This model is a fine-tuned version of sathiiiii/polyalign-llama3.2-3b-en-sft on the polyalign_dpo_train dataset. It achieves the following results on the evaluation set:

Loss: 0.2799
Rewards/chosen: -0.7350
Rewards/rejected: -5.4304
Rewards/accuracies: 0.9203
Rewards/margins: 4.6954
Logps/chosen: -93.6754
Logps/rejected: -84.5534
Logits/chosen: -2.1450
Logits/rejected: -2.0711

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: 8
gradient_accumulation_steps: 8
total_train_batch_size: 64
total_eval_batch_size: 8
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
0.2367	0.4181	3000	0.3013	-0.7480	-5.0461	0.9108	4.2981	-93.8062	-80.7107	-2.0358	-1.9547
0.2218	0.8362	6000	0.2799	-0.7350	-5.4304	0.9203	4.6954	-93.6754	-84.5534	-2.1450	-2.0711

Framework versions

Transformers 4.56.2
Pytorch 2.9.1+rocm6.3
Datasets 4.0.0
Tokenizers 0.22.2

saiteja33/llama-3.2-3b-dpo-en

polyalign

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

saiteja33/
llama-3.2-3b-dpo-en