model update

README.md

@@ -0,0 +1,137 @@
+---
+datasets:
+- tner/tweetner7
+metrics:
+- f1
+- precision
+- recall
+model-index:
+- name: cardiffnlp/twitter-roberta-base-2022-154m-tweetner7-2020
+  results:
+  - task:
+      name: Token Classification
+      type: token-classification
+    dataset:
+      name: tner/tweetner7
+      type: tner/tweetner7
+      args: tner/tweetner7
+    metrics:
+    - name: F1
+      type: f1
+      value: 0.6419150543257219
+    - name: Precision
+      type: precision
+      value: 0.6451010159990658
+    - name: Recall
+      type: recall
+      value: 0.6387604070305273
+    - name: F1 (macro)
+      type: f1_macro
+      value: 0.5829431071584856
+    - name: Precision (macro)
+      type: precision_macro
+      value: 0.5886989381701707
+    - name: Recall (macro)
+      type: recall_macro
+      value: 0.5796110916728531
+    - name: F1 (entity span)
+      type: f1_entity_span
+      value: 0.7753631609529343
+    - name: Precision (entity span)
+      type: precision_entity_span
+      value: 0.7791661800770758
+    - name: Recall (entity span)
+      type: recall_entity_span
+      value: 0.7715970856944605
+
+pipeline_tag: token-classification
+widget:
+- text: "Jacob Collier is a Grammy awarded artist from England."
+  example_title: "NER Example 1"
+---
+# cardiffnlp/twitter-roberta-base-2022-154m-tweetner7-2020
+
+This model is a fine-tuned version of [cardiffnlp/twitter-roberta-base-2022-154m](https://huggingface.co/cardiffnlp/twitter-roberta-base-2022-154m) on the 
+[tner/tweetner7](https://huggingface.co/datasets/tner/tweetner7) dataset.
+Model fine-tuning is done via [T-NER](https://github.com/asahi417/tner)'s hyper-parameter search (see the repository
+for more detail). It achieves the following results on the test set:
+- F1 (micro): 0.6419150543257219
+- Precision (micro): 0.6451010159990658
+- Recall (micro): 0.6387604070305273
+- F1 (macro): 0.5829431071584856
+- Precision (macro): 0.5886989381701707
+- Recall (macro): 0.5796110916728531
+
+The per-entity breakdown of the F1 score on the test set are below:
+- corporation: 0.5127020785219399
+- event: 0.43384759233286585
+- group: 0.6000666000666002
+- location: 0.6535326086956522
+- person: 0.8390577234310376
+- product: 0.6386386386386387
+- work_of_art: 0.40275650842266464 
+
+For F1 scores, the confidence interval is obtained by bootstrap as below:
+- F1 (micro): 
+ 
+- F1 (macro): 
+ 
+
+Full evaluation can be found at [metric file of NER](https://huggingface.co/cardiffnlp/twitter-roberta-base-2022-154m-tweetner7-2020/raw/main/eval/metric.json) 
+and [metric file of entity span](https://huggingface.co/cardiffnlp/twitter-roberta-base-2022-154m-tweetner7-2020/raw/main/eval/metric_span.json).
+
+### Usage
+This model can be used through the [tner library](https://github.com/asahi417/tner). Install the library via pip   
+```shell
+pip install tner
+```
+and activate model as below.
+```python
+from tner import TransformersNER
+model = TransformersNER("cardiffnlp/twitter-roberta-base-2022-154m-tweetner7-2020")
+model.predict(["Jacob Collier is a Grammy awarded English artist from London"])
+```
+It can be used via transformers library but it is not recommended as CRF layer is not supported at the moment.
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+ - dataset: ['tner/tweetner7']
+ - dataset_split: train_2020
+ - dataset_name: None
+ - local_dataset: None
+ - model: cardiffnlp/twitter-roberta-base-2022-154m
+ - crf: True
+ - max_length: 128
+ - epoch: 30
+ - batch_size: 32
+ - lr: 0.0001
+ - random_seed: 42
+ - gradient_accumulation_steps: 1
+ - weight_decay: 1e-07
+ - lr_warmup_step_ratio: 0.3
+ - max_grad_norm: 10
+
+The full configuration can be found at [fine-tuning parameter file](https://huggingface.co/cardiffnlp/twitter-roberta-base-2022-154m-tweetner7-2020/raw/main/trainer_config.json).
+
+### Reference
+If you use any resource from T-NER, please consider to cite our [paper](https://aclanthology.org/2021.eacl-demos.7/).
+
+```
+
+@inproceedings{ushio-camacho-collados-2021-ner,
+    title = "{T}-{NER}: An All-Round Python Library for Transformer-based Named Entity Recognition",
+    author = "Ushio, Asahi  and
+      Camacho-Collados, Jose",
+    booktitle = "Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations",
+    month = apr,
+    year = "2021",
+    address = "Online",
+    publisher = "Association for Computational Linguistics",
+    url = "https://aclanthology.org/2021.eacl-demos.7",
+    doi = "10.18653/v1/2021.eacl-demos.7",
+    pages = "53--62",
+    abstract = "Language model (LM) pretraining has led to consistent improvements in many NLP downstream tasks, including named entity recognition (NER). In this paper, we present T-NER (Transformer-based Named Entity Recognition), a Python library for NER LM finetuning. In addition to its practical utility, T-NER facilitates the study and investigation of the cross-domain and cross-lingual generalization ability of LMs finetuned on NER. Our library also provides a web app where users can get model predictions interactively for arbitrary text, which facilitates qualitative model evaluation for non-expert programmers. We show the potential of the library by compiling nine public NER datasets into a unified format and evaluating the cross-domain and cross- lingual performance across the datasets. The results from our initial experiments show that in-domain performance is generally competitive across datasets. However, cross-domain generalization is challenging even with a large pretrained LM, which has nevertheless capacity to learn domain-specific features if fine- tuned on a combined dataset. To facilitate future research, we also release all our LM checkpoints via the Hugging Face model hub.",
+}
+
+```

