This model is a fine-tuned version of W-61/llama-3-8b-base-sft-ultrachat-8xh200 on the HuggingFaceH4/ultrafeedback_binarized dataset. It achieves the following results on the evaluation set:

Loss: 0.7668
Beta Dpo/gap Mean: 15.9231
Beta Dpo/gap Std: 25.9660
Beta Dpo/beta Used Raw: 0.0986
Beta Dpo/beta Used: 0.1434
Beta Dpo/mask Keep Frac: 1.0
Logits/chosen: -0.8035
Logits/rejected: -0.7974

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: 8
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
num_devices: 8
gradient_accumulation_steps: 2
total_train_batch_size: 128
total_eval_batch_size: 64
optimizer: Use OptimizerNames.ADAMW_TORCH 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

Training results

Training Loss	Epoch	Step	Validation Loss	Beta Dpo/gap Mean	Beta Dpo/gap Std	Beta Dpo/beta Used Raw	Beta Dpo/beta Used	Beta Dpo/mask Keep Frac	Logits/chosen	Logits/rejected
1.1971	0.4188	200	0.6549	11.0198	18.6390	0.0997	0.1243	1.0	-0.7570	-0.7553
1.2165	0.8377	400	0.7668	15.9231	25.9660	0.0986	0.1434	1.0	-0.8035	-0.7974

Framework versions

Transformers 4.51.0
Pytorch 2.3.1+cu121
Datasets 2.21.0
Tokenizers 0.21.4

W-61/llama-3-8b-base-beta-dpo-ultrafeedback-8xh200

llama-3-8b-base-beta-dpo-ultrafeedback-8xh200-20260410-201956

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

W-61/
llama-3-8b-base-beta-dpo-ultrafeedback-8xh200