4000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.619289340101523e-05,
+      "loss": 0.1525,
+      "step": 4500
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.9481199428843408,
+      "eval_loss": 0.21237443387508392,
+      "eval_runtime": 2.2281,
+      "eval_samples_per_second": 2828.834,
+      "eval_steps_per_second": 176.83,
+      "step": 4728
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.576988155668359e-05,
+      "loss": 0.1218,
+      "step": 5000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5346869712351946e-05,
+      "loss": 0.1147,
+      "step": 5500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4923857868020306e-05,
+      "loss": 0.1195,
+      "step": 6000
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.94859590671109,
+      "eval_loss": 0.2216227501630783,
+      "eval_runtime": 2.2255,
+      "eval_samples_per_second": 2832.161,
+      "eval_steps_per_second": 177.038,
+      "step": 6304
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.4500846023688663e-05,
+      "loss": 0.114,
+      "step": 6500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.4077834179357023e-05,
+      "loss": 0.104,
+      "step": 7000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.3654822335025382e-05,
+      "loss": 0.0936,
+      "step": 7500
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9462160875773441,
+      "eval_loss": 0.24313929677009583,
+      "eval_runtime": 2.1752,
+      "eval_samples_per_second": 2897.6,
+      "eval_steps_per_second": 181.129,
+      "step": 7880
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.323181049069374e-05,
+      "loss": 0.099,
+      "step": 8000
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.28087986463621e-05,
+      "loss": 0.0929,
+      "step": 8500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2385786802030457e-05,
+      "loss": 0.0872,
+      "step": 9000
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.9465333967951769,
+      "eval_loss": 0.25265371799468994,
+      "eval_runtime": 2.2222,
+      "eval_samples_per_second": 2836.351,
+      "eval_steps_per_second": 177.3,
+      "step": 9456
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.1962774957698817e-05,
+      "loss": 0.0963,
+      "step": 9500
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.1539763113367176e-05,
+      "loss": 0.0733,
+      "step": 10000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.1116751269035532e-05,
+      "loss": 0.0812,
+      "step": 10500
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.0693739424703892e-05,
+      "loss": 0.0929,
+      "step": 11000
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.9490718705378391,
+      "eval_loss": 0.23417465388774872,
+      "eval_runtime": 2.1699,
+      "eval_samples_per_second": 2904.721,
+      "eval_steps_per_second": 181.574,
+      "step": 11032
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.0270727580372251e-05,
+      "loss": 0.0629,
+      "step": 11500
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 9.84771573604061e-06,
+      "loss": 0.0802,
+      "step": 12000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.424703891708968e-06,
+      "loss": 0.0757,
+      "step": 12500
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.9490718705378391,
+      "eval_loss": 0.24814845621585846,
+      "eval_runtime": 2.241,
+      "eval_samples_per_second": 2812.616,
+      "eval_steps_per_second": 175.816,
+      "step": 12608
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 9.001692047377328e-06,
+      "loss": 0.063,
+      "step": 13000
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 8.578680203045686e-06,
+      "loss": 0.0619,
+      "step": 13500
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 8.155668358714045e-06,
+      "loss": 0.0673,
+      "step": 14000
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.9474853244486753,
+      "eval_loss": 0.2526280879974365,
+      "eval_runtime": 2.1778,
+      "eval_samples_per_second": 2894.229,
+      "eval_steps_per_second": 180.918,
+      "step": 14184
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 23640,
+  "num_train_epochs": 15,
+  "save_steps": 500,
+  "total_flos": 904402998290376.0,
+  "trial_name": null,
+  "trial_params": null
+}

finetuned_entity_categorical_classification/checkpoint-14184/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:665da3b0732e752c339d7e2fda57582d0f87bae18392c4edd4fe1327453b2e44
+size 4600

