Yi-6B-zhihu5

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

• Loss: 2.5677

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-06
• 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: 20

Training results

Training Loss	Epoch	Step	Validation Loss
2.6816	1.0	96	2.6328
2.6582	2.0	192	2.6169
2.6676	3.0	288	2.5983
2.6413	4.0	384	2.5891
2.581	5.0	480	2.5825
2.5884	6.0	576	2.5776
2.704	7.0	672	2.5741
2.608	8.0	768	2.5715
2.5454	9.0	864	2.5698
2.5938	10.0	960	2.5688
2.6129	11.0	1056	2.5682
2.6334	12.0	1152	2.5679
2.5013	13.0	1248	2.5678
2.519	14.0	1344	2.5677
2.5076	15.0	1440	2.5677
2.5443	16.0	1536	2.5677
2.5972	17.0	1632	2.5677
2.5361	18.0	1728	2.5677
2.6119	19.0	1824	2.5677
2.6321	20.0	1920	2.5677

Framework versions

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