This model is a fine-tuned version of saves/llama32-3b-sft-zh on the llama32_3b_train_dpo_zh dataset. It achieves the following results on the evaluation set:

Loss: 0.1999
Rewards/chosen: -3.5686
Rewards/rejected: -27.1115
Rewards/accuracies: 0.9518
Rewards/margins: 23.5429
Logps/chosen: -214.1917
Logps/rejected: -356.6713
Logits/chosen: -0.8852
Logits/rejected: -1.0729

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
0.4794	0.3446	3000	0.1716	-2.1138	-18.4894	0.9445	16.3756	-199.6436	-270.4499	-0.6044	-0.7605
0.2678	0.6892	6000	0.1801	-3.2023	-25.7884	0.9526	22.5862	-210.5281	-343.4401	-0.8027	-0.9928

Framework versions

Transformers 4.57.1
Pytorch 2.11.0+cu126
Datasets 4.0.0
Tokenizers 0.22.2

saiteja33/llama32-3b-dpo-zh

llama32-3b-dpo-zh

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

saiteja33/
llama32-3b-dpo-zh