README.md

metadata

tags:
  - generated_from_trainer
metrics:
  - accuracy
model-index:
  - name: vietcuna-3b_2048
    results: []

vietcuna-3b_2048

This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5250
• Accuracy: 0.7375

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.18
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.5694	1.05	50	0.5834	0.7087
0.5614	2.1	100	0.5772	0.7165
0.5475	3.15	150	0.5684	0.7165
0.5503	4.2	200	0.5605	0.7087
0.5305	5.25	250	0.5784	0.7192
0.5353	6.3	300	0.5451	0.7323
0.5063	7.35	350	0.5441	0.7270
0.5141	8.4	400	0.5365	0.7244
0.5035	9.45	450	0.5354	0.7297
0.493	10.5	500	0.5322	0.7297
0.4763	11.55	550	0.5299	0.7375
0.5063	12.6	600	0.5295	0.7375
0.4787	13.65	650	0.5280	0.7297
0.4841	14.7	700	0.5266	0.7375
0.4732	15.75	750	0.5283	0.7297
0.4801	16.8	800	0.5259	0.7375
0.4651	17.85	850	0.5256	0.7375
0.4726	18.9	900	0.5260	0.7323
0.4758	19.95	950	0.5248	0.7375
0.4701	21.0	1000	0.5250	0.7375

Framework versions

• Transformers 4.35.0.dev0
• Pytorch 2.1.0+cu121
• Datasets 2.14.6
• Tokenizers 0.14.1