Initial Commit

README.md

@@ -0,0 +1,84 @@
+---
+license: mit
+tags:
+- generated_from_trainer
+datasets:
+- pritamdeka/cord-19-fulltext
+metrics:
+- accuracy
+model-index:
+- name: pubmedbert-fulltext-cord19
+  results:
+  - task:
+      name: Masked Language Modeling
+      type: fill-mask
+    dataset:
+      name: pritamdeka/cord-19-fulltext
+      type: pritamdeka/cord-19-fulltext
+      args: fulltext
+    metrics:
+    - name: Accuracy
+      type: accuracy
+      value: 0.7175316733550737
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# pubmedbert-fulltext-cord19
+
+This model is a fine-tuned version of [microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext](https://huggingface.co/microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext) on the pritamdeka/cord-19-fulltext dataset.
+It achieves the following results on the evaluation set:
+- Loss: 1.2667
+- Accuracy: 0.7175
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 5e-05
+- train_batch_size: 16
+- eval_batch_size: 16
+- seed: 42
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: linear
+- lr_scheduler_warmup_steps: 10000
+- num_epochs: 3.0
+- mixed_precision_training: Native AMP
+
+### Training results
+
+| Training Loss | Epoch | Step  | Validation Loss | Accuracy |
+|:-------------:|:-----:|:-----:|:---------------:|:--------:|
+| 1.7985        | 0.27  | 5000  | 1.2710          | 0.7176   |
+| 1.7542        | 0.53  | 10000 | 1.3359          | 0.7070   |
+| 1.7462        | 0.8   | 15000 | 1.3489          | 0.7034   |
+| 1.8371        | 1.07  | 20000 | 1.4361          | 0.6891   |
+| 1.7102        | 1.33  | 25000 | 1.3502          | 0.7039   |
+| 1.6596        | 1.6   | 30000 | 1.3341          | 0.7065   |
+| 1.6265        | 1.87  | 35000 | 1.3228          | 0.7087   |
+| 1.605         | 2.13  | 40000 | 1.3079          | 0.7099   |
+| 1.5731        | 2.4   | 45000 | 1.2986          | 0.7121   |
+| 1.5602        | 2.67  | 50000 | 1.2929          | 0.7136   |
+| 1.5447        | 2.93  | 55000 | 1.2875          | 0.7143   |
+
+
+### Framework versions
+
+- Transformers 4.17.0.dev0
+- Pytorch 1.10.0+cu111
+- Datasets 1.18.3
+- Tokenizers 0.11.0

all_results.json

@@ -0,0 +1,15 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.7175316733550737,
+    "eval_loss": 1.2667090892791748,
+    "eval_runtime": 290.3233,
+    "eval_samples": 25000,
+    "eval_samples_per_second": 86.111,
+    "eval_steps_per_second": 5.384,
+    "perplexity": 3.549153375531065,
+    "train_loss": 1.6709515473090277,
+    "train_runtime": 32029.5992,
+    "train_samples": 300000,
+    "train_samples_per_second": 28.099,
+    "train_steps_per_second": 1.756
+}

config.json

@@ -0,0 +1,25 @@
+{
+  "_name_or_path": "microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext",
+  "architectures": [
+    "BertForMaskedLM"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "classifier_dropout": null,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "torch_dtype": "float32",
+  "transformers_version": "4.17.0.dev0",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 30522
+}

eval_results.json

@@ -0,0 +1,10 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.7175316733550737,
+    "eval_loss": 1.2667090892791748,
+    "eval_runtime": 290.3233,
+    "eval_samples": 25000,
+    "eval_samples_per_second": 86.111,
+    "eval_steps_per_second": 5.384,
+    "perplexity": 3.549153375531065
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9de4e6b38a338718fe5b5c173b47a07e06c3c23bb05d3f1dd34136e0592278e5
+size 438141995

special_tokens_map.json

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

tokenizer_config.json

@@ -0,0 +1 @@
+{"do_lower_case": true, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "special_tokens_map_file": null, "name_or_path": "microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext", "do_basic_tokenize": true, "never_split": null, "tokenizer_class": "BertTokenizer"}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 3.0,
+    "train_loss": 1.6709515473090277,
+    "train_runtime": 32029.5992,
+    "train_samples": 300000,
+    "train_samples_per_second": 28.099,
+    "train_steps_per_second": 1.756
+}

