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: 5.6625
Rewards/chosen: -36.8871
Rewards/rejected: -38.7328
Rewards/accuracies: 0.625
Rewards/margins: 1.8457
Logps/rejected: -387.3283
Logps/chosen: -368.8713
Logits/rejected: -0.7502
Logits/chosen: -0.7660
Nll Loss: 1.3492

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: 4
eval_batch_size: 4
seed: 42
distributed_type: multi-GPU
num_devices: 8
gradient_accumulation_steps: 4
total_train_batch_size: 128
total_eval_batch_size: 32
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	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen	Nll Loss
27.4789	0.4188	200	6.6576	-34.2403	-34.9452	0.5766	0.7050	-349.4524	-342.4026	-0.7404	-0.7582	1.2307
23.6103	0.8377	400	5.6625	-36.8871	-38.7328	0.625	1.8457	-387.3283	-368.8713	-0.7502	-0.7660	1.3492

Framework versions

Transformers 4.51.0
Pytorch 2.3.1+cu121
Datasets 2.21.0
Tokenizers 0.21.4

jackf857/llama-3-8b-base-cpo-ultrafeedback-8xh200

llama-3-8b-base-cpo-ultrafeedback-8xh200

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

jackf857/
llama-3-8b-base-cpo-ultrafeedback-8xh200