bert-base-cased-finetuned-Stromberg_NLP_Twitter-PoS

This model is a fine-tuned version of bert-base-cased on the twitter_pos_vcb dataset. It achieves the following results on the evaluation set:

• Loss: 0.0533

• '''

  • Precision: 0.9580645161290322
  • Recall: 0.9519230769230769
  • F1: 0.954983922829582
  • Number': 312

• B

  • Precision: 0.9658270558694287
  • Recall: 0.9655240037652966
  • F1: 0.9656755060411109
  • Number: 25496

• Bd

  • Precision: 0.9630099728014506
  • Recall: 0.9572819033886085
  • F1: 0.9601373949200036
  • Number: 5548

• Bg

  • Precision: 0.9836065573770492
  • Recall: 0.9853434575313438
  • F1: 0.9844742413549753
  • Number: 5663

• Bn

  • Precision: 0.9182209469153515
  • Recall: 0.9116809116809117
  • F1: 0.9149392423159399
  • Number: 2106

• Bp

  • Precision: 0.9672037914691943
  • Recall: 0.9663488856619736
  • F1: 0.9667761495704902
  • Number': 15839

• Br

  • Precision: 0.94
  • Recall: 0.8785046728971962
  • F1: 0.9082125603864735
  • Number': 107

• Bs

  • Precision: 0.9848484848484849
  • Recall': 0.9701492537313433
  • F1: 0.9774436090225564
  • Number': 67

• Bz

  • Precision: 0.9865819209039548
  • Recall: 0.9850167459897762
  • F1: 0.9857987121813531
  • Number': 5673

• C

  • Precision: 0.9993461203138623,
  • Recall: 0.9993461203138623,
  • F1: 0.9993461203138623,
  • Number: 4588

• D

  • Precision: 0.9876836325864372
  • Recall: 0.9895926256318763
  • F1: 0.988637207575195
  • Number: 6726

• Dt

  • Precision: 1.0
  • Recall: 0.8
  • F1: 0.888888888888889
  • Number: 15

• H

  • Precision: 0.9487382595903587
  • Recall: 0.9305216426193119
  • F1: 0.9395416596626883
  • Number: 9010

• J

  • Precision: 0.9803528468323978
  • Recall: 0.980588754311382
  • F1: 0.9804707863816818
  • Number: 12467

• Jr

  • Precision: 0.9400386847195358
  • Recall: 0.9818181818181818
  • F1': 0.9604743083003953
  • Number': 495

• Js

  • Precision: 0.9612141652613828
  • Recall: 0.991304347826087
  • F1: 0.9760273972602741
  • Number': 575

• N

  • Precision: 0.9795543362923471
  • Recall: 0.9793769083475651
  • F1: 0.9794656142847902
  • Number': 38646

• Np

  • Precision: 0.9330242966751918
  • Recall: 0.9278334128119536
  • F1: 0.9304216147286205
  • Number': 6291

• Nps

  • Precision: 0.75
  • Recall: 0.23076923076923078
  • F1: 0.3529411764705882
  • Number: 26

• Ns

  • Precision: 0.9691858990616282
  • Recall: 0.9773657289002557
  • F1: 0.9732586272762003
  • Number': 7820

• O

  • Precision: 0.9984323288625675
  • Recall: 0.999302649930265
  • F1: 0.9988672998170254
  • Number: 5736

• Os

  • Precision: 1.0
  • Recall: 0.9952267303102625
  • F1: 0.9976076555023923
  • Number: 419

• P

  • Precision: 0.9887869520897044
  • Recall: 0.9918200408997955
  • F1: 0.9903011740684022
  • Number: 2934

• Rb

  • Precision: 0.9971910112359551
  • Recall: 0.9983929288871033
  • F1: 0.9977916081108211
  • Number: 2489

• Rl

  • Precision: 1.0
  • Recall: 0.9997228381374723
  • F1: 0.9998613998613999
  • Number: 3608

• Rp

  • Precision: 0.9979960600502683
  • Recall: 0.9980638586956522
  • F1: 0.9980299582215278
  • Number: 29440

• Rp$

  • Precision: 0.9975770162686051
  • Recall: 0.9972318339100346
  • F1: 0.9974043952240872
  • Number: 5780