trainer_state.json

@@ -0,0 +1,190 @@
+{
+  "best_metric": 1.2710014581680298,
+  "best_model_checkpoint": "/content/pubmedbert-fulltext-cord19/checkpoint-5000",
+  "epoch": 3.0,
+  "global_step": 56250,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.4970000000000003e-05,
+      "loss": 1.7985,
+      "step": 5000
+    },
+    {
+      "epoch": 0.27,
+      "eval_accuracy": 0.7176016320015706,
+      "eval_loss": 1.2710014581680298,
+      "eval_runtime": 289.9232,
+      "eval_samples_per_second": 86.23,
+      "eval_steps_per_second": 5.391,
+      "step": 5000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9965e-05,
+      "loss": 1.7542,
+      "step": 10000
+    },
+    {
+      "epoch": 0.53,
+      "eval_accuracy": 0.7069832712721614,
+      "eval_loss": 1.3359251022338867,
+      "eval_runtime": 290.2058,
+      "eval_samples_per_second": 86.146,
+      "eval_steps_per_second": 5.386,
+      "step": 10000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.460432432432432e-05,
+      "loss": 1.7462,
+      "step": 15000
+    },
+    {
+      "epoch": 0.8,
+      "eval_accuracy": 0.7034289815931059,
+      "eval_loss": 1.3488777875900269,
+      "eval_runtime": 290.0812,
+      "eval_samples_per_second": 86.183,
+      "eval_steps_per_second": 5.388,
+      "step": 15000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9202162162162166e-05,
+      "loss": 1.8371,
+      "step": 20000
+    },
+    {
+      "epoch": 1.07,
+      "eval_accuracy": 0.6891450706903661,
+      "eval_loss": 1.436103105545044,
+      "eval_runtime": 289.9321,
+      "eval_samples_per_second": 86.227,
+      "eval_steps_per_second": 5.391,
+      "step": 20000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.379675675675676e-05,
+      "loss": 1.7102,
+      "step": 25000
+    },
+    {
+      "epoch": 1.33,
+      "eval_accuracy": 0.7039453286069087,
+      "eval_loss": 1.3501886129379272,
+      "eval_runtime": 289.995,
+      "eval_samples_per_second": 86.208,
+      "eval_steps_per_second": 5.39,
+      "step": 25000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8393513513513515e-05,
+      "loss": 1.6596,
+      "step": 30000
+    },
+    {
+      "epoch": 1.6,
+      "eval_accuracy": 0.70652987764824,
+      "eval_loss": 1.334069013595581,
+      "eval_runtime": 290.2224,
+      "eval_samples_per_second": 86.141,
+      "eval_steps_per_second": 5.386,
+      "step": 30000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2990270270270274e-05,
+      "loss": 1.6265,
+      "step": 35000
+    },
+    {
+      "epoch": 1.87,
+      "eval_accuracy": 0.708723656830319,
+      "eval_loss": 1.322791576385498,
+      "eval_runtime": 291.0022,
+      "eval_samples_per_second": 85.91,
+      "eval_steps_per_second": 5.371,
+      "step": 35000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7584864864864868e-05,
+      "loss": 1.605,
+      "step": 40000
+    },
+    {
+      "epoch": 2.13,
+      "eval_accuracy": 0.7099296658889697,
+      "eval_loss": 1.307939052581787,
+      "eval_runtime": 290.1974,
+      "eval_samples_per_second": 86.148,
+      "eval_steps_per_second": 5.386,
+      "step": 40000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.2181621621621623e-05,
+      "loss": 1.5731,
+      "step": 45000
+    },
+    {
+      "epoch": 2.4,
+      "eval_accuracy": 0.7121370440536863,
+      "eval_loss": 1.2985501289367676,
+      "eval_runtime": 290.9835,
+      "eval_samples_per_second": 85.916,
+      "eval_steps_per_second": 5.371,
+      "step": 45000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.778378378378379e-06,
+      "loss": 1.5602,
+      "step": 50000
+    },
+    {
+      "epoch": 2.67,
+      "eval_accuracy": 0.7135782251340592,
+      "eval_loss": 1.2928680181503296,
+      "eval_runtime": 290.2984,
+      "eval_samples_per_second": 86.118,
+      "eval_steps_per_second": 5.384,
+      "step": 50000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.3762162162162162e-06,
+      "loss": 1.5447,
+      "step": 55000
+    },
+    {
+      "epoch": 2.93,
+      "eval_accuracy": 0.7142735090199965,
+      "eval_loss": 1.2874841690063477,
+      "eval_runtime": 290.4326,
+      "eval_samples_per_second": 86.078,
+      "eval_steps_per_second": 5.382,
+      "step": 55000
+    },
+    {
+      "epoch": 3.0,
+      "step": 56250,
+      "total_flos": 1.156661775e+17,
+      "train_loss": 1.6709515473090277,
+      "train_runtime": 32029.5992,
+      "train_samples_per_second": 28.099,
+      "train_steps_per_second": 1.756
+    }
+  ],
+  "max_steps": 56250,
+  "num_train_epochs": 3,
+  "total_flos": 1.156661775e+17,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:32791723132d9fc18f04a12e5de8e21d799f30360d75d11d4180d5cf9a33a815
+size 3055