You can deploy this model and start earning money today!

Strike0010/
hs-classifier-70b-dpo-v1.1-arm1

Name: hs-classifier-70b-dpo-v1.1-arm1
Availability: InStock

meta-llama/Llama-3.1-70B-Instruct

other

Views

No views yet

dpo-v1.1-arm1

This model is a fine-tuned version of meta-llama/Llama-3.1-70B-Instruct on the dpo_hs_v1_1_arm1 dataset. It achieves the following results on the evaluation set:

Loss: 3.9840
Rewards/chosen: 0.6766
Rewards/rejected: 0.1785
Rewards/accuracies: 0.7500
Rewards/margins: 0.4981
Logps/chosen: -16.3349
Logps/rejected: -20.9329
Logits/chosen: -1.0626
Logits/rejected: -1.0698

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: 7e-07
train_batch_size: 1
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 8
total_train_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
4.8073	0.4556	25	4.7574	0.2861	-0.2060	0.7500	0.4921	-20.2400	-24.7776	-1.0700	-1.0776
4.2015	0.9112	50	3.9840	0.6766	0.1785	0.7500	0.4981	-16.3349	-20.9329	-1.0626	-1.0698

Framework versions

PEFT 0.17.1
Transformers 4.57.1
Pytorch 2.10.0+cu128
Datasets 4.0.0
Tokenizers 0.22.2

Strike0010/hs-classifier-70b-dpo-v1.1-arm1

dpo-v1.1-arm1

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

Strike0010/
hs-classifier-70b-dpo-v1.1-arm1