• Sr

  • Precision: 0.9998923110058152
  • Recall: 0.9998384752059442
  • F1: 0.9998653923812088
  • Number: 18573

• T

  • Precision: 0.9987569919204475
  • Recall: 0.9984811874352779
  • F1: 0.9986190706345371
  • Number: 28970

• W

  • Precision: 0.0
  • Recall: 0.0
  • F1: 0.0
  • number: 1

• X

  • Precision: 0.9466666666666667,
  • Recall: 0.9594594594594594,
  • F1 0.9530201342281879,
  • Number: 74}

• Ym

  • Precision: 0.0
  • Recall: 0.0
  • F1: 0.0
  • Number: 5

• ' '

  • Precision: 0.9951481772882245
  • Recall: 0.9949524745984923
  • F1: 0.9950503163208444
  • Number: 15255

• '`'

  • Precision: 0.9540229885057471
  • Recall: 0.9595375722543352
  • F1: 0.956772334293948
  • Number: 173

• Overall

  • Precision: 0.9828
  • Recall: 0.9820
  • F1: 0.9824
  • Accuracy: 0.9860

Model description

For more information on how it was created, check out the following link: https://github.com/DunnBC22/NLP_Projects/blob/main/Token%20Classification/Monolingual/StrombergNLP-Twitter_pos_vcb/NER%20Project%20Using%20StrombergNLP%20Twitter_pos_vcb%20Dataset.ipynb

Intended uses & limitations

This model is intended to demonstrate my ability to solve a complex problem using technology.

