README.md

metadata

license: apache-2.0
tags:
  - generated_from_trainer
metrics:
  - precision
  - recall
  - f1
  - accuracy
model-index:
  - name: spellcorrector_1009_v4
    results: []

spellcorrector_1009_v4

This model is a fine-tuned version of google/canine-s on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3307
• Precision: 0.9681
• Recall: 0.9681
• F1: 0.9681
• Accuracy: 0.9573

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: 2e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.2767	1.0	1951	0.2331	0.96	0.9562	0.9581	0.9383
0.2181	2.0	3902	0.2028	0.9524	0.9562	0.9543	0.9450
0.1776	3.0	5853	0.2019	0.96	0.9562	0.9581	0.9498
0.1519	4.0	7804	0.2038	0.9526	0.9602	0.9563	0.9498
0.1277	5.0	9755	0.2091	0.9567	0.9681	0.9624	0.9521
0.1133	6.0	11706	0.2187	0.9449	0.9562	0.9505	0.9540
0.1041	7.0	13657	0.2378	0.9762	0.9801	0.9781	0.9545
0.0906	8.0	15608	0.2371	0.9603	0.9641	0.9622	0.9558
0.0806	9.0	17559	0.2509	0.976	0.9721	0.9741	0.9532
0.0689	10.0	19510	0.2624	0.9681	0.9681	0.9681	0.9563
0.0623	11.0	21461	0.2623	0.976	0.9721	0.9741	0.9559
0.06	12.0	23412	0.2783	0.9643	0.9681	0.9662	0.9564
0.0537	13.0	25363	0.2938	0.976	0.9721	0.9741	0.9569
0.0507	14.0	27314	0.2976	0.9603	0.9641	0.9622	0.9565
0.0491	15.0	29265	0.3075	0.9681	0.9681	0.9681	0.9576
0.0426	16.0	31216	0.3182	0.9681	0.9681	0.9681	0.9571
0.0426	17.0	33167	0.3154	0.9681	0.9681	0.9681	0.9572
0.0387	18.0	35118	0.3266	0.9681	0.9681	0.9681	0.9573
0.0336	19.0	37069	0.3317	0.9681	0.9681	0.9681	0.9574
0.0341	20.0	39020	0.3307	0.9681	0.9681	0.9681	0.9573

Framework versions

• Transformers 4.28.0
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3