Yi-6B-ruozhiba-1e-5

This model is a fine-tuned version of 01-ai/Yi-6B on the ruozhiba dataset. It achieves the following results on the evaluation set:

• Loss: 1.9852

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: 1e-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: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss
1.9439	2.0	110	2.0206
1.8731	3.0	165	1.9055
1.7574	4.0	220	1.8510
1.7266	5.0	275	1.8366
1.6036	7.0	385	1.8308
1.7214	8.0	440	1.8380
1.5245	9.0	495	1.8495
1.5239	10.0	550	1.8638
1.4286	11.0	605	1.8771
1.3534	12.0	660	1.9030
1.3895	14.0	770	1.9447
1.3721	15.0	825	1.9617
1.3598	16.0	880	1.9719
1.3015	17.0	935	1.9796
1.3456	18.0	990	1.9831
1.2136	19.0	1045	1.9848

Framework versions

• PEFT 0.7.1
• Transformers 4.36.2
• Pytorch 2.2.2+cu118
• Datasets 2.14.6
• Tokenizers 0.15.2