Training and evaluation data

Dataset Source: https://huggingface.co/datasets/strombergnlp/twitter_pos_vcb

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-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
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	''' Precision	''' Recall	''' F1	''' Number	B Precision	B Recall	B F1	B Number	Bd Precision	Bd Recall	Bd F1	Bd Number	Bg Precision	Bg Recall	Bg F1	Bg Number	Bn Precision	Bn Recall	Bn F1	Bn Number	Bp Precision	Bp Recall	Bp F1	Bp Number	Br Precision	Br Recall	Br F1	Br Number	Bs precision	Bs Recall	Bs F1	Bs Number	Bz Precision	Bz Recall	Bz F1	Bz Number	C Precision	C Recall	C F1	C Number	D Precision	D Recall	D F1	D Number	Dt Precision	Dt Recall	Dt F1	Dt Number	H Precision	H Recall	H F1	H Number	J Precision	J Recall	J F1	J Number	Jr Precision	Jr Recall	Jr F1	Jr Number	Js Precision	Js Recall	Js F1	Js Number	N Precision	N Recall	N F1	N Number	Np Precision	Np Recall	Np F1	Np Number	Nps Precision	Nps Recall	Nps F1	Nps Number	Ns Precision	Ns Recall	Ns F1	Ns Number	O Precision	O Recall	O F1	O Number	Os Precision	Os Recall	Os F1	Os Number	P Precision	P Recall	P F1	P Number	Rb Precision	Rb Recall	Rb f1	Rb Number	Rl Precision	Rl Recall	Rl F1	Rl Number	Rp Precision	Rp Recall	Rp F1	Rp Number	Rp$ Precision	Rp$ Recall	Rp$ F1	Rp$ Number	Sr Precision	Sr Recall	Sr F1	Sr Number	T Precision	T recall	T F1	T Number	W Precision	W Recall	W F1	W Number	X Precision	X Recall	X F1	X Number	Ym Precision	Ym Recall	Ym F1	Ym Number	' ' Precision	' ' Recall	' ' F1	' ' Number	'`' Precision	'`' Recall	'`' F1	'`' Number	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
0.0617	1.0	7477	0.0595	0.9331	0.9391	0.9361	312	0.9563	0.9536	0.9550	25496	0.9716	0.9322	0.9515	5548	0.9811	0.9786	0.9798	5663	0.8725	0.9231	0.8971	2106	0.9556	0.9586	0.9571	15839	0.8879	0.8879	0.8879	107	0.8590	1.0	0.9241	67	0.9793	0.9834	0.9814	5673	0.9985	0.9991	0.9988	4588	0.9818	0.9886	0.9852	6726	1.0	0.8	0.8889	15	0.9391	0.9105	0.9246	9010	0.9707	0.9766	0.9736	12467	0.9212	0.9677	0.9438	495	0.9227	0.9757	0.9484	575	0.9754	0.9738	0.9746	38646	0.9158	0.9200	0.9179	6291	0.0	0.0	0.0	26	0.9657	0.9688	0.9673	7820	0.9972	0.9990	0.9981	5736	1.0	0.9928	0.9964	419	0.9771	0.9908	0.9839	2934	0.9948	0.9968	0.9958	2489	1.0	0.9997	0.9999	3608	0.9970	0.9976	0.9973	29440	0.9974	0.9954	0.9964	5780	0.9998	0.9998	0.9998	18573	0.9977	0.9982	0.9979	28970	0.0	0.0	0.0	1	0.8861	0.9459	0.9150	74	0.0	0.0	0.0	5	0.9936	0.9926	0.9931	15255	0.9540	0.9595	0.9568	173	0.9779	0.9772	0.9775	0.9821
0.0407	2.0	14954	0.0531	0.9605	0.9359	0.9481	312	0.9599	0.9646	0.9622	25496	0.9674	0.9459	0.9565	5548	0.9834	0.9825	0.9830	5663	0.8920	0.9259	0.9087	2106	0.9728	0.9569	0.9648	15839	0.9592	0.8785	0.9171	107	0.9429	0.9851	0.9635	67	0.9890	0.9825	0.9858	5673	0.9991	0.9993	0.9992	4588	0.9855	0.9896	0.9875	6726	1.0	0.8	0.8889	15	0.9498	0.9303	0.9399	9010	0.9776	0.9797	0.9786	12467	0.9125	0.9899	0.9496	495	0.9481	0.9843	0.9659	575	0.9788	0.9771	0.9779	38646	0.9252	0.9285	0.9268	6291	0.5	0.2308	0.3158	26	0.96534	0.9769	0.9711	7820	0.9976	0.9993	0.9984	5736	0.9929	0.9952	0.9940	419	0.9861	0.9928	0.9895	2934	0.9972	0.9984	0.9978	2489	1.0	0.9997	0.9999	3608	0.9986	0.9982	0.9984	29440	0.9964	0.9978	0.9971	5780	0.9999	0.9999	0.9999	18573	0.9985	0.9983	0.9984	28970	0.0	0.0	0.0	1	0.9114	0.9730	0.9412	74	0.0	0.0	0.0	5	0.9949	0.9961	0.9955	15255	0.9651	0.9595	0.9623	173	0.9817	0.9808	0.9813	0.9850
0.0246	3.0	22431	0.0533	0.9581	0.9519	0.9550	312	0.9658	0.9655	0.9657	25496	0.9630	0.9573	0.9601	5548	0.9836	0.9853	0.9845	5663	0.9182	0.9117	0.9149	2106	0.9672	0.9663	0.9668	15839	0.94	0.8785	0.9082	107	0.9848	0.9701	0.9774	67	0.9866	0.9850	0.9858	5673	0.9993	0.9993	0.9993	4588	0.9877	0.9896	0.9886	6726	1.0	0.8	0.8889	15	0.9487	0.9305	0.9395	9010	0.9804	0.9806	0.9805	12467	0.9400	0.9818	0.9605	495	0.9612	0.9913	0.9760	575	0.9796	0.9794	0.9795	38646	0.9330	0.9278	0.9304	6291	0.75	0.2308	0.3529	26	0.9692	0.9774	0.9733	7820	0.9984	0.9993	0.9989	5736	1.0	0.9952	0.9976	419	0.9888	0.9918	0.9903	2934	0.9972	0.9984	0.9978	2489	1.0	0.9997	0.9999	3608	0.9980	0.9981	0.9981	29440	0.9976	0.9972	0.9974	5780	0.9999	0.9998	0.9999	18573	0.9988	0.9985	0.9986	28970	0.0	0.0	0.0	1	0.9467	0.9595	0.9530	74	0.0	0.0	0.0	5	0.9951	0.9950	0.9951	15255	0.9540	0.9595	0.9568	173	0.9828	0.9820	0.9824	0.9860

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0
• Datasets 2.11.0
• Tokenizers 0.13.3