finetuned_entity_categorical_classification/checkpoint-1576/added_tokens.json

@@ -0,0 +1,7 @@
+{
+  "[CLS]": 101,
+  "[MASK]": 103,
+  "[PAD]": 0,
+  "[SEP]": 102,
+  "[UNK]": 100
+}

finetuned_entity_categorical_classification/checkpoint-1576/config.json

@@ -0,0 +1,83 @@
+{
+  "_name_or_path": "distilbert-base-uncased",
+  "activation": "gelu",
+  "architectures": [
+    "DistilBertForSequenceClassification"
+  ],
+  "attention_dropout": 0.1,
+  "dim": 768,
+  "dropout": 0.1,
+  "hidden_dim": 3072,
+  "id2label": {
+    "0": "Beauty_and_Fitness",
+    "1": "People_and_Society",
+    "2": "Travel_and_Transportation",
+    "3": "Shopping",
+    "4": "Adult",
+    "5": "Sports",
+    "6": "Science",
+    "7": "Food_and_Drink",
+    "8": "News",
+    "9": "Sensitive Subjects",
+    "10": "Autos_and_Vehicles",
+    "11": "Law_and_Government",
+    "12": "Business_and_Industrial",
+    "13": "Health",
+    "14": "Real Estate",
+    "15": "Books_and_Literature",
+    "16": "Computers_and_Electronics",
+    "17": "Internet_and_Telecom",
+    "18": "Home_and_Garden",
+    "19": "Jobs_and_Education",
+    "20": "Online Communities",
+    "21": "Finance",
+    "22": "Arts_and_Entertainment",
+    "23": "Games",
+    "24": "Hobbies_and_Leisure",
+    "25": "Reference",
+    "26": "Pets_and_Animals"
+  },
+  "initializer_range": 0.02,
+  "label2id": {
+    "Adult": 4,
+    "Arts_and_Entertainment": 22,
+    "Autos_and_Vehicles": 10,
+    "Beauty_and_Fitness": 0,
+    "Books_and_Literature": 15,
+    "Business_and_Industrial": 12,
+    "Computers_and_Electronics": 16,
+    "Finance": 21,
+    "Food_and_Drink": 7,
+    "Games": 23,
+    "Health": 13,
+    "Hobbies_and_Leisure": 24,
+    "Home_and_Garden": 18,
+    "Internet_and_Telecom": 17,
+    "Jobs_and_Education": 19,
+    "Law_and_Government": 11,
+    "News": 8,
+    "Online Communities": 20,
+    "People_and_Society": 1,
+    "Pets_and_Animals": 26,
+    "Real Estate": 14,
+    "Reference": 25,
+    "Science": 6,
+    "Sensitive Subjects": 9,
+    "Shopping": 3,
+    "Sports": 5,
+    "Travel_and_Transportation": 2
+  },
+  "max_position_embeddings": 512,
+  "model_type": "distilbert",
+  "n_heads": 12,
+  "n_layers": 6,
+  "pad_token_id": 0,
+  "problem_type": "single_label_classification",
+  "qa_dropout": 0.1,
+  "seq_classif_dropout": 0.2,
+  "sinusoidal_pos_embds": false,
+  "tie_weights_": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.34.0",
+  "vocab_size": 30522
+}

finetuned_entity_categorical_classification/checkpoint-1576/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e20e493b3480b24380280eb5e21c5fd12d4881adfb9cb57a5dd4559f3e85680
+size 535881018

finetuned_entity_categorical_classification/checkpoint-1576/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:732fd46b0cb9afaec1d14b6595994279ae8d82e40715c5b60b128db60718e69b
+size 267932842

finetuned_entity_categorical_classification/checkpoint-1576/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:774c81fc50874cca433d334a482dfa90f04ccfdc642ea608612d6233e8d60700
+size 1064

finetuned_entity_categorical_classification/checkpoint-1576/special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "cls_token": "[CLS]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
+}

finetuned_entity_categorical_classification/checkpoint-1576/tokenizer_config.json

@@ -0,0 +1,56 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "101": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "102": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "103": {
+      "content": "[MASK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [],
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "do_lower_case": true,
+  "mask_token": "[MASK]",
+  "model_max_length": 512,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "DistilBertTokenizer",
+  "unk_token": "[UNK]"
+}

finetuned_entity_categorical_classification/checkpoint-1576/trainer_state.json

@@ -0,0 +1,46 @@
+{
+  "best_metric": 0.2187376469373703,
+  "best_model_checkpoint": "finetuned_entity_categorical_classification/checkpoint-1576",
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 1576,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.957698815566836e-05,
+      "loss": 1.5567,
+      "step": 500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.915397631133672e-05,
+      "loss": 0.3944,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.873096446700508e-05,
+      "loss": 0.2773,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.9374900840869427,
+      "eval_loss": 0.2187376469373703,
+      "eval_runtime": 2.2114,
+      "eval_samples_per_second": 2850.256,
+      "eval_steps_per_second": 178.169,
+      "step": 1576
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 23640,
+  "num_train_epochs": 15,
+  "save_steps": 500,
+  "total_flos": 100414675899720.0,
+  "trial_name": null,
+  "trial_params": null
+}