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "cner_output/model/baseline/t_roberta_base_2022/best_model",
+  "_name_or_path": "cardiffnlp/twitter-roberta-base-2022-154m",
   "architectures": [
     "RobertaForTokenClassification"
   ],

eval/metric.json

@@ -0,0 +1 @@
+{"micro/f1": 0.6419150543257219, "micro/f1_ci": {}, "micro/recall": 0.6387604070305273, "micro/precision": 0.6451010159990658, "macro/f1": 0.5829431071584856, "macro/f1_ci": {}, "macro/recall": 0.5796110916728531, "macro/precision": 0.5886989381701707, "per_entity_metric": {"corporation": {"f1": 0.5127020785219399, "f1_ci": {}, "precision": 0.5336538461538461, "recall": 0.49333333333333335}, "event": {"f1": 0.43384759233286585, "f1_ci": {}, "precision": 0.4461538461538462, "recall": 0.4222020018198362}, "group": {"f1": 0.6000666000666002, "f1_ci": {}, "precision": 0.6067340067340067, "recall": 0.5935441370223979}, "location": {"f1": 0.6535326086956522, "f1_ci": {}, "precision": 0.6362433862433863, "recall": 0.6717877094972067}, "person": {"f1": 0.8390577234310376, "f1_ci": {}, "precision": 0.8188838188838189, "recall": 0.8602507374631269}, "product": {"f1": 0.6386386386386387, "f1_ci": {}, "precision": 0.621832358674464, "recall": 0.6563786008230452}, "work_of_art": {"f1": 0.40275650842266464, "f1_ci": {}, "precision": 0.4573913043478261, "recall": 0.359781121751026}}}

eval/metric.test_2020.json

@@ -0,0 +1 @@
+{"micro/f1": 0.638988177069013, "micro/f1_ci": {}, "micro/recall": 0.6030098598858329, "micro/precision": 0.6795321637426901, "macro/f1": 0.5967724585385162, "macro/f1_ci": {}, "macro/recall": 0.5601050750558658, "macro/precision": 0.6414056489059751, "per_entity_metric": {"corporation": {"f1": 0.5411764705882353, "f1_ci": {}, "precision": 0.6174496644295302, "recall": 0.4816753926701571}, "event": {"f1": 0.47265625000000006, "f1_ci": {}, "precision": 0.4898785425101215, "recall": 0.45660377358490567}, "group": {"f1": 0.5499999999999999, "f1_ci": {}, "precision": 0.6184738955823293, "recall": 0.49517684887459806}, "location": {"f1": 0.6394984326018808, "f1_ci": {}, "precision": 0.6623376623376623, "recall": 0.6181818181818182}, "person": {"f1": 0.810580204778157, "f1_ci": {}, "precision": 0.8246527777777778, "recall": 0.7969798657718121}, "product": {"f1": 0.6682577565632458, "f1_ci": {}, "precision": 0.7035175879396985, "recall": 0.6363636363636364}, "work_of_art": {"f1": 0.49523809523809514, "f1_ci": {}, "precision": 0.5735294117647058, "recall": 0.43575418994413406}}}

eval/metric_span.json

@@ -0,0 +1 @@
+{"micro/f1": 0.7753631609529343, "micro/f1_ci": {}, "micro/recall": 0.7715970856944605, "micro/precision": 0.7791661800770758, "macro/f1": 0.7753631609529343, "macro/f1_ci": {}, "macro/recall": 0.7715970856944605, "macro/precision": 0.7791661800770758}

eval/metric_span.test_2020.json

@@ -0,0 +1 @@
+{"micro/f1": 0.7462194116029693, "micro/f1_ci": {}, "micro/recall": 0.7042034250129735, "micro/precision": 0.7935672514619883, "macro/f1": 0.7462194116029693, "macro/f1_ci": {}, "macro/recall": 0.7042034250129735, "macro/precision": 0.7935672514619883}

tokenizer_config.json

@@ -6,7 +6,7 @@
   "errors": "replace",
   "mask_token": "<mask>",
   "model_max_length": 512,
-  "name_or_path": "cner_output/model/baseline/t_roberta_base_2022/best_model",
+  "name_or_path": "cardiffnlp/twitter-roberta-base-2022-154m",
   "pad_token": "<pad>",
   "sep_token": "</s>",
   "special_tokens_map_file": null,

trainer_config.json

@@ -0,0 +1 @@
+{"dataset": ["tner/tweetner7"], "dataset_split": "train_2020", "dataset_name": null, "local_dataset": null, "model": "cardiffnlp/twitter-roberta-base-2022-154m", "crf": true, "max_length": 128, "epoch": 30, "batch_size": 32, "lr": 0.0001, "random_seed": 42, "gradient_accumulation_steps": 1, "weight_decay": 1e-07, "lr_warmup_step_ratio": 0.3, "max_grad_norm": 10}