README.md

metadata

license: mit
tags:
  - generated_from_trainer
metrics:
  - f1
model-index:
  - name: newsdiscourse-model
    results: []

newsdiscourse-model

This model is a fine-tuned version of roberta-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.7049
• F1: 0.5472

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: 3e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5.0

Training results

Training Loss	Epoch	Step	Validation Loss	F1
No log	0.14	100	1.3954	0.3736
No log	0.28	200	1.2494	0.4403
No log	0.43	300	1.1859	0.4049
No log	0.57	400	1.3166	0.4692
1.2082	0.71	500	1.3011	0.4667
1.2082	0.85	600	1.2637	0.4909
1.2082	1.0	700	1.1596	0.4643
1.2082	1.14	800	1.2012	0.4954
1.2082	1.28	900	1.1207	0.4856
0.9276	1.42	1000	1.3099	0.5074
0.9276	1.57	1100	1.2627	0.4821
0.9276	1.71	1200	1.1202	0.5034
0.9276	1.85	1300	1.1611	0.5022
0.9276	1.99	1400	1.2114	0.5191
0.8288	2.14	1500	1.2759	0.5078
0.8288	2.28	1600	1.3322	0.5286
0.8288	2.42	1700	1.2991	0.5301
0.8288	2.56	1800	1.2623	0.5004
0.8288	2.71	1900	1.3173	0.5245
0.6347	2.85	2000	1.3929	0.5318
0.6347	2.99	2100	1.3334	0.5383
0.6347	3.13	2200	1.5554	0.5275
0.6347	3.28	2300	1.5034	0.5592
0.6347	3.42	2400	1.5118	0.5716
0.4923	3.56	2500	1.4939	0.5211
0.4923	3.7	2600	1.5325	0.5485
0.4923	3.85	2700	1.5297	0.5553
0.4923	3.99	2800	1.5026	0.5420
0.4923	4.13	2900	1.5561	0.5461
0.3913	4.27	3000	1.6237	0.5420
0.3913	4.42	3100	1.6495	0.5506
0.3913	4.56	3200	1.7337	0.5502
0.3913	4.7	3300	1.7264	0.5446
0.3913	4.84	3400	1.7198	0.5450
0.3154	4.99	3500	1.7049	0.5472

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.1+cu117
• Datasets 2.13.1
• Tokenizers 0.13.3