This model is a fine-tuned version of saves/gemma2-2b-sft-zh on the gemma2_2b_train_dpo_zh dataset. It achieves the following results on the evaluation set:

Loss: 0.2930
Rewards/chosen: -6.2456
Rewards/rejected: -28.9376
Rewards/accuracies: 0.9433
Rewards/margins: 22.6921
Logps/chosen: -260.4318
Logps/rejected: -367.5450
Logits/chosen: -4.3646
Logits/rejected: -4.4725

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-07
train_batch_size: 1
eval_batch_size: 1
seed: 42
distributed_type: multi-GPU
num_devices: 2
gradient_accumulation_steps: 4
total_train_batch_size: 8
total_eval_batch_size: 2
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.1
num_epochs: 1.0

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/chosen	Logps/rejected	Logits/chosen	Logits/rejected
0.0513	0.3741	3000	0.3045	-4.9056	-25.4557	0.9369	20.5501	-247.0327	-332.7262	-4.2631	-4.2778
0.0185	0.7482	6000	0.2908	-6.0224	-29.2477	0.9424	23.2253	-258.2001	-370.6453	-4.4148	-4.5147

Framework versions

Transformers 4.57.1
Pytorch 2.11.0+cu126
Datasets 4.0.0
Tokenizers 0.22.2

saiteja33/gemma2-2b-dpo-zh

gemma2-2b-dpo-zh

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

saiteja33/
gemma2-2b-dpo-zh