metadata

license: apache-2.0
base_model: Twitter/twhin-bert-large
tags:
  - generated_from_trainer
metrics:
  - accuracy
  - f1
model-index:
  - name: financial-twhin-bert-large-3labels-pesudo
    results: []

financial-twhin-bert-large-3labels-dev2

This model is a fine-tuned version of Twitter/twhin-bert-large on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.3035
Accuracy: 0.8948
F1: 0.8932

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: 3.671821890679145e-05
train_batch_size: 16
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.1
num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.8977	0.15	100	0.6404	0.7428	0.6732
0.7036	0.3	200	0.8094	0.6326	0.6576
0.6344	0.46	300	0.5222	0.7961	0.8068
0.5275	0.61	400	0.4322	0.8372	0.8425
0.4899	0.76	500	0.5061	0.8249	0.8320
0.4723	0.91	600	0.3711	0.8631	0.8577
0.4345	1.06	700	0.3421	0.8833	0.8819
0.3524	1.21	800	0.3322	0.8811	0.8829
0.3288	1.37	900	0.4062	0.8696	0.8724
0.3566	1.52	1000	0.3035	0.8948	0.8932
0.3253	1.67	1100	0.3064	0.8934	0.8938
0.2861	1.82	1200	0.3336	0.8927	0.8932
0.2955	1.97	1300	0.3526	0.8970	0.8959
0.1759	2.12	1400	0.4498	0.8948	0.8942
0.171	2.28	1500	0.3863	0.8955	0.8950
0.1636	2.43	1600	0.4361	0.8869	0.8889
0.1431	2.58	1700	0.4124	0.8963	0.8967
0.1661	2.73	1800	0.4062	0.8984	0.8989
0.1549	2.88	1900	0.4041	0.8999	0.9002

Framework versions

Transformers 4.37.2
Pytorch 2.1.0+cu121
Datasets 2.17.1
Tokenizers 0.15.2