adl-hw3-finetune-gemma-2-chinese-kyara-2

This model is a fine-tuned version of zake7749/gemma-2-2b-it-chinese-kyara-dpo on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.9932

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
2.3974	0.1778	25	2.1847
2.1716	0.3556	50	2.0972
2.1123	0.5333	75	2.0470
2.0123	0.7111	100	2.0131
1.9901	0.8889	125	1.9926
1.9153	1.0667	150	1.9910
1.7569	1.2444	175	1.9843
1.7971	1.4222	200	1.9748
1.8106	1.6	225	1.9597
1.7733	1.7778	250	1.9526
1.7275	1.9556	275	1.9500
1.6153	2.1333	300	1.9988
1.5536	2.3111	325	1.9955
1.5153	2.4889	350	1.9992
1.5445	2.6667	375	1.9893
1.544	2.8444	400	1.9932

Framework versions

• PEFT 0.13.2
• Transformers 4.45.1
• Pytorch 2.5.0+cu121
• Datasets 3.1.0
• Tokenizers 0.20.2